--- # Product africa_flood_occurrence # url: https://odcv2-explorer.staging.iwmi.org/products/africa_flood_occurrence.odc-product.yaml name: africa_flood_occurrence license: CC-BY-4.0 metadata_type: eo3 description: 'Annual flood occurrence across Africa at ~500 m, 2001 to 2024, from two parallel IWMI series on one grid: flood_occurrence (the finer count, per-year maximum 17 to 37) and inter_annual_flood_occurrence (consistently smaller). Both start at 1 and neither is documented by the publisher, but their distributions are those of counts. Rebuilt from IWMI Africa GeoPortal tile services, which serve no downloadable raster.' metadata: product: name: africa_flood_occurrence properties: odc:file_format: GeoTIFF measurements: - name: flood_occurrence dtype: int32 units: '1' nodata: -2147483648 aliases: - flood - occurrence - name: inter_annual_flood_occurrence dtype: float32 units: '1' nodata: -9.999e+03 default_iwmi: theme: Flood caveat: "What the values MEAN is undocumented. There is no legend, no raster attribute table (the service answers \"There is no raster attribute table available\"), no description on the ArcGIS item, and nothing in the catalogue. The publisher's own tags -- \"Flood, Occurrence, Inter-Annual\" -- and three measured properties all point to a count of flood detections within the year rather than a class code:\n\ Values start at 1 and decay monotonically, with a stable shape in every one of the 24 years: 38-60% of valid cells are 1, 18-23% are 2, 8-14% are 3.\nThe maximum is NOT fixed. It runs from 17 in 2001 to 37 in 2020, varying year by year -- a closed classification would have the same top code throughout, whereas a count naturally reaches higher in a worse year. (The service's own advertised maxValues of 17 is therefore per-service, not a scheme-wide limit; reading it once, off 2001, is how this product first got documented wrongly.)\nThe wettest years by this measure are 2020, 2022 and 2024, which is consistent with the flooding those years are known for in eastern Africa.\nStill, this is inference from the data, not documentation. Confirm with IWMI before publishing anything that depends on it, and do not report \"37 floods\" as though the unit were established.\nThe rasters are extremely sparse and that is expected, not a defect. Only 0.007% to 0.054% of the 271,970,520 cells carry a value -- 17,877 cells in 2002, 147,486 in 2020 -- because only cells where flooding was detected are populated and everything else is nodata. So every statistic must be computed over the valid cells alone; a mean over the frame is meaningless, and a \"percentage of Africa flooded\" taken from cell counts without honouring the nodata mask will be wrong by orders of magnitude.\nThat eightfold swing in populated cells between years is the product's main signal, and it is also a warning: because the count of valid cells varies so much, any comparison between years must state whether it is comparing flooded AREA (valid cell count) or flood INTENSITY (the values themselves). They move differently -- 2003 and 2019 have similar areas but different maxima.\nNodata is the int32 minimum, -2147483648, not 0. The services report a minimum value of 1, so 0 is outside the real range -- but it is also the value an unmasked read would plausibly produce, and treating \"no flood detected\" and \"not assessed\"\ \ as the same thing is exactly the error this avoids.\nPermanent water appears to be excluded. Sampled hits concentrate where large floods are expected -- the Niger inland delta, the Lower Shire, the Sudd, Lake Chad, the Barotse floodplain -- but the Okavango delta returns none at all, which is consistent with permanent wetland being masked out of a flood product. That means absence of a value does not mean absence of water.\nTHE TWO BANDS ARE NOT INTERCHANGEABLE, and their relationship was measured across all 24 years and 1,462,852 shared cells rather than sampled. Both are integral counts -- inter_annual_flood_occurrence is stored float32 but every value is a whole number from 1 to 11, so do not read the float type as precision. Comparing them cell by cell:\n54.8% of shared cells hold exactly the same value. flood_occurrence is larger in about 43%, by +1 in 17.4% of cells and decaying from there. And in 42,161 cells -- 2.9% -- flood_occurrence is SMALLER, always by exactly one, never by more. So the safe statement is flood_occurrence >= inter_annual_flood_occurrence - 1; neither band dominates the other outright.\nA first check on two tiles suggested flood_occurrence was never smaller at all. That was wrong, and it is recorded here because the difference matters: a monotonic relationship would imply one band is a strict refinement of the other, and it is not.\nTheir footprints are identical in 21 of the 24 years and differ in 2002, 2005 and 2022 by 1,359 to 1,618 cells. So the valid mask is per-band, not per-dataset.\nOne unexplained structural feature: the maximum ratio between the bands is exactly 4.0 in 23 of the 24 years, and 7.0 in 2022. A cap that lands on a round number in 23 years running is unlikely to be coincidence, but nothing published explains it, so it is noted rather than interpreted.\nRebuilt from a tile cache rather than downloaded, so the values are whatever was published into the LERC pyramid. LERC is lossless and the level used is the service's native resolution, but the ArcGIS service recorded in each file's `source` tag stays the authority.\n\"_Clean\" in the source service names implies a filtering or QA step that is not described anywhere. Whatever was removed, and on what basis, is unknown." period: 2001-2024 region: Africa ... --- # Product africa_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/africa_water_accounting.odc-product.yaml name: africa_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for continental Africa at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: africa_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Africa both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 326 mm of rainfall against the dry season's 336 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.0007935 mm, so agreement between them corroborates nothing.\nDry-season water yield averages +60 mm and yearly +119 mm.\nOnly 51.4% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\n\ The 18 source services do not report identical corner coordinates: they differ by up to 1e-06 degrees, about 111.32 mm on the ground -- 0.00011 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: Africa ... --- # Product benificial_fraction_africa # url: https://odcv2-explorer.staging.iwmi.org/products/benificial_fraction_africa.odc-product.yaml name: benificial_fraction_africa license: CC-BY-4.0 metadata_type: eo3 description: benificial_fraction_africa explorer description metadata: product: name: benificial_fraction_africa measurements: - name: benificialfraction dtype: float32 units: percentage nodata: -9999 aliases: - BenificialFraction default_iwmi: theme: Water accounting region: Africa ... --- # Product blue_et_africa # url: https://odcv2-explorer.staging.iwmi.org/products/blue_et_africa.odc-product.yaml name: blue_et_africa license: CC-BY-4.0 metadata_type: eo3 description: The fraction of total ET that is from sources other than rainfall i.e. blue water sources including irrigation metadata: product: name: blue_et_africa measurements: - name: bluenet dtype: int16 units: mm/month nodata: -9999 aliases: - BlueET default_iwmi: theme: Evapotranspiration region: Africa ... --- # Product burkina_faso_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/burkina_faso_water_accounting.odc-product.yaml name: burkina_faso_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for Burkina Faso at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: burkina_faso_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Burkina Faso both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 772 mm of rainfall against the dry season's 38 mm.