Crop yield forecasting before harvesting is critical for the creation, implementation, and optimization of policies related to food safety as well as for agro-product storage and marketing. Crop growth and development are influenced by the weather. Therefore, models using weather variables can provide reliable predictions of crop yields. It can be tough to select the best crop production forecasting model. Therefore, in this study, five alternative models, viz., stepwise multiple linear regression (SMLR), an artificial neural network (ANN), the least absolute shrinkage and selection operator (LASSO), an elastic net (ELNET), and ridge regression, were compared in order to discover the best model for rice yield prediction. The outputs from individual models were used to build ensemble models using the generalized linear model (GLM), random forest (RF), cubist and ELNET methods. For the previous 21 years, historical rice yield statistics and meteorological data were collected for three districts under three separate agro-climatic zones of Chhattisgarh, viz., Raipur in the Chhattisgarh plains, Surguja in the northern hills, and Bastar in the southern plateau. The models were calibrated using 80% of these datasets, and the remaining 20% was used for the validation of models. The present study concluded that for rice crop yield forecasting, the performance of the ANN was good for the Raipur ($R_{cal}^2$ = 1, $R_{val}^2$= 1 and $RMS{E}_{cal}$ = 0.002, $RMS{E}_{val}$ = 0.003) and Surguja ($R_{cal}^2$ = 1, $R_{val}^2$= 0.99 and $RMS{E}_{cal}$ = 0.004, $RMS{E}_{val}$ = 0.214) districts as compared to the other models, whereas for Bastar, ELNET ($R_{cal}^2$ = 90, $R_{val}^2$= 0.48) and LASSO ($R_{cal}^2$ = 93, $R_{val}^2$= 0.568) performed better. The performance of the ensemble model was better compared to the individual models. For Raipur and Surguja, the performance of all the ensemble methods was comparable, whereas for Bastar, random forest (RF) performed better, with R² = 0.85 and 0.81 for calibration and validation, respectively, as compared to the GLM, cubist, and ELNET approach.

Early information exchange regarding predicted crop production could play a role in lowering the danger of food insecurity. In this study total six multivariate models were developed using past time series yield data and weather indices viz. SMLR, PCA-SMLR, ANN, PCA-ANN, SMLR-ANN and PCA-SMLR-ANN for three major soybean producing districts of Uttarakhand viz. Almora, Udham Singh Nagar and Uttarkashi. Further analysis was done by fixing 80% of the data for calibration and the remaining dataset for validation to predict soybean yield. Phenology wise average values were computed using the daily weather data. These average values are subsequently employed in the computation of both weighted and unweighted weather indices. The PCA-SMLR-ANN, SMLR-ANN and PCA-ANN models were found to be the best soybean yield predictor model for Almora, Udham Singh Nagar and Uttarkashi districts, respectively. The overall ranking based on the performances of the models for all locations can be given as: SMLR-ANN > PCA-ANN > PCA-SMLR-ANN ≈ ANN > PCA-SMLR > SMLR. The study results indicated that hybrid models outperformed the individual models well for all the study regions.

Timely identification and management of plant diseases are critical for sustaining crop yields and ensuring food security. This study proposes an innovative approach to detect Rhizoctonia aerial blight (RAB) caused by Rhizoctonia solani, a prevalent disease-causing substantial loss in soybean crops in Uttarakhand, India. By integrating smartphone imageries into sophisticated algorithms, our automated system offers a scalable solution for disease detection. We evaluated nine machine learning algorithms, including logistic regression, Support Vector Machine (SVM), VGG-16 (with and without augmentation), ResNet-18 (with, without augmentation and larger image sizes) and ResNet-34 (with and without augmentation), for disease classification. Results demonstrate the effectiveness of our approach, with classification accuracies ranging from 66.77% (logistic regression) to 95.64% (ResNet-34). Particularly, the ResNet-34 model with data augmentation achieved the highest accuracy, showcasing its potential for accurate disease detection. Leveraging advanced computer technologies and smartphone imaging, this study presents a practical solution for enhancing crop management practices and minimizing yield losses due to diseases. The source code for our implementation is available in supportive file.

Groundnut is a key oilseed crop in the world and India is one of the largest groundnuts producing country in terms of area and yield. Keeping that in view, five models were developed for five districts of Andhra Pradesh to forecast the groundnut yield viz., Stepwise Multiple Linear Regression (SMLR), Ridge regression, Least Absolute Shrinkage and Selection Operator (LASSO), Elastic Net (ELNET) and Artificial Neural Network (ANN). The historical data on the weather parameters are obtained from NASA POWER web portal and groundnut yields for these districts of the state during both Kharif and Rabi seasons obtained through Season and Crop Report, Government of Andhra Pradesh for the period, 2001 to 2020. In total 30 weather indices were generated through five weather variables. The assessment of models was done by fixing 75 % of the data for calibration and left 25 % data for validation. The findings inferred that based on the values of R², RMSE, nRMSE and EF, Ridge regression, ELNET and ANN models showed better performance for Ananthapur, Chittoor and Kadapa districts and SMLR and LASSO models showed better performance for Kurnool and Nellore districts during both Kharif and Rabi seasons at calibration and validation stages.