The effectiveness of meteorological drought indices at regional scales is hampered by the sparse distribution of in-situ measurement sites. Remote sensing (RS)-based drought indices typically standardize indirect drought indicators derived from vegetation responses and land surface temperature (LST) fluctuations; nevertheless, they are restricted by the low temporal sampling frequency and cloud cover. They lack a distinct physical interpretation crucial for water management and informed policy formulation. Existing soil moisture products still suffer from coarse spatial resolutions, rendering them unsuitable for local irrigation management. Moreover, existing drought indices, particularly those derived from high-resolution remote sensing sensors, remain inadequately validated across diverse landscapes and climates over Africa.
In this study, daily soil moisture (SM) data were acquired from 74 sites of the International Soil Moisture Network (ISMN) network in Africa. Keetch-Byram Drought Index (KBDI), derived by maximum daily temperature and total daily precipitation from ERA5, is employed as a meteorological drought index. Shortwave infrared Transformed Reflectance (STR), derived by Harmonized Landsat Sentinel (HLS) shortwave infrared reflectance sensitive to water content, is included as a representative RS-based index. A hybrid Scaled Drought Condition Index (SDCI) was used in this study, which is calculated by nominalized Landsat LST and green Normalized Difference Vegetation Index (NDVI) from the Harmonized Landsat and Sentinel (HLS) and precipitation from ERA5. Three RS products are also involved for comparison, including JPL ECOSTRESS evaporative stress index (ESI), NASA Soil Moisture Active Passive (SMAP) SM product, and European Space Agency (ESA) Soil Water Index (SWI) from Copernicus Global Land Products.
The ground assessment revealed that SMAP SM demonstrates the most stable and consistently good performance (R2 = 0.53 ± 0.19) when compared to the SM observations, followed by SWI. Conversely, ESI showed an insignificant correlation with SM. While all indices showed improved performance on croplands, they fall short in capturing SM variability within the tropical monsoon climate zone. Furthermore, KBDI adeptly captures the drying phase following the rainy season in each annual SM cycle, whereas during the rainy season, it shows limited correlation at the daily scale. The STR, exclusively generated from RS shortwave infrared reflectance, demonstrates a discerning correlation, although with a wider dispersion of values under high fractional vegetation cover. In comparison, hybrid SDCI demonstrates a convergent correlation with SM through the integration of thermal and precipitation data; however, in tropical areas, SDCI shows an insignificant correlation with SM during rainy seasons. To conclude, no single drought index demonstrates optimal performance across diverse land cover types and climate zones. Further assessment is required to examine the appropriateness of various indexes across different time scales. This study serves as a roadmap for proposing an enhanced and flexible drought index designed for high spatial resolution thermal infrared data. This index holds potential for application in forthcoming water cycle missions, such as TRISHNA and LSTM.