\nThe yearly bands are redundant rather than independent. dry + wet reproduces yearly to within 0.01324 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -89 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 50.4% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\nThe 18 source services do not report identical corner coordinates: they differ by up to 3e-06 degrees, about 333.96 mm on the ground -- 0.00032 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: Burkina Faso ... --- # Product central_africa_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/central_africa_water_accounting.odc-product.yaml name: central_africa_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for Central Africa at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: central_africa_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. No period is declared for this region anywhere -- not in a webmap title, an item description or the service metadata. The span given here is inherited from the sibling sets that DO declare one (Southern Region of Africa, and the continental set, both 2003-2021), on the assumption that these regional cuts come from the same processing run. Treat the period as an inference, not the publisher. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 578 mm of rainfall against the dry season's 702 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.0001831 mm, so agreement between them corroborates nothing.\nDry-season water yield averages +225 mm and yearly +351 mm.\nOnly 51.8% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\n\ Rebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority." period: 2003-2021 region: Central Africa ... --- # Product dem_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/dem_limpopo.odc-product.yaml name: dem_limpopo license: CC-BY-4.0 metadata_type: eo3 description: Digital elevation model of the Limpopo basin at 90 m, a 3D representation of the basin's topography. Indexed from IWMI's basin data cube; the underlying raster is served from IWMI's public object storage rather than copied here. metadata: product: name: dem_limpopo measurements: - name: gray dtype: int32 units: metre nodata: -9999 aliases: - elevation - dem default_iwmi: theme: Terrain & soils caveat: "Referenced, not held: the raster lives in IWMI's iwmi-dt-odc-products-public bucket in af-south-1, so this product breaks if that bucket moves. The source cube also declares 30 m for this product while the file is 90 m -- trust the dataset grid, not that number." ... --- # Product east_africa_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/east_africa_water_accounting.odc-product.yaml name: east_africa_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for East Africa at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: east_africa_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. No period is declared for this region anywhere -- not in a webmap title, an item description or the service metadata. The span given here is inherited from the sibling sets that DO declare one (Southern Region of Africa, and the continental set, both 2003-2021), on the assumption that these regional cuts come from the same processing run. Treat the period as an inference, not the publisher. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 280 mm of rainfall against the dry season's 349 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.01439 mm, so agreement between them corroborates nothing. The exception is a scatter of individual bad cells: runoff breaks the sum in 2972 cells of 5656923, by up to 155 mm. Fewer than a hundredth of a percent, but mask aggressively if you difference the seasons against the yearly band.\nDry-season water yield averages +30 mm and yearly +84 mm.\nOnly 51.0% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\n\ The 18 source services do not report identical corner coordinates: they differ by up to 4e-06 degrees, about 445.28 mm on the ground -- 0.00042 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: East Africa ... --- # Product eflow_warnings_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/eflow_warnings_limpopo.odc-product.yaml name: eflow_warnings_limpopo license: CC-BY-4.0 metadata_type: eo3 description: 'Rasterized e-flow warnings derived from simulated SWAT discharges implemented by IWMI under the LIMCOM Digital Twin project. ' metadata: product: name: eflow_warnings_limpopo measurements: - name: LegendIndex dtype: uint8 units: '1' nodata: 0 default_iwmi: theme: Environmental flows ... --- # Product et_fraction_africa # url: https://odcv2-explorer.staging.iwmi.org/products/et_fraction_africa.odc-product.yaml name: et_fraction_africa license: CC-BY-4.0 metadata_type: eo3 description: Indicates which fraction of the total inflow of water is consumed and which part is converted into renewable resources metadata: product: name: et_fraction_africa measurements: - name: etfraction dtype: float32 units: percentage nodata: -9999 aliases: - ETFraction default_iwmi: theme: Water accounting region: Africa ... --- # Product ethiopia_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/ethiopia_water_accounting.odc-product.yaml name: ethiopia_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for Ethiopia at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: ethiopia_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Ethiopia both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 445 mm of rainfall against the dry season's 385 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.007408 mm, so agreement between them corroborates nothing.\nDry-season water yield averages +33 mm and yearly +198 mm.\nOnly 53.6% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\n\ The 18 source services do not report identical corner coordinates: they differ by up to 5e-06 degrees, about 556.60 mm on the ground -- 0.00053 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: Ethiopia ... --- # Product et_monthly_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/et_monthly_limpopo.odc-product.yaml name: et_monthly_limpopo license: CC-BY-4.0 metadata_type: eo3 description: Monthly Actual Evapotranspiration and Interception for the Limpopo River Basin metadata: product: name: et_monthly_limpopo measurements: - name: L2_AETI_M dtype: float32 units: et nodata: -9999 default_iwmi: theme: Evapotranspiration ... --- # Product evaporation_africa # url: https://odcv2-explorer.staging.iwmi.org/products/evaporation_africa.odc-product.yaml name: evaporation_africa license: CC-BY-4.0 metadata_type: eo3 description: 'Evaporation Africa ' metadata: product: name: evaporation_africa measurements: - name: evaporation dtype: float32 units: mm/month nodata: -9999 aliases: - Evaporation default_iwmi: theme: Evapotranspiration region: Africa ... --- # Product evaporative_stress_index_africa # url: https://odcv2-explorer.staging.iwmi.org/products/evaporative_stress_index_africa.odc-product.yaml name: evaporative_stress_index_africa license: CC-BY-4.0 metadata_type: eo3 description: Evaporative Stress Index Africa metadata: product: name: evaporative_stress_index_africa measurements: - name: evaporative_stress_index dtype: float32 units: '1' nodata: -9999 aliases: - Evaporaive_Stress_Index default_iwmi: theme: Drought region: Africa ... --- # Product gefis_eflow_sites # url: https://odcv2-explorer.staging.iwmi.org/products/gefis_eflow_sites.odc-product.yaml name: gefis_eflow_sites license: CC-BY-4.0 metadata_type: eo3 description: "Environmental flow assessment sites from IWMI's GEFIS platform: 1,689 point locations across 36 countries, each with the assessment method and model, the e-flow requirement volume, naturalised mean annual runoff at the site, the present-day ecological condition, and whether supporting national legislation exists. Site observations rather than modelled grid output -- this is the evidence base behind gefis_environmental_flows." metadata: product: name: gefis_eflow_sites properties: odc:vector: true odc:file_format: GeoJSON measurements: - name: feature_count dtype: int32 units: '1' nodata: -1 default_iwmi: theme: Environmental flows zonal: false caveat: "58% of the features arrive with their text already destroyed. The payload is valid UTF-8 as served and contains 4,575 literal U+FFFD replacement characters, so the loss happened upstream in the platform's own export, not in transit and not here -- the original bytes are gone and cannot be recovered from this source. It hits accented-language records hardest: 455 of 457 French sites, 182 of 246 Mexican, 159 of 164 Canadian. Expect \"Reventaz?n\" for Reventazon and \"T?rraba\"\ \ for Terraba. Do not treat these place, basin or river names as authoritative, and do not key a join on them.