A comparative study of NOAA–AVHRR derived drought indices using change vector analysis

The objective of this study was to compare the spatial occurrences of droughts, detected by remotely sensed drought-indices over the desert-steppe and desert geo-botanical zones of Mongolia. All indices were derived from reflectance and thermal data sets, obtained from the NOAA–AVHRR data between 1982 and 1999. One group of the drought-indices is based on vegetation state derived from the reflective channels. This group includes the Normalized Difference Vegetation Index (NDVI), Anomaly of Normalized Difference Vegetation Index (NDVIA), Standardized Vegetation Index (SVI), and Vegetation Condition Index (VCI). Another group, based on surface brightness temperature derived from the thermal channel of NOAA–AVHRR, includes the Temperature Condition Index (TCI). The third group is based on combination between the reflective and thermal channels includes the ratio between Land Surface Temperature (LST) and NDVI (LST/NDVI), the Vegetation Health Index (VH), and the Drought Severity Index (DSI). Change detection procedure was performed by using the Change Vector Analysis in the temporal domain. Comparison analysis among the drought-indices reveals that there is no spatial coincidence between them, even when the vegetation growing period was divided into 2-month sub-periods — beginning, middle, and end. Based on the statistical analysis, higher correlations were found among the reflective indices while lesser or no relationships were found between the thermal and combination of the thermal and reflective indices. Furthermore, no agreement was found between the spatial extent of the satellite-derived drought-indices and the meteorological-based Palmer Drought Severity Index (PDSI) and also between the traditional ground-observed drought-affected-areas (DAA) maps. It was found that the combination of satellite-derived drought-indices can identify wider drought-occurred areas rather than the PDSI and the DAA maps. In summary, this study concludes that it is difficult to point out the most reliable drought index, and that the ground observations cannot provide sufficient information for validation of satellite derived drought indices.