Crop Yield Prediction Techniques using Remote Sensing Data

Abstract

Crop yield prediction is an art of forecasting the yield of crop before harvesting. Prediction of crop yield will be very useful for the government to make food policies, market price, import and export policies and proper warehousing well in time. The socio-economical impact of crop loss due to any natural disaster i.e. flood, drought can be minimized and humanitarian food assistance can be planned. The paper present a literature survey of various stastical method, empirical models,artificial neural network and machine learning regression techniques which are used with the data provided by the satellites. Many models are developed and results calculated are compared with the benchmark models are also presented.

Standard Methods and Innovative Techniques in Phytopathometry, the Art and Science of Plant Disease Measurement

“…In China, Cheng et al (2022) affirmed that field observation and traditional linear regression crop yield models gave poor approximations of winter wheat yield. Remote sensing (RS) on the other hand can be used to effectively assess crop conditions timely and reliably (Table 1); and estimate its yield based on very high resolutions of the spectral, textual and structural characteristics of the crop than conventional approaches (Singh and Kumar, 2020;Seklar, 2020;Gumma et al, 2022;Udemy, 2023). This technique allows for large scale continuous crop and field monitoring throughout the growth cycle per season (Khaki et al, 2021).…”

“…RS estimates crop yield by capturing crop biomass accumulation through pre-harvest vegetation index, temperature condition index, enhanced vegetation index, leaf area index, land surface temperature and normalized difference vegetation Index (NDVI) which are strong indicators of total greenness in plants (Abul-Jabbar et al, 2004;Seklar, 2020;Campos et al, 2023) and these satellite-based techniques outperformed traditional vegetation indices (Cheng et al, 2022). RS techniques used for crop monitoring and yield prediction include multispectral, hyperspectral data, radar and lidar imagery, random forest and decision tree regressor and gradient boosting predictors, AI and ML, GIS, GPS (loraiswamy et al, 2000;Singh and Kumar, 2020;Ilyas et al, 2023) and may be applied to more than one crop simultaneously (Khaki et al, 2021). These workers utilized this approach to estimate the yield of corn and soybean at the same time.…”