\nNo attribute is a usable feature id, so feature ids here are positional. EFUID is the source's own identifier and is unique across the 1,552 features that carry one, but 137 have none, and ODC falls back to the row position per-row -- which mixes small integers into the same key space as real EFUIDs. Passing EFUID as the id column was measured to collapse 1,689 features into 1,613 keys, silently dropping 76. Join on EFUID by all means, but expect 137 rows that cannot be joined, and do not assume the positional ids survive a re-export by the platform.\nCoverage is opportunistic, not systematic: it is where assessments have been published, so France (457), South Africa (343) and Australia (296) hold 65% of the sites and most of the world is absent. 384 sites fall in Africa. A count of sites per country measures reporting effort, not the state of environmental flows there.\nThe volumes are not comparable as served. E_flow_Requirement_Volume and the runoff at each site carry their own per-row units in `ef_unit` and `mar_unit`; read those before aggregating anything.\nNo observation date travels with the sites -- \"year of published\" is populated on 344 of 1,689 -- so the dataset time is the platform reference, not a survey epoch, exactly as for gefis_environmental_flows." region: Global ... --- # Product gefis_environmental_flows # url: https://odcv2-explorer.staging.iwmi.org/products/gefis_environmental_flows.odc-product.yaml name: gefis_environmental_flows license: CC-BY-4.0 metadata_type: eo3 description: "Global environmental flow requirements at 0.1 degree from IWMI's GEFIS platform: modelled natural annual runoff, environmental flow as a percentage and a volume, sustainable groundwater abstraction, and the environmental management class -- each also broken out for management classes A to D. Referenced from eflows.iwmi.org rather than copied into this cube." metadata: product: name: gefis_environmental_flows properties: odc:file_format: GeoTIFF measurements: - name: natural_annual_runoff dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: ef_percentage dtype: float32 units: percent nodata: -340282346638528860000000000000000000000 - name: ef_mcm dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: sga dtype: float32 units: mcm*1e-3 nodata: -340282346638528860000000000000000000000 aliases: - sgwa - name: emc_classes dtype: uint8 units: '1' nodata: 15 - name: ef_percentage_a dtype: float32 units: percent nodata: -340282346638528860000000000000000000000 - name: ef_total_a dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: base_flow_a dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: sga_a dtype: float32 units: mcm*1e-3 nodata: -340282346638528860000000000000000000000 aliases: - sgwa_a - name: ef_percentage_b dtype: float32 units: percent nodata: -340282346638528860000000000000000000000 - name: ef_total_b dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: base_flow_b dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: sga_b dtype: float32 units: mcm*1e-3 nodata: -340282346638528860000000000000000000000 aliases: - sgwa_b - name: ef_percentage_c dtype: float32 units: percent nodata: -340282346638528860000000000000000000000 - name: ef_total_c dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: base_flow_c dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: sga_c dtype: float32 units: mcm*1e-3 nodata: -340282346638528860000000000000000000000 aliases: - sgwa_c - name: ef_percentage_d dtype: float32 units: percent nodata: -340282346638528860000000000000000000000 - name: ef_total_d dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: base_flow_d dtype: float32 units: mcm nodata: -340282346638528860000000000000000000000 - name: sga_d dtype: float32 units: mcm*1e-3 nodata: -340282346638528860000000000000000000000 aliases: - sgwa_d default_iwmi: theme: Environmental flows caveat: "No observation period is declared anywhere in the platform: the datetime on this dataset is the 2024 IWMI publication reference (doi 10.5337/2024.204), not a period the data describes, so do not read it as an epoch. The user manual contains no year at all. GEFIS is PCR-GLOBWB model output, not observation.\n\ Do not aggregate these by unweighted pixel arithmetic. The grid is EPSG:4326 at a constant 0.1 degrees, so cell ground area falls away from the equator, and the volume bands (mcm) are per-cell totals. The platform's own zonal engine corrects for this -- its manual states it \"utilizes latitude-weighted pixel area algorithms to correct for the distortion inherent in unprojected raster datasets\" -- so a plain mean or sum over cells in this cube will NOT reproduce the platform's published figures, and will overweight high latitudes.\nDo not read these bands at reduced resolution. The COG overviews average the float nodata sentinel into cells at the edge of the data, so a decimated read returns values that do not exist in the data: sampling at 1/8 gives runoff down to -5.9 and environmental flow percentages up to 112.5, where the full-resolution ranges are 0.0095..987 and 0.375..100. Any web-map rendering, OWS tile or overview-based statistic inherits that contamination. Read at native resolution, or mask before resampling.\nThe frame is global but only about 17.4% of cells carry data -- these are river cells, so a whole-frame mean is meaningless. The percentage bands cover slightly more cells than the volume bands (17.72% against 17.40%), so the bands do not share one mask; mask each on its own nodata rather than reusing one.\nemc_classes holds only 1, 2, 3 and 4, with the declared nodata of 15 as the fill -- so the nodata is right, and the four classes line up with the management classes A to D that the _a.._d bands break out.\nReferenced from eflows.iwmi.org, so this product breaks if that host moves." region: Global ... --- # Product ghana_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/ghana_water_accounting.odc-product.yaml name: ghana_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for Ghana at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: ghana_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Ghana both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 932 mm of rainfall against the dry season's 278 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.006134 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -148 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 68.4% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\nThe 18 source services do not report identical corner coordinates: they differ by up to 2e-06 degrees, about 222.64 mm on the ground -- 0.00021 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: Ghana ... --- # Product global_sst_anomaly # url: https://odcv2-explorer.staging.iwmi.org/products/global_sst_anomaly.odc-product.yaml name: global_sst_anomaly metadata_type: eo3_forecast description: Limpopo ENSO briefing (global) — sst_anomaly (degC) metadata: product: name: global_sst_anomaly properties: enso:scope: global measurements: - name: sst_anomaly dtype: float32 units: degC nodata: -9.999e+03 default_iwmi: theme: Climate ... --- # Product green_et_africa # url: https://odcv2-explorer.staging.iwmi.org/products/green_et_africa.odc-product.yaml name: green_et_africa license: CC-BY-4.0 metadata_type: eo3 description: Total ET resulting from direct rainfall (i.e. total green ET, sum of interception, soil evaporation and transpiration), excluding the ET from irrigation water withdrawals metadata: product: name: green_et_africa measurements: - name: greenet dtype: int16 units: mm/month nodata: -9999 aliases: - GreenET default_iwmi: theme: Evapotranspiration region: Africa ... --- # Product hccm_cropland_africa # url: https://odcv2-explorer.staging.iwmi.org/products/hccm_cropland_africa.odc-product.yaml name: hccm_cropland_africa license: CC-BY-4.0 metadata_type: eo3 description: "High Confidence Cropland Mask for Africa at ~300 m: the >=66% agreement of the three best-scoring cropland masks (WaPOR land cover, Copernicus CGLS-LC100 v3, GLC-SHARE) after accuracy and cultivated-area assessment. Continental extent, 17.5W to 51.4E and 34.8S to 37.3N. Indexed from IWMI's object storage rather than copied here." metadata: product: name: hccm_cropland_africa measurements: - name: cropmask dtype: float32 units: '1' nodata: NaN aliases: - hccm - cropland default_iwmi: theme: Cropland & land use caveat: Not a single-year map. The three contributing masks are dated 2019, 2019 and 2013, so the indexed validity spans 2013-2019 and no year within it should be read as the observation date. The file is also a striped GeoTIFF rather than a COG, so windowed reads are expensive, and it is referenced from IWMI's