CHIRPS Precipitation, Copernicus DEM, (Ag)ERA5 Meteorological Data, GEOS-5 Meteorological Data, IMERG Precipitation, Landsat satellites, MODIS sensors, MSG satellites, Sentinel-2 satellites, VIIRS sensors and WorldCover Land Cover

Landsat

For WaPOR version 3, Landsat is a secondary source for calculation of [Albedo]((../Intermediate_Data_Components/Albedo), NDVI and the Land Surface Temperature (LST) at 20m (L3).

Purpose of data

Landsat data is used for the production of the following data components:
‐ Surface albedo for Level 3 (selected areas, currently Ethiopia, Kenya, Mozambique, Rwanda, Sri Lanka, Libya and Yemen)
‐ NDVI/fAPAR composites for Level 3 a(selected areas, currently Ethiopia, Kenya, Mozambique, Rwanda, Sri Lanka, Libya and Yemen) ‐ Land surface temperature at Level 3 (selected areas, currently Yemen and Libya)

Approach

The Landsat program started in 1972 with the Landsat 1 satellite. Currently the 9th Landsat satellite (launched in 2021) is orbiting the Earth, carrying the Operational Land Imager (OLI-2) and Thermal Infrared Sensor (TIRS-2). Furthermore Landsat 8 (2013-present) and Landsat 7 (2009-present) are also still operational. The Landsat Next mission is planned for launch in late 2030.

For WaPOR version 3, the 10m Sentinel-2 data is the preferred option to calculate 20m L3 Albedo and NDVI. However, in some areas, cloud cover hampers data acquisition. To minimize temporal data gaps, Landsat is used as a secondary source of information.

For the Land Surface Temperature, VIIRS has become the preferred option as it acquires data multiple times a day. VIIRS data is available since 2012, but at L3 requires thermal sharpening using Sentinel-2, which is available since mid 2015. For this reason, for L3 areas that require a long historical archive, Landsat is used.

All Landsat optical and thermal bands except the panchromatic band are used. Depending on availability, we either use Landsat Level 1 or Level 2 data. - The Landsat Level 1 standard terrain corrected product is converted from digital numbers to radiance and atmospheric corrected using SMAC (Simplified Model for Atmospheric Correction) after H. Rahman & G. Dedieu (1994). All gaps in Landsat 7 data that are affected by the scan line corrector (SLC) error are filled using the method proposed by Chen et al (2011). To derive land surface temperature, for Landsat 5 and 7, the method described in Jiménez-Muñoz et al. (2009) is used and for Landsat 8, the method described in Rozenstein, et al. (2014) is used.
- The Landsat L2 Surface Reflectance and Landsat Surface Temperature products only require cloud masking.

Generally Landsat Level 2 multispectral data is used to calculate NDVI and albedo while Landsat Level 1 brightness temperature is used to estimate land surface temperature. Cloud masking is done manually or by an in-house neural network approach (Kappa mask). KappaMask is a cloud detector developed by KappaZeta LTD for Sentinel-2 Level-1C and Level-2A input products. The project was funded by European Space Agency, Contract No. 4000132124/20/I-DT (Domnich et al, 2021).

Challenges

The amount of Landsat data available in archives was not constant over time, location, and sensor, due to sensor issues, reception capabilities (ground stations) and selective acquisition (payload, cost constraints). Landsat 7 developed a problem with its scan-line corrector, leading to reduced data quality, affecting Landsat 7 data from June 2003 onward.

Another challenge when producing time series is the high cloud cover that occurs over certain areas.

Alternative sources

Alternative sources of freely available, moderate resolution multi-spectral (in VNIR) satellite sensors exist that could be used as alternatives if significant data gaps in the Landsat archives occur for a Level 3 ROI.

These are as follows:
- Sentinel-2A and 2B data are available from 2015 and 2017 respectively. Its VNIR bands are at 10m resolution and compare well with those of Landsat 8. The Sentinel-2 constellation can serve as back up for recent data acquisitions. Unfortunately, Sentinel-2 satellites do not have thermal bands.
- ASTER onboard Terra has VNIR data at 15 m resolution and thermal infrared bands at 90m resolution. It has a temporal extent from 2000 to present but data has only been acquired for areas that have been requested in the past. Spectrally, ASTER’s Green, Red and NIR bands compare well with those of Landsat 7 ETM.
- CBERS (China Brazil Resource Satellite) has VNIR and thermal sensors (CCD, PANMUX, IRMSS) which are spectrally similar to Landsat. , CBERS-2B and CBERS 4 form potential alternative data sources within the period of interest. Limited data is freely available from INPE for Africa.