Studying Drought Phenomena in the Continental United States in 2011 and 2012 Using Various Drought Indices
Lifu Zhang,Wenzhe Jiao,Hongming Zhang,Changping Huang,Qingxi Tong
DOI: https://doi.org/10.1016/j.rse.2016.12.010
IF: 13.5
2017-01-01
Remote Sensing of Environment
Abstract:Numerous drought indices have been developed to monitor drought conditions. However, different drought indices have differing characteristics, and are suitable for specific environments. The aim of this study was to compare the occurrence of drought, as detected by remote sensing across the Continental United States (CONUS). We used drought events during 2011 and 2012 to compare various indices developed for the study of drought phenomena. Three in situ drought indices, the Palmer drought severity index (PDSI), Z-index and standardized precipitation indices (SPI) with different time scales were used to evaluate drought conditions in different climate divisions. The drought indices compared in this study include the vegetation condition index (VCI), the temperature condition index (TCI), the perpendicular drought index (PDI) and modified PDI (MPDI) derived from moderate-resolution imaging spectroradiometer (MODIS) data, the precipitation condition index (PCI) derived from tropical rainfall measuring mission (TRMM) data, and the soil moisture condition index (SMCI) derived from advanced microwave scanning radiometer - earth observing system (AMSR-E). Other synthesized drought indices, which combine indices such as VCI, TCI, SMCI and PCI, were also compared in this study. These included the vegetation health index (VHI), temperature vegetation dryness index (TVDI), scaled drought condition index (SDCI), the microwave integrated drought index (MIDI), the synthesized drought index (SDI), the optimized meteorological drought index (OMDI), and the optimized vegetation drought index (OVDI). The results include a wide variety of drought conditions based on different drought indices. Meteorological drought indices indicated that more regions were under severe drought than agricultural drought indices did. Drought indices appear to be more similar to standard drought indices than PDI, MPDI and TVDI. The results also indicate that different indices have strengths and weaknesses in different climates across CONUS. SMCI has a good correlation with short-term SPI, and the sensitivity of SMCI is strongly dependent on terrain, as it performs worse in regions with heavy tree cover than in regions with a low density of vegetation. TCI, VCI, PDI and MPDI are more similar to 3-month SPI data, but correlate weakly with station-based indices located in areas of high precipitation, higher soil permeability, large-scale agriculture and forests. PCI is more strongly correlated with short-term drought conditions in almost all climate divisions than other single indices. VCI would be more reliable if there were only red and NIR surface reflectance bands available, while VHI would be a better choice if only NDVI and LST data were available. Condition index-based drought indices (PCI, TCI, VCI, SMCI, VHI, SDCI, SDCI, MIDI, OVDI, OMDI) performed better than other categories of drought indices, and the use of time series analysis may be a contributing factor to this difference in performance.