iwmi-datasets bucket rather than held in this cube's own storage. ... --- # Product hccm_crop_probability_africa # url: https://odcv2-explorer.staging.iwmi.org/products/hccm_crop_probability_africa.odc-product.yaml name: hccm_crop_probability_africa license: CC-BY-4.0 metadata_type: eo3 description: Agreement, in percent, between the three cropland masks that scored highest in the two-fold assessment behind the High Confidence Cropland Mask -- WaPOR land cover, Copernicus CGLS-LC100 v3 and GLC-SHARE. Africa at ~300 m. The HCCM is this layer thresholded at 66%. Indexed from IWMI's object storage rather than copied here. metadata: product: name: hccm_crop_probability_africa measurements: - name: crop_probability dtype: uint8 units: percent nodata: 15 aliases: - agreement - probability_crop default_iwmi: theme: Cropland & land use caveat: "Agreement between three masks, not a calibrated probability: the only values it can take are 0, 33, 67 and 100, so do not read it as a continuous likelihood. nodata is 15, which is safe precisely because no real value can land there. Inherits the 2013-2019 input span of the HCCM, is a 615 MB striped GeoTIFF rather than a COG -- windowed reads are expensive -- and is referenced from IWMI's iwmi-datasets bucket rather than held here." ... --- # Product hccm_formal_irrigated_africa # url: https://odcv2-explorer.staging.iwmi.org/products/hccm_formal_irrigated_africa.odc-product.yaml name: hccm_formal_irrigated_africa license: CC-BY-4.0 metadata_type: eo3 description: Formal irrigated areas over Africa at ~300 m, derived alongside the High Confidence Cropland Mask. Overlaid on hccm_rainfed_irrigated_africa it separates rainfed, supplementary and formal irrigation. Indexed from IWMI's object storage rather than copied here. metadata: product: name: hccm_formal_irrigated_africa measurements: - name: formal_irrigated dtype: float32 units: '1' nodata: NaN aliases: - formal_irri - formal default_iwmi: theme: Cropland & land use caveat: Only interpretable together with hccm_rainfed_irrigated_africa and hccm_cropland_africa -- alone it does not distinguish supplementary from formal irrigation. Inherits the 2013-2019 input span of the HCCM, is a striped GeoTIFF rather than a COG, and is referenced from IWMI's iwmi-datasets bucket rather than held in this cube. ... --- # Product hccm_rainfed_irrigated_africa # url: https://odcv2-explorer.staging.iwmi.org/products/hccm_rainfed_irrigated_africa.odc-product.yaml name: hccm_rainfed_irrigated_africa license: CC-BY-4.0 metadata_type: eo3 description: The High Confidence Cropland Mask for Africa partitioned into rainfed and irrigated cropland, ~300 m. Read with hccm_formal_irrigated_africa to separate rainfed, supplementary and formal irrigation. Indexed from IWMI's object storage rather than copied here. metadata: product: name: hccm_rainfed_irrigated_africa measurements: - name: irrigated_rainfed dtype: float32 units: '1' nodata: NaN aliases: - rainfed_irrigated - partition default_iwmi: theme: Cropland & land use caveat: Which value means rainfed and which means irrigated is not stated in the source Readme, so check with the producer before labelling a class. Inherits the 2013-2019 input span of the HCCM, is a striped GeoTIFF rather than a COG, and is referenced from IWMI's iwmi-datasets bucket rather than held here. ... --- # Product hecras_flood_dolo_ado # url: https://odcv2-explorer.staging.iwmi.org/products/hecras_flood_dolo_ado.odc-product.yaml name: hecras_flood_dolo_ado license: CC-BY-4.0 metadata_type: eo3 description: hecras_flood_dolo_ado metadata: product: name: hecras_flood_dolo_ado measurements: - name: HECRAS dtype: float32 units: '1' nodata: -340282000000000000000000000000000000000 aliases: - hecras default_iwmi: theme: Flood region: Ethiopia ... --- # Product incomati_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/incomati_water_accounting.odc-product.yaml name: incomati_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Incomati basin at ~1 km: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: incomati_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period comes from the publisher's own sibling items, \"Incomati Seasonal Maps (2003 - 2021)\" and \"Incomati Yearly Maps (2003 - 2021)\". Every cell is a 19-year mean, so no year is described and no trend can be read. The dataset time spans the period only to record it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where dry ends and wet begins, so these cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 648 mm of rainfall against the dry season's 120 mm, and the two sum to the yearly layer to within 0.0003 mm.\nThe yearly bands are redundant, not independent. Because dry + wet reproduces yearly exactly, the yearly layer carries no information the seasonal pair does not. Do not treat agreement between them as corroboration of anything.\nDry-season water yield is negative on average, about -45 mm. That is P - ET with ET exceeding rainfall, which is physically ordinary for a dry season, but it means the band is not a \"yield\" in the sense of water available.\nOnly 55.0% of the frame carries data -- the rest is outside the basin -- so any whole-frame statistic is meaningless.\nOne grid, with one wrinkle: 17 of the 18 source services report their corner coordinates at full precision and `Incomati_Water_Yield_yearly_avg` reports its rounded to six decimals, a difference of 4e-6 degrees or about 0.44 mm on the ground -- 0.0004 of a pixel. The dataset declares the precise grid the other 17 agree on. Nothing is resampled and the discrepancy is far below anything that could matter, but it is recorded so nobody rediscovers it as a bug.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the one pyramid level that matches the services' native resolution. The ArcGIS service in each file's `source` tag stays the authority.\n\ The basin is transboundary -- South Africa, Eswatini and Mozambique -- so this is not a South African dataset even though its headwaters are in Mpumalanga." period: 2003-2021 region: Incomati ... --- # Product incremental_channel_contributions_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/incremental_channel_contributions_limpopo.odc-product.yaml name: incremental_channel_contributions_limpopo license: CC-BY-4.0 metadata_type: eo3 description: Rasterized average monthly incremental channel contributions simulated by SWAT by the IWMI under the LIMCOM Digital Twin project. metadata: product: name: incremental_channel_contributions_limpopo measurements: - name: LegendIndex dtype: uint8 units: '1' nodata: 0 default_iwmi: theme: Basins & hydrology ... --- # Product interception_africa # url: https://odcv2-explorer.staging.iwmi.org/products/interception_africa.odc-product.yaml name: interception_africa license: CC-BY-4.0 metadata_type: eo3 description: Interception Africa Water Use metadata: product: name: interception_africa measurements: - name: interception dtype: float32 units: '1' nodata: -9999 aliases: - Interception default_iwmi: theme: Evapotranspiration region: Africa ... --- # Product irrigated_areas_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/irrigated_areas_limpopo.odc-product.yaml name: irrigated_areas_limpopo license: CC-BY-4.0 metadata_type: eo3 description: Monthly irrigated areas extent map produced by IWMI under the digital twin project. metadata: product: name: irrigated_areas_limpopo measurements: - name: map dtype: uint8 units: '1' nodata: 0 aliases: - Map - MAP - name: prob dtype: float32 units: '1' nodata: -32768 aliases: - irrigated_prob - PROBABILITY - name: filtered dtype: uint8 units: '1' nodata: 0 aliases: - irrigated_filtered - FILTERED default_iwmi: theme: Cropland & land use ... --- # Product kenya_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/kenya_water_accounting.odc-product.yaml name: kenya_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for Kenya at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: kenya_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Kenya both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 424 mm of rainfall against the dry season's 216 mm.