WaPOR version 3 - Annemarie Klaasse - eLeaf - 05 May 2023.pdf
1. WaPOR version 3
5 May 2023 - Annemarie Klaasse, eLEAF
Contributors: Kristin Abraham, Joost Brombacher, Natalia Cárdenas, Jelle Degen, Lucas
Ellerbroek, Rutger Kassies, Joris van de Kerkhof, Roeland de Koning, Sheeba Lawrence, Ockert
Malan, Joost Meijer, Sabina Mirt, Henk Pelgrum, Sjef Sanders, Sotirios Soulantikas, Arnaud van
Dommele, Melvin van ‘t Holt, Steven Wonink, Karlis Zalite
2. Introduction
Together with VITO, eLEAF developed the WaPOR data version 0 (2016), version 1 (2018) and
version 2 (2019). Now it is time for version 3 (2023)
This presentation:
- Why do we launch a version 3?
- What are the major changes?
- What are the improvements?
3. Why version 3?
WaPOR version 2 was launched in June 2019
Version 3 builds further on version 2 incorporating:
User feedback
New satellite datasets (improved resolution and coverage)
New earth system modeling data (better quality)
Scientific advances (algorithms and approaches)
Technological advances (cloud computing and big data infrastructure)
And version 3 includes a new dataset:
Relative Soil Moisture (beta product)
4. WaPOR data - primary goals
NRT Near real-time (updated every dekad)
Consistent over time (2018 – preferably 2010 - until present)
Consistent over space (countries, ecosystems)
Consistent over scale (from field to global scale)
No data gaps (continuous temporal and spatial coverage)
Open-access inputs
5. WaPOR data production overview
• Data sourcing
• Pre-processing
Input data
• Gapfilling
• Harmonization
Intermediates
• Biophysical models
• Aggregation
WaPOR data
components
Collect all available data
(multiple sources)
Determine best value for each day
(daily intermediates with full coverage)
Run the models
(create WaPOR outputs)
6. WaPOR data production – data volumes in L2 (5 products)
• Data sourcing
• Pre-processing
Input data
• Gapfilling
• Harmonization
Intermediates
• Biophysical models
• Aggregation
WaPOR data
components
60,488,406 files sourced
84,694,743 raster input files
56,478,132 raster files
in assembly 106,005 raster blocks
packaged
7. Resulting in the following deliverables
Source code (Bitbucket)
Methodology and data manuals (Wiki)
Data components (delivered NRT after 3 days and as final after 6 dekads):
Interception (I)
Evaporation (E)
Transpiration (T)
Net Primary Productivity (NPP)
Precipitation (PCP)
Reference Evapotranspiration (RET)
Quality layers (QUAL_NDVI & QUAL_LST)
Actual
Evapotranspiration
(AETI)
Water
Productivity
FAO WaPOR database
Hand in Hand portal
8. Resulting in the following deliverables
Source code (Bitbucket)
Methodology and data manuals (Wiki)
Data components (delivered NRT after 3 days and as final after 6 dekads):
Interception (I)
Evaporation (E)
Transpiration (T)
Net Primary Productivity (NPP)
Precipitation (PCP)
Reference Evapotranspiration (RET)
Quality layers (QUAL_NDVI & QUAL_LST)
Actual
Evapotranspiration
(AETI)
Water
Productivity
FAO WaPOR database
Hand in Hand portal
9. What did change in version 3?
New inputs:
( Ag)ERA5 meteo
VIIRS
Sentinel-2
Copernicus DEM
WorldCover land cover
Preparation of intermediates:
VIIRS atmospheric correction
Cloud masking
Coefficients for albedo and fAPAR
Updated statics
Gapfilling and smoothening
Thermal sharpening
Modelling:
Soil moisture parameterization
Tenacity factor
Data components:
Extent
Spatial resolution
Delivery projection
Relative Soil Moisture (RSM) added as beta
product
Data production:
Cloud processing
Tile based processing
10. NDVI & albedo
More spatial detail with new sensors: VIIRS
(L1), Sentinel-2 (L2+L3) and Landsat (L3)
Improved data quality with new Kappa cloud
masking procedure
Improved smoothening and interpolation
(pixel based temporal fill algorithm: the
Whittaker smoother (Eilers, 2003))
Improved gap-filling approach (Weiss, Daniel
J., et al., 2014) using both spatial and
temporal information
Improved consistency between levels and
sensors by using new albedo coefficients
Original data
(instantaneous NDVI)
Version 3
(smoothened NDVI)
11. Land Surface Temperature (LST) - indicator of root zone soil moisture content
VIIRS replacing MODIS and Landsat
MODIS VIIRS Landsat
1000m 375m 100-120m
Daily + Daily + 1 x 8-16 days
2005 – present 2012 – present 1983 - present
# beyond
lifetime
# low spatial
resolution
# no LST
product at
375m
# already
resampled to
30m
12. Thermal sharpening of VIIRS 375 data with PyDMS to create high resolution LST:
Data Mining Sharpener (DMS) methodology by Gao et al (2012) and open-access code PyDMS published
by the European Space Agency (ET4FAO, Guzinski et al, 2019)
Land Surface Temperature (LST) - indicator of root zone soil moisture content
VIIRS 375m VIIRS 10m
VIIRS 100m Landsat 100m
Libya, 18 September 2022
Satellite missions carrying high spatial-temporal thermal infrared sensors such as LSTM (~2029), TRISHNA
(~2025) and SBG (~2027) are expected to offer better spatial-temporal coverage from 2025 onwards.
13. Data components (delivery)
Data component L1 L2 L3
Reference Evapotranspiration (RET) 20 km → 10 km
Precipitation (PCP) 5km
Water productivity datasets: Actual
Evapotranspiration (E, T, I, AETI) and Net
Primary Productivity (NPP) and Relative
Soil Moisture (RSM)
250m → 300m 100m 30m → 20m
GLOBAL Africa and Near East
All data delivered in UTM blocks, maintaining the best (equal area) projection
14. Relative soil moisture
In WaPOR it is used to estimate soil moisture stress, which limits vegetation
transpiration and can hamper biomass growth. Relative Soil Moisture is releases as a
beta product:
Values range between 0 and 1, where 0 is equal to soil moisture content at wilting point and 1 is
equal to soil moisture content at field capacity.
Values apply to the root zone
Update in the soil moisture model:
Modeled wind speed can get unrealistically low. In reality, this will create a high surface heat flux,
which creates turbulence, which provides a negative feedback again. For these situations with free
convection, an alternative method to calculate the aerodynamic resistance is used (Garrat, 1992)
15. Relative soil moisture: WaPOR vs SMAP
Soil Moisture Active/Passive (SMAP) L4
9km resolution root zone soil moisture
WaPOR RSM converted to
absolute values (cm3/cm3)
Office du Niger, Mali
16. L2 (100m) results
Full coverage
Africa & Near East
at 100m
Transpiration
Evaporation
Transpiration 1st dekad of May 2022
Egypt pivots
Version 2 Version 3
18. L2 (100m) results
Full coverage
Africa & Near East
at 100m
v2 vs v3
Example with desert and irrigated area
19. The full team
Kristin Abraham Joost Brombacher
Sotirios Soulantikas
Jelle Degen Rutger Kassies
Annemarie Klaasse Sheeba Lawrence
Roeland de Koning Henk Pelgrum
Steven Wonink Karlis Zalite
Sabina Mirt
Ockert Malan
Lucas Ellerbroek
Natalia Cárdenas
Joost Meijer
Joris v.d. Kerkhof
Sjef Sanders Arnaud v. Dommelen Melvin v.t Holt