Possibilities of Sorghum Yield Estimation Using Remote Sensing in Rain-fed Farming,Gedarif State, Sudan
【摘要】：Sudan is still using traditional methods for estimate crops yield depend on direct field survey, whereas there are available attractive techniques tested in many areas over the world not only with high efficiency, but also cost-effective and quick for the same purpose. The aim of this research is to investigate the utility of using remotely sensed data of Moderate-resolution Imaging Spectro-radiometer (MODIS) satellite for estimating Sorghum yield in Rain-fed farming in Gedarif State, Sudan, where Gedarif State provides about65%of Sudan main food crop "Sorghum", besides, sensing the difficulties may encounter the localization of the technique in the area.
Rain-fed farming in Gedarif State is considered as hard area for conducting remote sensing studies because of high variation in yield, farm sizes, and precipitation. In addition, references data are limited to four years2007-2010.
The study used simple linear regression model investigation two methods; the first Method included, Method1-1:the study measured the NDVI of Sorghum (in vegetation period) for the years2007-2008only. The result was low relatively0.50. The large area and high variation in yield may have caused to this result. Method1-2:Rainfall added to the NDVI, the model by then has a significant performance and R2increased to0.70for same term above.
In Method2:the large area of Gedarif State has reclassified to smaller partitioning ones (Strata Sample), two, in order to decrease the mentioned high variation. As a result, Gedarif Rain-fed farming is reclassified to North and South of Gedarif according to Rainfall precipitation. This reclassification necessarily influenced in different ways by other factors beyond the Rainfall, such as localities level borders, farming sizes, and management systems. The result was positive in most cases. R2recorded a good performance in North and South of Gedarif in year2007, and North of Gedarif in2008, R2=0.60,0.68, and0.59, for previous cases respectively. Only South of Gedarif2008recorded poor performance=0.49.The interpretation was suggested for the high infection of Southern area of Gedarif by weeds. Weeds are detected by sensor as high NDVI, where actual yield is low. The result of South Gedarif could caused to the low R2for the two years together2007-2008to0.50.
The final model represent four years,2007-2010recorded R2=0.70. The availability of time series reference (to more than four years) could enhance the result, yet in the same time affect the reality of study.