————————————————–
3.3 Coarse-resolution surface soil moisture
product
In 2012, the ESA CCI project for SSM was established to fulfill global
long-term SSM monitoring by merging multiple available active and
passive microwave-based SSM products (Wagner et al., 2012). That same
year, the first SSM product from the ESA CCI (v0.1) was publicly
released. By involving new sensors and improving the merging scheme, the
subsequent SSM dataset has been updated over an extended spatiotemporal
coverage. The daily SSM product provides a consistent SSM record from
1978 to the present. The latest version (v04.4) of the ESA CCI SSM
product at depths of 0.5–5 cm was used in this study, with a spatial
resolution of 0.25 degree
(https://www.esa-soilmoisture-cci.org).
The SSM data were interpolated and resampled to 25 km × 25 km regular
grids (Figure 2).
3.4 MODIS products
The Moderate Resolution Imaging Spectroradiometer (MODIS) is a key
instrument onboard the Terra and Aqua satellites. Fine-resolution
ancillary variables LST, NDVI, BSA and LC information were collected
from the Version 6 products of Aqua MODIS
(https://lpdaac.usgs.gov/). The
daily LST and 16-day NDVI were provided by MYD11A1 and MYD13A2 at 1-km
resolution, while the 16-day albedo and annual LC were provided by
MCD43A3 and MCD12Q1 at 500-m resolution. The BSA data were calculated
from shortwave radiation of MCD43A3, which uses a linear combination of
the black-sky and white-sky albedo data, with weights of 0.34 for the
former and of 0.66 for the latter. All MODIS products were reprojected
consistently with the ESA CCI product. Missing values were filled using
the aforementioned spatiotemporal prediction method to ensure complete
coverage. The LST and NDVI data were resampled and aggregated into 1 km
× 1 km and 25 km × 25 km regular grids. The average aggregations of BSA
and modal aggregations of LC were achieved at both fine and coarse
grids.
3.5 DEM products
The DEM at 90-m resolution provided by the NASA Shuttle Radar
Topographic Mission (SRTM) within the study area was employed. The Void
Filled DEM product was downloaded fromhttps://www.usgs.gov/centers/eros.
The DEM data were resampled into 1 km × 1 km and 25 km × 25 km regular
grids by using average aggregations. The basic terrain factors at 1 km
and 25 km, including aspect and slope, were calculated from the DEM
information.