Multi-sensor remote sensing of the extent and persistence of the 2005 Amazon drought
Monday, 15 December 2014
A strong drought occurred in the southwestern Amazon Basin forest in 2005. Impact of the drought was detected in optical/NIR vegetation indices (e.g., MODIS EVI), but interpretation of this data has been mixed. A significant positive anomaly in MODIS land surface temperature (LST), and a decline in morning-overpass microwave (Ku-band) backscatter have also been reported. It has been shown that the regional aggregate dry season morning-overpass microwave backscatter had a negative anomaly that persisted for several years. To refine and focus the analysis of microwave backscatter data associated with this drought, we examine spatial and temporal correlations between QuikSCAT microwave backscatter data and MODIS land surface temperature (LST) data – MOD11A1 (Terra) and MYD11A1 (Aqua), daily 1-km, altogether 4 LST observations per day. We use the MODIS LST anomaly as an ‘impact classifier’ and evaluate the strength and persistence of the microwave backscatter anomaly for different degrees of ‘drought thermal impact’, after masking for deforestation, which also causes a significant decrease in microwave backscatter. To explore potential crown mortality, we also evaluate changes in MODIS-derived surface composition, expressed as a change in green vegetation and non-photosynthetic vegetation in drought-impacted areas.