Operational semi-physical spectral-spatial wheat yield model development

Abstract

ABSTRACT:Spectral yield models based on Vegetation Index (VI) and the mechanistic crop simulation models are being widely used for crop yield prediction. However, past experience has shown that the empirical nature of the VI based models and the intensive data requirement of the complex mechanistic models has limited their use for regional and spatial crop yield prediction especially for operational use. The present study was aimed at development of an intermediate method based on the use of remote sensing and the physiological concepts such as the photo-synthetically active solar radiation (PAR) and the fraction of PAR absorbed by the crop (fAPAR) in Monteith's radiation use efficiency based equation (Monteith, 1977) for operational wheat yield forecasting by the Department of Agriculture (DoA). Net Primary Product (NPP) has been computed using the Monteith model and stress has been applied to convert the potential NPP to actual NPP. Wheat grain yield has been computed using the actual NPP and Harvest index. Kalpana-VHRR insolation has been used for deriving the PAR. Maximum radiation use efficiency has been collected from literature and wheat crop mask was derived at MNCFC, New Delhi using RS2-AWiFS data. Water stress has been derived from the Land Surface Water Index (LSWI) which has been derived periodically from the MODIS surface reflectance data (NIR and SWIR1). Temperature stress has been derived from the interpolated daily mean temperature. Results indicated that this model underestimated the yield by 3.45 % as compared to the reported yield at state level and hence can be used to predict wheat yield at state level. This study will be able to provide the spatial wheat yield map, as well as the district-wise and state level aggregated wheat yield forecast. It is possible to operationalize this remote sensing based modified Monteith's efficiency model for future yield forecasting with around 0.15 t ha-1 RMSE at state level.

Statistical Machine Learning Methods and Remote Sensing for Sustainable Development Goals: A Review

“…Tripathy et al (2014) also use a semi-empirical method, which incorporates physiological measures, spectral measures, and spatial features for estimating wheat yield [21]. The authors note that while spectral (empirical) and physics-based (mechanistic) models based on vegetation indices are widely used, they are respectively limited by being data-intensive and complex.…”

Evaluation of Pre-Harvest Production Forecasting of Mustard Crop in Major Producing States of India, Under Fasal Project

“…District level R&M crop yield was estimated using three different procedures: i) Agro-meteorological regression models (Singh et al, 2017), ii) Remote sensing index (VCI) based empirical models (Dubey et al, 2018) and iii) Semi physical Model (Tripathy et al, 2014;Chaurasiya et al, 2017). First approach was used by IMD (in collaboration with state agricultural universities), the second and third approaches were used by MNCFC.…”

Optimizing Crop Yield Estimation through Geospatial Technology: A Comparative Analysis of a Semi-Physical Model, Crop Simulation, and Machine Learning Algorithms

“…The model operates under the assumption of constant radiation conversion efficiency and harvest index, overlooking factors such as nitrogen deficiency influenced by water stress and temperature stress during reproductive and grain-filling stages. While nitrogen deficiency is addressed through fAPAR, the primary challenges affecting the accuracy of yield estimations are water and temperature stress [54].…”