Research Projects

The uncertainties of non-catastrophic weather events can cause major concern for the farmers and policy makers. It is well documented that water stress, triggered by increasing temperatures, reduction in number of rainy days coupled with increasing length of dry spells have adverse impact on major cereal crops e.g. rice and wheat production in India. Notwithstanding, crop yield are directly influenced by the climatic pattern. Statistical models based on historical data on crop yields and weather provide a common alternative to process-based models. But as the time horizon elongates into the future, it becomes difficult to accurately forecast and make decisions without considering climate variables. In terms of methodology, the standard parametric or their combinatory hybrid structure possesses obvious limitation of being data and distribution dependent. The recent scenario highlighted that, climatic variability has greater impact on agricultural production. The statistical approach for yield prediction, based on climatic variability could help us reveal differences in factors’ effect for better crop production management. A quantitative understanding of crop responses to climate requires the development of models for various characteristics of crop by taking into account its time series behaviour along with exogenous climate factors. It is well known that climate variability is one of the main factors causing yields to change from year to year. Indeed, farmers’ expectations of crop yields are mostly based on subjective weather observations. When forecasting is applied to the dynamic behaviour of crop yield, it should be able to take advantage not only of the historical data, but also of the impact of various driving forces from the external environment. Remote sensing data act as important regressor variables for crop yield forecasting. The primary advantage of radar data over optical data is that it is not affected by weather. Radar sensors also provide information that is complementary to that contained in visible-infrared imagery. In this study, we will combine the benefits of SAR data and optical data with different field and weather parameters to improve the model's prediction accuracy. Under FASAL project, National/State/District level crop production forecasts are calculated using multi-date remote sensing data with different parametric techniques. In PMFBY scheme, crop health/yield loss estimation technique has been improved by using satellite data using in coordination with concerned states. In these projects, they mainly used parametric models such as regression technique. But remote sensing data are fuzzy and noisy due to its characteristic and its capturing technique. Wavelet technique along-with optimization based ensembled model has the potential to handle the noisy data and increase the forecasting efficiency. In the proposed project, novel wavelet-based AI models will be developed to encounter the lacuna of parametric model.