\nThe yearly bands are redundant rather than independent. dry + wet reproduces yearly to within 0.007874 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -17 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 63.5% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\nThe 18 source services do not report identical corner coordinates: they differ by up to 4e-06 degrees, about 445.28 mm on the ground -- 0.00042 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: Kenya ... --- # Product lake_tana_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/lake_tana_water_accounting.odc-product.yaml name: lake_tana_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Lake Tana basin at ~110 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2010-2020. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: lake_tana_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2010-2020 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Lake Tana both carry 2010-2020 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 1006 mm of rainfall against the dry season's 282 mm.\nThe yearly bands are redundant rather than independent. dry + wet reproduces yearly to within 0.0002441 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -181 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 45.9% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority." period: 2010-2020 region: Lake Tana ... --- # Product land_use_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/land_use_limpopo.odc-product.yaml name: land_use_limpopo license: CC-BY-4.0 metadata_type: eo3 description: "Land use and land cover of the Limpopo basin at 90 m: a granular account of the basin's surface composition, as classes. Indexed from IWMI's basin data cube; the underlying raster is served from IWMI's public object storage rather than copied here." metadata: product: name: land_use_limpopo measurements: - name: landuse dtype: uint8 units: '1' nodata: 0 aliases: - land_use - lulc default_iwmi: theme: Cropland & land use caveat: "Referenced, not held: the raster lives in IWMI's iwmi-dt-odc-products-public bucket in af-south-1, so this product breaks if that bucket moves. The class codes themselves are undocumented in the source metadata -- 68 distinct values are present and no legend travels with the file, so do not label a class without checking with the producer." ... --- # Product letaba_catchment # url: https://odcv2-explorer.staging.iwmi.org/products/letaba_catchment.odc-product.yaml name: letaba_catchment license: CC-BY-4.0 metadata_type: eo3 description: Letaba catchment boundary (Groot and Klein Letaba), the area of interest the WaPOR 20 m downscaler is run over. Indexed as an ODC vector product so it can be used as a zonal unit alongside the raster products in this cube. The shipped geometry is a hand-digitised approximation that over-covers the published catchment by about 19%. metadata: product: name: letaba_catchment properties: odc:vector: true odc:file_format: GeoJSON measurements: - name: feature_count dtype: int32 units: '1' nodata: -1 default_iwmi: theme: Basins & hydrology caveat: 'The shipped boundary is hand-digitised and marked approximate: ~15,900 km2 against a published ~13,400 km2, so it over-covers by roughly 19%. Fine for bringing a pipeline up, wrong for reporting or area accounting.' ... --- # Product limpopo_cropland_10m # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_cropland_10m.odc-product.yaml name: limpopo_cropland_10m license: CC-BY-4.0 metadata_type: eo3 description: Annual cropland extent mask at 10 m for the Limpopo basin. Each year is the union of the land-cover products that actually cover that year -- no product is carried into years it does not represent -- so the inputs differ from year to year and are recorded on every dataset as crop:sources. Tuned for smallholder recall rather than high-confidence agreement. Coverage and method vary by dataset; check crop:sources before comparing years. metadata: product: name: limpopo_cropland_10m measurements: - name: cropland dtype: uint8 units: '1' nodata: 255 aliases: - crop_mask - cropland_mask flags_definition: cropland: bits: - 0 values: '0': not_cropland '1': cropland description: Cropland present in this year default_iwmi: theme: Cropland & land use caveat: The product is defined for the whole Limpopo basin but what is indexed covers only the Middle Olifants AOI, about 0.4% of it -- a query expecting basin coverage returns nothing outside that box, silently. Inputs also differ from year to year and three of the five are near-static, so read a year-on-year difference as partly method, not only ground change. ... --- # Product limpopo_irrigated_field_survey # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_irrigated_field_survey.odc-product.yaml name: limpopo_irrigated_field_survey license: CC-BY-NC-ND-4.0 metadata_type: eo3 description: 'Surveyed field locations for the Limpopo River Basin irrigated-area mapping: 525 sites in Zimbabwe, Mozambique, Botswana and South Africa plus a denser campaign in the Letaba catchment, each with its site id, country and how the site was chosen (sampling grid or known irrigation scheme). Locations only -- whether a field was found to be irrigated is not recorded. Recovered from the published field-trip maps; the source of record is the Earth Engine asset named in the caveat.' metadata: product: name: limpopo_irrigated_field_survey properties: odc:vector: true odc:file_format: GeoJSON measurements: - name: feature_count dtype: int32 units: '1' nodata: -1 default_iwmi: theme: Cropland & land use zonal: false caveat: "Survey locations only. Nothing here says whether a field was found to be irrigated -- the `class` attribute records how a site was CHOSEN (`grid`: a sampling-grid point; `known_scheme`: a visit to a known irrigation scheme), not what was observed there. The accuracy assessment in the source paper rests on a separate set of 190 labelled reference points (92 irrigated, 98 not) that is not in this dataset. Do not use these points as training or validation labels.\nRecovered, not sourced. The points were read back out of the two published field-trip web maps (Folium pages in the IWMIHQ/limpopo-irrigation-mapper Space), marker by marker, so this copy holds only what was drawn on those maps. The authoritative copy is the Earth Engine asset projects/tethys-app-1/assets/surveyed_irrigated_fields; check its feature count against the 525 here before relying on this version.\nUneven coverage. Sites are dense in South Africa's Letaba catchment (120 points from a separate campaign) and sparser elsewhere; the paper notes validation data were more available in South Africa than in Botswana, Zimbabwe or Mozambique. The 120 Letaba points carry no `country` value -- their labels never had one -- and are all in South Africa.\nPositions are as published, to five decimal places (roughly 1 m), with no stated GPS accuracy; treat them as field-level, not plot-level. No survey date travels with the points, so the dataset time is the campaign date given at indexing, not a per-site observation time. Sub-basin membership is not recorded on the points; join to limpopo_subbasins if it is needed.\nSource: Kiala et al., Operational Sentinel-2 system for monthly near-real-time irrigated area mapping in the Limpopo River Basin, Discover Water (2026) 6:41, doi:10.1007/s43832-026-00360-z. Both shortened data-availability links in that paper were broken on 2026-09-28. The article is CC BY-NC-ND 4.0, and this copy of its survey sites is offered under the same terms until IWMI states otherwise." region: Limpopo ... --- # Product limpopo_jfm_et0 # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_jfm_et0.odc-product.yaml name: limpopo_jfm_et0 metadata_type: eo3 description: Limpopo ENSO briefing (limpopo) — et0_anomaly (mm/season) metadata: product: name: limpopo_jfm_et0 properties: enso:scope: limpopo measurements: - name: jfm_et0_anomaly dtype: float32 units: mm/season nodata: -9.999e+03 default_iwmi: theme: Climate ... --- # Product limpopo_jfm_rainfall # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_jfm_rainfall.odc-product.yaml name: limpopo_jfm_rainfall metadata_type: eo3 description: Limpopo ENSO briefing (limpopo) — rainfall_anomaly (%) metadata: product: name: limpopo_jfm_rainfall properties: enso:scope: limpopo measurements: - name: jfm_rainfall_anomaly dtype: float32 units: '%' nodata: -9.999e+03 default_iwmi: theme: Climate ... --- # Product limpopo_jfm_temperature # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_jfm_temperature.odc-product.yaml name: limpopo_jfm_temperature metadata_type: eo3 description: Limpopo ENSO briefing (limpopo) — temperature_anomaly (degC) metadata: product: name: limpopo_jfm_temperature properties: enso:scope: limpopo measurements: - name: jfm_temperature_anomaly