Redefining agricultural drought monitoring and forecasting in Kenya.

Drought is one of the most important environmental hazards in the Horn of Africa region, causing annual human and livestock losses and multi-million dollar economic losses. Monitoring and timely detection of drought plays a key role in natural resource management and mitigating its impacts. One of the common methods for monitoring agricultural drought is the use of the Vegetation Condition Index (VCI) based on remote sensing. While useful, the VCI has also several substantial limitations and cannot be a robust and generalizable method for different regions due to differences in climate, land cover, and spatio-temporal dynamics. 

In this study, a new and robust framework for drought detection based on MODIS time-history satellite images is developed. This method uses the NDVI and statistical analysis based on percentiles to define dynamic thresholds that depend on the climatic conditions of each region. Thus, the proposed method is not dependent on fixed values ​​and is able to adaptively consider spatial and temporal changes in vegetation cover. 

The proposed framework has been tested in several counties in Kenya and its results have been validated with field reports and ground data. The results show that the proposed method has a higher ability to identify drought robustly than methods based on fixed thresholds and can be used as an effective tool for drought monitoring in regions with diverse climates and land cover.