Land management for crop production is an essential human activity that supports life on Earth. The main challenge to be faced by the agriculture sector in coming years is to feed the rapidly growing population while maintaining the key resources such as soil fertility, efficient land use, and water. Climate change is also a critical factor that impacts agricultural production. Among others, a major effect of climate change is the potential alterations in the growth cycle of crops which would likely lead to a decline in the agricultural output. Due to the increasing demand for proper agricultural management, this study explores the effects of meteorological variation on wheat yield in Chakwal and Faisalabad districts of Punjab, Pakistan and used normalised difference vegetation index (NDVI) as a predictor for yield estimates. For NDVI data (2001-14), the NDVI product of Moderate Resolution Imaging spectrometer (MODIS) 16-day composites data has been used. The crop area mapping has been realised by classifying the satellite data into different land use/land covers using iterative self-organising (ISO) data clustering. The land cover for the wheat crop was mapped using a crop calendar. The relation of crop yield with NDVI and the impact of meteorological parameters on wheat growth and its yield has been analysed at various development stages. A strong correlation of rainfall and temperature was found with NDVI data, which determined NDVI as a strong predictor of yield estimation. The wheat yield estimates were obtained by linearly regressing the reported crop yield against the time series of MODIS NDVI profiles. The wheat NDVI profiles have shown a parabolic pattern across the growing season, therefore parabolic least square fit (LSF) has been applied prior to linear regression. The coefficients of determination (R2) between the reported and estimated yield was found to be 0.88 and 0.73, respectively, for Chakwal and Faisalabad. This indicates that the method is capable of providing yield estimates with competitive accuracies prior to crop harvest, which can significantly aid the policy guidance and contributes to better and timely decisions.
Climate change; Yield estimation; Regression; Land cover; Mapping