Title :
A thirteen-year analysis of drought in the horn of Africa with MODIS NDVI and NWDI measurements
Author :
Taeyoung Choi ; Qu, J.J. ; Xiaoxiong Xiong
Author_Institution :
MODIS Calibration Support, Sigma Space Corp., Lanham, MD, USA
Abstract :
Satellite remotely sensed visible and infrared surface reflectance measurements are utilized to detect long-term changes of vegetation drought conditions in the horn of Africa from 2000 to early 2013. The horn of Africa area has been experiencing drought problems due to the warming of the Indian Ocean, which produces hot and dry air that descend across the area. A region of interests (ROI) is defined based on MODIS land surface reflectance 8-day L3 Global 500m (MOD09A1) product, which covers most of the horn of Africa area. The ROI is defined by latitudes from 0 to 10 degrees and longitudes from 40 to 51 degrees excluding the ocean area. Based on a land classification scheme, the ROI is composed of mostly open shrub land, barren land and some grass land. The MODIS reflective solar band observations are utilized to measure well-known indexes called normalized difference vegetation index (NDVI) and normalized difference water index (NDWI). A linear regression of the filtered data shows that the NDWI has non-negligible degradation over the thirteen years observation period approximately from -0.05 to -0.10, whereas NDVI remained the same level. The NDWI long-term degradation indicates the portion of soil over vegetation has been increased because of the dry condition in the ROI. On top of the degradation, a recent NDWI value in early 2013 shows a significant drop from the linear trend, which signifies that there is a recent severe drought condition on-going in the horn of Africa. Utilizing hypothesis testing, we show that the NDWI provides more statistically meaningful information than the NDVI. It suggests that NDWI is a better drought indicator over in the Horn of Africa areas. The results also show that the ROI is mostly affected by the long-term drought condition especially during the second yearly growing season from September to December.
Keywords :
hydrology; land surface temperature; rain; remote sensing; soil; vegetation; AD 2000 to 2013; Indian Ocean warming; MOD09A1 product; MODIS NDVI measurement; MODIS NWDI measurement; MODIS land surface reflectance; MODIS reflective solar band observations; barren land; drought analysis; drought indicator; dry air; filtered data linear regression; grass land; horn of Africa area; hot air; land classification scheme; normalized difference vegetation index; normalized difference water index; open shrub land; region-of-interests; satellite remotely sensed infrared surface reflectance measurement; satellite remotely sensed visible surface reflectance measurement; soil over vegetation portion; vegetation drought conditions; Africa; Indexes; Land surface; MODIS; Monitoring; Reflectivity; Vegetation mapping; Drought; Horn of Africa; MODIS; NDVI; NDWI; Student´s T-test;
Conference_Titel :
Agro-Geoinformatics (Agro-Geoinformatics), 2013 Second International Conference on
Conference_Location :
Fairfax, VA
DOI :
10.1109/Argo-Geoinformatics.2013.6621926