dtype: float32 units: degC nodata: -9.999e+03 default_iwmi: theme: Climate ... --- # Product limpopo_jfm_warm_days # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_jfm_warm_days.odc-product.yaml name: limpopo_jfm_warm_days metadata_type: eo3 description: Limpopo ENSO briefing (limpopo) — warm_days_anomaly (days) metadata: product: name: limpopo_jfm_warm_days properties: enso:scope: limpopo measurements: - name: jfm_warm_days_anomaly dtype: float32 units: days nodata: -9.999e+03 default_iwmi: theme: Climate ... --- # Product limpopo_rainfall_anomaly # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_rainfall_anomaly.odc-product.yaml name: limpopo_rainfall_anomaly metadata_type: eo3_forecast description: Limpopo ENSO briefing (limpopo) — rainfall_anomaly (mm/month) metadata: product: name: limpopo_rainfall_anomaly properties: enso:scope: limpopo measurements: - name: rainfall_anomaly dtype: float32 units: mm/month nodata: -9.999e+03 default_iwmi: theme: Climate ... --- # Product limpopo_rivers # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_rivers.odc-product.yaml name: limpopo_rivers license: CC-BY-4.0 metadata_type: eo3 description: 'River network of the Limpopo basin as an ODC vector product: 911 line segments spanning 25.3E to 34.2E and 26.3S to 19.9S, keyed by seg_id. A hydrological reference and routing network rather than a zonal unit -- these are lines, so they cannot be used for the per-polygon statistics limpopo_subbasins is for.' metadata: product: name: limpopo_rivers properties: odc:vector: true odc:file_format: GeoJSON measurements: - name: feature_count dtype: int32 units: '1' nodata: -1 default_iwmi: theme: Basins & hydrology zonal: false caveat: "The source endpoint returns 26,068 features but only 911 have any geometry; the other 25,157 are empty stubs with all properties null, and are not indexed. Do not take a count from the raw GeoJSON. The only other attribute, HYBAS_NAME, is the constant \"Limpopo\" on every real feature and carries no information. Lines, not polygons: not a zonal unit, and the dataset footprint is the network's convex hull rather than the lines themselves." ... --- # Product limpopo_subbasins # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_subbasins.odc-product.yaml name: limpopo_subbasins license: CC-BY-4.0 metadata_type: eo3 description: 'Sub-basins of the Limpopo river basin as an ODC vector product: 27 hydrological units covering roughly 412,000 km2 across South Africa, Botswana, Zimbabwe and Mozambique, with area, perimeter and population attributes. Intended as the zonal reporting units for raster products in this cube.' metadata: product: name: limpopo_subbasins properties: odc:vector: true odc:file_format: GeoJSON measurements: - name: feature_count dtype: int32 units: '1' nodata: -1 default_iwmi: theme: Basins & hydrology ... --- # Product limpopo_temperature_anomaly # url: https://odcv2-explorer.staging.iwmi.org/products/limpopo_temperature_anomaly.odc-product.yaml name: limpopo_temperature_anomaly metadata_type: eo3_forecast description: Limpopo ENSO briefing (limpopo) — temperature_anomaly (degC) metadata: product: name: limpopo_temperature_anomaly properties: enso:scope: limpopo measurements: - name: temperature_anomaly dtype: float32 units: degC nodata: -9.999e+03 default_iwmi: theme: Climate ... --- # Product madagascar_seasonal_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/madagascar_seasonal_water_accounting.odc-product.yaml name: madagascar_seasonal_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Seasonal water accounting for Madagascar at ~1 km: rainfall, total ET, green and blue ET, water yield and runoff, for each of two cropping seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: madagascar_seasonal_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: s1_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: s1_et dtype: float32 units: mm nodata: -9.999e+03 - name: s1_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - s1_green_et - name: s1_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - s1_blue_et - name: s1_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: s1_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: s2_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: s2_et dtype: float32 units: mm nodata: -9.999e+03 - name: s2_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - s2_green_et - name: s2_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - s2_blue_et - name: s2_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: s2_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "This is a 2003-2021 AVERAGE, not a time series, and the dataset datetime spans that period only to record it. Every cell is a 19-year mean for its season, so nothing here describes a particular year and no trend can be taken from it. The evidence for the averaging is the source's own naming: the sibling published items are \"Seasonal Maps of Madagascar (2003 - 2021)\" and every layer carries an \"Average\" tag.\nThe season boundaries are undeclared. Nothing in the service metadata, the DCAT feed or the ArcGIS item records says which months either season covers, so s1 and s2 cannot be aligned to a calendar or compared against another dataset's seasons. What is known is measured rather than assumed: s2 is much the wetter, at a mean 1,204 mm of rainfall against 241 mm for s1.\nThese rasters were rebuilt from a tile cache, not downloaded. The service refuses exportImage, so the values here are whatever was published into the LERC pyramid, at the one pyramid level that matches the service's native resolution. LERC is lossless and both WA+ identities close to float32 precision, so the reconstruction is sound -- but it is a copy of a cache, and the ArcGIS service recorded in each file's `source` tag remains the authority.\nRunoff has a different footprint from the other ten bands: 555,659 valid cells against 556,286, so 627 cells carry rainfall and ET but no runoff. Do not assume one mask across the product; mask each band on its own nodata.\nOnly 50.9% of the frame carries data -- the rest is ocean around the island -- so a whole-frame statistic is meaningless.\nOne name to distrust: layers on this portal are not reliably named. A sibling service is called `__GW_productivity_5km_conditional_output` while serving a 93 m grid. The grid declared above was read from the files, not their titles." period: 2003-2021 region: Madagascar ... --- # Product maputo_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/maputo_water_accounting.odc-product.yaml name: maputo_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Maputo basin at ~1 km: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: maputo_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period comes from the publisher's own sibling items, \"Maputo Seasonal Maps (2003 - 2021)\" and \"Maputo Yearly Maps (2003 - 2021)\". Every cell is a 19-year mean, so no year is described and no trend can be read. The dataset time spans the period only to record it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where dry ends and wet begins, so these cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 681 mm of rainfall against the dry season's 171 mm, and the two sum to the yearly layer to within 0.0001 mm.\nThe yearly bands are redundant, not independent. Because dry + wet reproduces yearly exactly, the yearly layer carries no information the seasonal pair does not. Do not treat agreement between them as corroboration of anything.\nDry-season water yield averages almost exactly zero, about +1 mm: across the basin ET consumes essentially all the dry-season rainfall. The band still runs from -574 to +138 mm cell by cell, so the near-zero mean is a cancellation, not a sign that little is happening.\nOnly 64.3% of the frame carries data -- the rest is outside the basin -- so any whole-frame statistic is meaningless.\n\ One grid, with one wrinkle: 17 of the 18 source services report their corner coordinates at full precision and `Maputo_Water_Yield_yearly_avg` reports its rounded to six decimals, a difference of 5e-6 degrees or about 0.55 mm on the ground -- 0.0005 of a pixel. The dataset declares the precise grid the other 17 agree on. Nothing is resampled and the discrepancy is far below anything that could matter, but it is recorded so nobody rediscovers it as a bug.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the one pyramid level that matches the services' native resolution. The ArcGIS service in each file's `source` tag stays the authority.