Wheat yield estimation using assimilation of remotely sensed information into a crop model

Abstract

Accurate and timely information of crop area and its production plays an important role to summarize the upcoming situation of market supply and demand. It also provides the foundation to policy makers, stakeholders, government planners and agribusiness community for ideal management of their interests. Remote sensing provides the information about discrete time instant event over a larger area while crop models explain continuous crop growth status on daily basis as function of weather, soil and management. This study was planned with the objective to estimate the area under wheat (Triticum aestivum L.) cultivation in Faisalabad district using satellite data and estimate its grain yield by assimilating the remotely sensed information into the CERES-Wheat model under spatiotemporal heterogenous conditions and variable management practices. Before applying the crop model at regional scale, it was calibrated using a field experimental data. In this experiment, three wheat cultivars (Punjab-2011, Aas-2011 and Galaxy-2013) were grown under six levels of deficit irrigation. The experiment was conducted with three replications under strip plot arrangement. Results of experiment proved that stem elongation stage is the least sensitive to drought, and grain formation stage of wheat crop is more sensitive to drought as compared to booting and stem elongation stages. Same grain yield can be achieved by applying 75% of irrigation compared to farmer irrigation practice. Wheat cultivars Punjab-2011 and Galaxy-2013 produced significantly more grain yield than wheat cultivar Aas-2011 and both cultivars are statistically at par to each other. CERES-Wheat model was calibrated and validated using data of field experiment to simulate the crop growth parameters and soil water balance. Model performed very well and simulated results were close to the observed data. Landsat based temporal satellite images were used to estimate the area under wheat cultivation and water index of crop during the crop growth period. Green (G), red (R), near infrared (NIR), shortwave infrared 1 (SWIR1) and shortwave infrared 2 (SWIR2) bands, and NDVI and NDWI indices of both satellites landsat7 and landsat8 were used to develop the metrics (minimum, 10, 25, 50, 75, 90, maximum and mean). Median NDWI derived from landsat based temporal images were used to quantify the applied volume of irrigation in wheat cultivated area of Faisalabad district. Calibrated crop model was used to simulate the wheat growth and estimate the wheat yield. Estimated wheat yield by crop model is 5% lower than the estimate of Crop Reporting Service (CRS), Punjab. It shows that assimilation of remotely sensed information into a crop model can be used to estimate the wheat productivity with good accuracy.