Use of ordinal logistic regression in crop yield forecasting

The performance of ordinal logistic regression and discriminant function analysis has been compared in crop yield forecasting of wheat crop for Kanpur district of Uttar Pradesh. Crop years were divided into two or three groups based on the detrended yield. Crop yield forecast models have been developed using probabilities obtained through ordinal logistic regression along with year as regressors and validated using subsequent years data. In discriminant function approach two types of models were developed, one using scores and another using posterior probabilities. Performance of the models obtained at different weeks was compared using Adj R2, PRESS (Predicted error sum of square), number of misclassifications and forecasts were compared using RMSE (Root Mean Square Error) and MAPE (Mean absolute percentage error) of forecast. Ordinal logistic regression based approach was found to be better than discriminant function analysis approach.

Crop yield forecasting of paddy and sugarcane through modified Hendrick and Scholl technique for south Gujarat

Regression models by modified Hendrick and Scholl technique were developed on paddy and sugarcane for six districts of south Gujarat. The data on the yield and weather parameters were analyzed for 27 years. The 25 year data was used for development of the model. The validation of model was done using data set of 2010 and 2011. The stepwise regression analysis was executed by trial and error method to obtain the finest combination of predictors, significant at 5% level. The multiple regression techniques was used for fitting of the model and decided best by highest R2 and lowest percent error. All crop yield forecasting models gave good estimates and produced error percent within acceptable range. Analysis revealed that the model error percent of paddy and sugarcane for respective crop growing districts were -10.0 to 8.1% and -12.2 to 1.5% respectively. Crop yield forecasting for year 2012 based on validated model was made for the districts of Navsari, Surat, Bharuch, Valsad, Narmada and Tapi.

Integration of Sentinel-3 and MODIS Vegetation Indices with ERA-5 Agro-meteorological Indicators for Operational Crop Yield Forecasting

Timely crop yield forecasts at national level are substantial to support food policies, to assess agricultural production and to subsidize regions affected by food shortage. This study presents an operational crop yield forecasting system for Poland that employs freely available satellite and agro-meteorological products provided by the Copernicus programme. The crop yield predictors consist of: (1) vegetation condition indicators provided daily by Sentinel-3 OLCI (optical) and SLSTR (thermal) imagery, (2) a backward extension of Sentinel-3 data (before 2018) derived from cross-calibrated MODIS data, (3) air temperature, total precipitation, surface radiation, and soil moisture derived from ERA-5 climate reanalysis generated by the European Centre for Medium-Range Weather Forecasts. The crop yield forecasting algorithm is based on thermal time (growing degree days derived from ERA-5 data) to better follow the crop development stage. The recursive feature elimination is used to derive an optimal set of predictors for each administrative unit, which are ultimately employed by the Extreme Gradient Boosting regressor to forecast yields using official yield statistics as a reference. According to intensive leave-one-year-out cross validation for 2000–2019 period, the relative RMSE for NUTS-2 units are: 8% for winter wheat, and 13% for winter rapeseed and maize. Respectively, for the LAU units it equals 14% for winter wheat, 19% for winter rapeseed, and 27% for maize. The system is designed to be easily applicable in other regions and to be easily adaptable to cloud computing environments (such as DIAS or Amazon AWS), where data sets from the Copernicus programme are directly accessible.

Developing decision support systems for crop yield forecasts

This chapter discusses existing yield forecasting systems in which the yield forecasts are driven by integration of different data sources, such as output of crop modeling, remote sensing and gridded climate datasets. It first provides overviews of the two predominant modeling approaches— crop simulation modeling and statistical modeling— to forecasting crop yield, with an emphasis on their respective use for operational crop yield forecasting systems. The chapter then briefly describes the accuracy and lead time of the existing yield forecasting models. Lastly, it provides a case study that integrates digital tools, field surveys, and crop modeling to provide on-time maize yield forecasts in small fields in Tanzania. The chapter concludes with a summary and future perspectives for research.

USA Crop Yield Estimation with MODIS NDVI: Are Remotely Sensed Models Better Than Simple Trend Analyses?

Crop yield forecasting is performed monthly during the growing season by the United States Department of Agriculture’s National Agricultural Statistics Service. The underpinnings are long-established probability surveys reliant on farmers’ feedback in parallel with biophysical measurements. Over the last decade though, satellite imagery from the Moderate Resolution Imaging Spectroradiometer (MODIS) has been used to corroborate the survey information. This is facilitated through the Global Inventory Modeling and Mapping Studies/Global Agricultural Monitoring system, which provides open access to pertinent real-time normalized difference vegetation index (NDVI) data. Hence, two relatively straightforward MODIS-based modeling methods are employed operationally. The first model constitutes mid-season timing based on the maximum peak NDVI value, while the second is reflective of late-season timing by integrating accumulated NDVI over a threshold value. Corn model results nationally show the peak NDVI method provides a R2 of 0.88 and a coefficient of variation (CV) of 3.5%. The accumulated method, using an optimally derived 0.58 NDVI threshold, improves the performance to 0.93 and 2.7%, respectively. Both these models outperform simple trend analysis, which is 0.48 and 7.4%, correspondingly. For soybeans the R2 results of the peak NDVI model are 0.62, and 0.73 for the accumulated using a 0.56 threshold. CVs are 6.8% and 5.7%, respectively. Spring wheat’s R2 performance with the accumulated NDVI model is 0.60 but just 0.40 with peak NDVI. The soybean and spring wheat models perform similarly to trend analysis. Winter wheat and upland cotton show poor model performance, regardless of method. Ultimately, corn yield forecasting derived from MODIS imagery is robust, and there are circumstances when forecasts for soybeans and spring wheat have merit too.