\nThe basin is transboundary -- South Africa, Eswatini and Mozambique -- so this is not a South African dataset even though much of it drains KwaZulu-Natal." period: 2003-2021 region: Maputo ... --- # Product mara_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/mara_water_accounting.odc-product.yaml name: mara_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Mara basin at ~278 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2017. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: mara_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2017 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Mara both carry 2003-2017 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 702 mm of rainfall against the dry season's 370 mm.\nThe yearly bands are redundant rather than independent. dry + wet reproduces yearly to within 0.0005188 mm, so agreement between them corroborates nothing.\nDry-season water yield averages +68 mm and yearly +393 mm.\nOnly 36.6% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority." period: 2003-2017 region: Mara ... --- # Product north_africa_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/north_africa_water_accounting.odc-product.yaml name: north_africa_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for North Africa at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: north_africa_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. No period is declared for this region anywhere -- not in a webmap title, an item description or the service metadata. The span given here is inherited from the sibling sets that DO declare one (Southern Region of Africa, and the continental set, both 2003-2021), on the assumption that these regional cuts come from the same processing run. Treat the period as an inference, not the publisher. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 56 mm of rainfall against the dry season's 33 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.01892 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -19 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 53.3% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\nThe 18 source services do not report identical corner coordinates: they differ by up to 2e-06 degrees, about 222.64 mm on the ground -- 0.00021 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: North Africa ... --- # Product number_of_rainy_days # url: https://odcv2-explorer.staging.iwmi.org/products/number_of_rainy_days.odc-product.yaml name: number_of_rainy_days license: CC-BY-4.0 metadata_type: eo3 description: Rainy days are defined as days when precipitation (P) exceeds 0.1(P > 0.1).However, this threshold can be adjusted based on local recommendations to better reflect regional climatic conditions or specific hydrological modeling requirements. metadata: product: name: number_of_rainy_days measurements: - name: nRD dtype: int16 units: days/month nodata: -9999 aliases: - nRD default_iwmi: theme: Climate region: Africa ... --- # Product rift_valley_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/rift_valley_water_accounting.odc-product.yaml name: rift_valley_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Rift Valley at ~110 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2010-2020. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: rift_valley_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2010-2020 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Rift Valley both carry 2010-2020 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 669 mm of rainfall against the dry season's 363 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.0004272 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -152 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 33.7% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority." period: 2010-2020 region: Rift Valley ... --- # Product soil_map_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/soil_map_limpopo.odc-product.yaml name: soil_map_limpopo license: CC-BY-4.0 metadata_type: eo3 description: Soil map of the Limpopo basin at 90 m, describing the basin's pedosphere. Indexed from IWMI's basin data cube; the underlying raster is served from IWMI's public object storage rather than copied here. metadata: product: name: soil_map_limpopo measurements: - name: soilmap dtype: float32 units: '1' nodata: 0 aliases: - soil - soil_map default_iwmi: theme: Terrain & soils caveat: "Referenced, not held: the raster lives in IWMI's iwmi-dt-odc-products-public bucket in af-south-1, so this product breaks if that bucket moves. What the values mean is undocumented: float32, around 12,200 distinct values clustered near 262, with no legend or units in the source metadata. Treat it as an identifier of unknown scheme until the producer confirms otherwise, and do not do arithmetic on it." ... --- # Product souss_massa_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/souss_massa_water_accounting.odc-product.yaml name: souss_massa_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Souss-Massa basin at ~67 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2009-2018. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: souss_massa_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2009-2018 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Souss-Massa both carry 2009-2018 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 234 mm of rainfall against the dry season's 67 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 6.104e-05 mm, so agreement between them corroborates nothing.\nA few cells break the water balance itself. Green + blue ET misses total ET in 1001 cells of 7172227, by up to 1.004 mm; and water yield differs from rainfall minus ET in 61 cells of 7172227, by up to 1393 mm. That is far too few for the band mapping to be wrong -- it passes at the 99.9th percentile, which is what these definitions are checked on -- but they are real defects in the published data, so do not trust an extremum taken from these bands without masking.\nDry-season water yield averages -62 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 48.9% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\nThe 18 source services do not report identical corner coordinates: they differ by up to 2e-06 degrees, about 222.64 mm on the ground -- 0.00333 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2009-2018 region: Souss-Massa ... --- # Product southern_africa_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/southern_africa_water_accounting.odc-product.yaml name: southern_africa_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for southern Africa at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: southern_africa_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Southern Africa both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 555 mm of rainfall against the dry season's 135 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.009018 mm, so agreement between them corroborates nothing. The exception is a scatter of individual bad cells: runoff breaks the sum in 1825 cells of 5700875, by up to 238 mm. Fewer than a hundredth of a percent, but mask aggressively if you difference the seasons against the yearly band.\nDry-season water yield averages -77 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 58.4% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\nThe 18 source services do not report identical corner coordinates: they differ by up to 3e-06 degrees, about 333.96 mm on the ground -- 0.00032 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: Southern Africa ... --- # Product transpiration_africa # url: https://odcv2-explorer.staging.iwmi.org/products/transpiration_africa.odc-product.yaml name: transpiration_africa license: CC-BY-4.0 metadata_type: eo3 description: Transpiration Africa metadata: product: name: transpiration_africa measurements: - name: transpiration dtype: float32 units: '1' nodata: -9999 aliases: - Transpiration default_iwmi: theme: Evapotranspiration region: Africa ... --- # Product vegetation_condition_index_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/vegetation_condition_index_limpopo.odc-product.yaml name: vegetation_condition_index_limpopo license: CC-BY-4.0 metadata_type: eo3 description: Monthly vegetation condition index calculated using MODIS data to visualize drought extent map produced by IWMI under the digital twin project metadata: product: name: vegetation_condition_index_limpopo measurements: - name: VCI dtype: float32 units: '1' nodata: NaN default_iwmi: theme: Drought ... --- # Product volta_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/volta_water_accounting.odc-product.yaml name: volta_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Volta basin at ~248 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2010-2019. