A Study On Spatial-temporal Sensitivity And Applicability Of Remote Sensing Based Drought Indices

Drought is considered to be one of the most complex,yet least understood,natural hazards,and is the cause of considerable environmental,societal,and economic problems.International interest in droughts has increased in recent decades due to their frequency and to resulting losses.Numerous remote sensing based drought indices have been developed to monitor drought conditions,as the development of the remote sensing technology.However,different drought indices have differing characteristics,and are suitable for specific environments.Studying the sensitivities and feasibilities of various drought indices can not only help us better understand the drought phenomenon,but also help us better understand the mechanism of establishing more reliable drought monitoring indices.What is more,the aim of most efforts is to establish drought indices with simple input data,and requiring only simple calculations.To address this issue,this paper was focused on the study area of different climate division regions in the continental of United States,analyzing the suitability of various drought indices.Main conclusions and results can be deduced as follows:(1)Studying drought phenomena in the Continental United States in 2011 and 2012 using various drought indicesComparison of the spatial distributions of data from July 2011 and July 2012 showed that the remotely sensed drought indices and the locally measured drought indices varied greatly,while the CCIs showed a stronger correlation with station based indices than single drought indices.The drought indices compared were capable of detecting the main drought event of July 2011 and July 2012 throughout the CONUS.The CCIs indicated larger areas to be under drought conditions than did the USDM data.The VUA-based SMCI showed greater correlations with the station-based drought indices than NSIDC-based SMCI,and the MODIS-based TCI showed stronger correlations with the station-based drought indices than VUA-based TCI in the CONUS.For the comparison of drought indices in regions of varying climate divisions,PCI is a good choice as a short-term drought index as it is correlated more strongly with station-based indices.VCI is a poor choice for monitoring drought in humid regions as it can be strongly affected by atmospheric changes.However,for low-latitude,semi-arid climates,VCI is more reliable than other methods.TCI should be avoided when monitoring desert environments,and SMCI should be avoided in regions covered by dense vegetation.The combined indices investigated had little variation among them,and empirical weighting was shown to be a suitable approach to combining a number of single indices.Compared with VegDRI,the combined indices showed limited improvement over single indices.Future development of combined indices could focus on empirical weighting based on climate.(2)Evaluating the utility of the VIUPD-derived vegetation condition index(VCI)for drought monitoring in the continental United StatesThe VIUPD-derived VCI had the strongest correlation with in situ drought indices in different sub-periods of the growing season and in a greater number of climate divisions,with higher r-values than the NDVI-derived VCI.This indicates that the VIUPD has the potential to replace the commonly used NDVI for monitoring drought conditions in different environments.Compared with the other single remote-sensing-based drought indices(TCI,PCI,and SMCI),the VIUPD-derived VCI was more suitable for longer-term drought monitoring,such as agricultural droughts,because it had stronger correlations with SPI-3,SPI-6,SPI-9,SPI-12,SPI-24,and PDSI.The VIUPD-derived VCI can also overcome some of the defects of SMCI for short-term drought monitoring in areas with a high density of vegetation cover.The VIUPD-derived VCI could also be combined with TCI.The VIUPD-derived VCI is recommended as the optimum remote-sensing-based single drought index for long-term drought over the CONUS or for other regions worldwide.(3)The sensitivity of satellite solar-induced chlorophyll fluorescence(SIF)for drought monitoring.When correlated with in-situ based drought indices,the SIF* data with higher spatial resolution did not show better performance than SIF from GOME-2 data.It is not the higher spatial resolution the better performance in monitoring drought conditions for SIF data.When comparing with MODIS data based VCI,the performances of SIF and SIF* were much worse.Even though the SIF* data was fused from GOME-2,MYD13C2,MOD16A2 and MYD11C3 data,it did not show very good results.(4)Monitoring 2011 and 2012 drought conditions in CONUS using MODIS data based BVI.From the overall performance,NDVI changes in the whole year is relatively weaker than the BVI.From the comparative analysis can be drawn,BVI to some extents,can be in response to the occurrence of drought events.When drought occurs BVIvalue will be reduced.In some cases,the trend of BVI will be inconsistent with the trend of drought,but on the whole,BVI changes with the drought situation is more active than NDVI changes.