Abstract:Seasonal climate forecasts (SCFs) can be used to identify appropriate risk management strategies and to reduce the sensitivity of rural industries and communities to climate risk. However, these forecasts have low utility among farmers in agricultural decision making, unless translated into a more understood portfolio of farm management options. Towards achieving this translation, we developed a mathematical programming model that integrates seasonal climate forecasts to assess ‘what-if?’ crop choice scenarios for famers. We used the Rayapalli village in southern India as a case study. The model maximises expected profitability at village level subject to available resource constraints. The main outputs of the model are the optimal cropping patterns and corresponding agricultural management decisions such as fertiliser, biocide, labour and machinery use. The model is set up to run in two steps. In the first step the initial climate forecast is used to calculate the optimal farm plan and corresponding agricultural management decisions at a village scale. The second step uses a ‘revised forecast’ that is given six weeks later during the growing season. In scenarios where the forecast provides no clear expectation for a dry or wet season the model utilises the total agricultural land available. A significant area is allocated to redgram (pigeon pea) and the rest to maize and paddy rice. In a forecast where a dry season is more probable, cotton is the predominant crop selected. In scenarios where a ‘normal’ season is expected, the model chooses predominantly cotton and maize in addition to paddy rice and redgram. As part of the stakeholder engagement process, we operated the model in an iterative way with participating farmers. For ‘deficient’ rainfall season, farmers were in agreement with the model choice of leaving a large portion of the agriculture land as fallow with only 40ha (total area 136ha) of cotton and subsistence paddy rice area. While the model crop choice was redgram in ‘above normal and wet seasons, only a few farmers in the village favoured redgram mainly because of high labour requirements, and the farmers perceptions about risks related to pests and diseases. This highlighted the discrepancy between the optimal cropping pattern, calculated with the model and the farmer's actual decisions which provided useful insights into factors affecting farmer decision making that are not always captured by models. We found that planning for a ‘normal’ season alone is likely to result in losses and opportunity costs and an adaptive climate risk management approach is prudent. In an interactive feedback workshop, majority of participating farmers agreed that their knowledge on the utility and challenges of SCF have highly improved through the participation in this research and most agreed that exposure to the model improved their understanding of the role of SCF in crop choice decisions and that the modelling tool was useful to discuss climate risk in agriculture.

An Online Optimization-Based Decision Support Tool for Small Farmers in India: Learning in Non-stationary Environments

Comparative Analysis of Multi-Agent Reinforcement Learning Policies for Crop Planning Decision Support

Optimizing Crop Planting Schedule Considering Planting Window and Storage Capacity

Optimizing Crop Management with Reinforcement Learning and Imitation Learning

Learning-based agricultural management in partially observable environments subject to climate variability

Farmer's Assistant: A Machine Learning Based Application for Agricultural Solutions

An Acquisition Based Optimised Crop Recommendation System with Machine Learning Algorithm

Stochastic model-based optimization of irrigation scheduling

Short and Medium Range Irrigation Scheduling Using Stochastic Simulation‐Optimization Framework With Farm‐Scale Ecohydrological Model and Weather Forecasts

Development of a growth model-based decision support system for crop management

A SWAT-based Reinforcement Learning Framework for Crop Management

A nonlinear mathematical model integrated with the CROPWAT decision support tool to optimize cropping patterns under different climate conditions

Improving the representation of smallholder farmers' adaptive behaviour in agent-based models: Learning-by-doing and social learning

Development of Growth Model-Based Decision Support System for Crop Management

Risk-averse Stochastic Optimization for Farm Management Practices and Cultivar Selection Under Uncertainty

CROPS: A Deployable Crop Management System Over All Possible State Availabilities

Optimizing Nitrogen Management with Deep Reinforcement Learning and Crop Simulations

Agricultural Risk Decision Support System for Resource-Poor Farmers in Burkina Faso, West Africa

Engaging farmers on climate risk through targeted integration of bio-economic modelling and seasonal climate forecasts

Integrated decision support for promoting crop rotation based sustainable agricultural management using geoinformatics and stochastic optimization