18 November 2015 Rice yield estimation using Landsat ETM+ Data
Author Affiliations +
Paddy rice areas in Larkana district in Sindh province, Pakistan, were mapped over eight years. Landsat 7 ETM+ satellite imagery was classified for rice areas using training data collected through visual interpretation and using a bagged decision tree approach. Within the rice areas, we estimated yield for the 2013 season using regression models based on Landsat-derived normalized difference vegetation index (NDVI) and ratio vegetation index (RVI) values against historic, reported yield values. The annual cropped rice area estimated from satellite imagery was between 19% and 24% lower than the area reported by the Crop Reporting Service, Sindh. A positive and strong relationship with coefficient of determination (R2) of 0.94 was observed between the reported rice crop yield and NDVI at the peak of the growing season for the years 2006 to 2013. A fair relation (R2=0.875) between rice crop yield and RVI was observed for the same years. A strong relationship between observed and predicted rice production with model efficiency=0.925, mean bias error=−85,016t, and RMSE=80,726t was obtained. Thus, Landsat ETM+ has a high potential for estimating rice yield and production at the district level in Pakistan and elsewhere.
© 2015 Society of Photo-Optical Instrumentation Engineers (SPIE)
Altaf Ali Siyal, Altaf Ali Siyal, Jan Dempewolf, Jan Dempewolf, Inbal Becker-Reshef, Inbal Becker-Reshef, } "Rice yield estimation using Landsat ETM+ Data," Journal of Applied Remote Sensing 9(1), 095986 (18 November 2015). https://doi.org/10.1117/1.JRS.9.095986 . Submission:

Back to Top