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: volta_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2010-2019 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Volta both carry 2010-2019 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 788 mm of rainfall against the dry season's 115 mm.\nThe seasonal and yearly bands DO NOT agree, and this geography is unusual in that. dry + wet differs from yearly by up to 38 mm (99.9th percentile 26.9 mm) across 1843507 cells of 6764119 -- not a few bad pixels but a systematic offset, so the yearly layers are evidently a different processing run from the seasonal ones rather than their sum. Do not mix the two: pick the yearly bands or the seasonal pair and stay within it.\n\ Dry-season water yield averages -106 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 48.8% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority." period: 2010-2019 region: Volta ... --- # Product wapor_aoi_vector # url: https://odcv2-explorer.staging.iwmi.org/products/wapor_aoi_vector.odc-product.yaml name: wapor_aoi_vector metadata_type: eo3 description: Areas of interest / irrigation scheme boundaries used to drive the WaPOR 20 m downscaler, indexed as an ODC vector product. metadata: product: name: wapor_aoi_vector properties: odc:vector: true odc:file_format: GeoJSON measurements: - name: feature_count dtype: int32 units: '1' nodata: -1 ... --- # Product wapor_eta_20m # url: https://odcv2-explorer.staging.iwmi.org/products/wapor_eta_20m.odc-product.yaml name: wapor_eta_20m license: CC-BY-4.0 metadata_type: eo3 description: WaPOR Level 1 AETI_D downscaled from 300 m to 20 m using the IWMI SwinIR + Prithvi ensemble with Sentinel-2 and terrain predictors. Dekadal (~10 day) actual evapotranspiration and interception. metadata: product: name: wapor_eta_20m measurements: - name: eta dtype: float32 units: mm/day nodata: -9.999e+03 aliases: - ETa20m - aeti - AETI_D default_iwmi: theme: Evapotranspiration caveat: 'Downscaled output, not an independent observation: values inherit the biases of the 300 m WaPOR input, and the ensemble was trained over the Letaba catchment only. Treat it as unvalidated outside that footprint.' ... --- # Product wapor_limpopo # url: https://odcv2-explorer.staging.iwmi.org/products/wapor_limpopo.odc-product.yaml name: wapor_limpopo license: CC-BY-4.0 metadata_type: eo3 description: Downscaled WaPOR product. Extracted for Limpopo Region. 20m scale. metadata: product: name: wapor_limpopo measurements: - name: wapor dtype: int16 units: '1' nodata: -32768 aliases: - Wapor default_iwmi: theme: Evapotranspiration ... --- # Product west_africa_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/west_africa_water_accounting.odc-product.yaml name: west_africa_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for West Africa at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: west_africa_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. No period is declared for this region anywhere -- not in a webmap title, an item description or the service metadata. The span given here is inherited from the sibling sets that DO declare one (Southern Region of Africa, and the continental set, both 2003-2021), on the assumption that these regional cuts come from the same processing run. Treat the period as an inference, not the publisher. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 114 mm of rainfall against the dry season's 534 mm.\nThe yearly bands are redundant rather than independent: dry + wet reproduces yearly to within 0.0002441 mm, so agreement between them corroborates nothing.\nDry-season water yield averages +178 mm and yearly +124 mm.\nOnly 65.8% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\n\ Rebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority." period: 2003-2021 region: West Africa ... --- # Product zambezi_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/zambezi_water_accounting.odc-product.yaml name: zambezi_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for the Zambezi basin at ~1104 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: zambezi_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Zambezi both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 888 mm of rainfall against the dry season's 39 mm.\nThe yearly bands are redundant rather than independent. dry + wet reproduces yearly to within 0.01562 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -220 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 55.9% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority." period: 2003-2021 region: Zambezi ... --- # Product zambia_water_accounting # url: https://odcv2-explorer.staging.iwmi.org/products/zambia_water_accounting.odc-product.yaml name: zambia_water_accounting license: CC-BY-4.0 metadata_type: eo3 description: "Water accounting for Zambia at ~1060 m: rainfall, total ET, green and blue ET, water yield and runoff, each as a yearly average and split into dry and wet seasons, averaged over 2003-2021. Water Accounting Plus output rebuilt from IWMI's Africa GeoPortal tile services, which serve no downloadable raster." metadata: product: name: zambia_water_accounting properties: odc:file_format: GeoTIFF measurements: - name: yearly_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_et dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_green_et - name: yearly_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - yearly_blue_et - name: yearly_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: yearly_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: dry_et dtype: float32 units: mm nodata: -9.999e+03 - name: dry_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_green_et - name: dry_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - dry_blue_et - name: dry_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: dry_runoff dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall dtype: float32 units: mm nodata: -9.999e+03 - name: wet_et dtype: float32 units: mm nodata: -9.999e+03 - name: wet_rainfall_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_green_et - name: wet_incremental_et dtype: float32 units: mm nodata: -9.999e+03 aliases: - wet_blue_et - name: wet_water_yield dtype: float32 units: mm nodata: -9.999e+03 - name: wet_runoff dtype: float32 units: mm nodata: -9.999e+03 default_iwmi: theme: Water accounting caveat: "A 2003-2021 AVERAGE, not a time series. The period is the publisher's own: its sibling \"Seasonal Maps\" and \"Yearly Maps\" items for Zambia both carry 2003-2021 in their titles. Every cell is a multi-year mean, so no individual year is described and no trend can be read from it.\nSeason months are undeclared. Nothing in the service metadata or the item records says where the dry season ends and the wet begins, so these bands cannot be aligned to a calendar or compared with another dataset's seasons. What is known is measured: the wet season averages 983 mm of rainfall against the dry season's 30 mm.\nThe yearly bands are redundant rather than independent. dry + wet reproduces yearly to within 0.01166 mm, so agreement between them corroborates nothing.\nDry-season water yield averages -281 mm, i.e. negative: ET exceeds rainfall and draws on stored moisture. Ordinary for a dry season, but it means that band is not \"water available\".\nOnly 54.4% of the frame carries data -- the rest lies outside the mapped area -- so any whole-frame statistic is meaningless.\nRebuilt from a tile cache rather than downloaded, so these values are whatever was published into the LERC pyramid, read at the pyramid level that matches the services' native resolution. The ArcGIS service named in each file's `source` tag remains the authority.\nThe 18 source services do not report identical corner coordinates: they differ by up to 3e-06 degrees, about 333.96 mm on the ground -- 0.00032 of a pixel. The dataset declares the grid the majority agree on. Nothing is resampled; this is recorded only so the discrepancy is not rediscovered as a bug." period: 2003-2021 region: Zambia ...