Despite the remarkable improvements in skillful weather forecasts, their uses in irrigation water management at a farm scale are still limited. This is attributable to the scale mismatch between weather and hydrologic models as well complexities in farmscale ecohydrological processes. Here, we have developed a simulation‐optimization algorithm for minimizing irrigation water application using short to medium range forecasts by determining the conditional probability density functions of the rainfall and subsequently the soil moisture for the days in forecast range. With the forecasted soil moisture information, the model ensures that the probability of crops undergoing water stress is less than a prescribed threshold. The optimization model includes a farmscale ecohydrological model, which computes daily evapotranspiration, runoff, and leakage based on a probabilistic description of rainfall through Monte‐Carlo simulations. After calibrating and validating the ecohydrological model with past data obtained from the farmers, this proposed optimization framework was employed to test the outcome for two pilot sites in Nashik, Maharashtra, India. We found that using the proposed framework, irrigation water use can be reduced by 10%–30% as compared to that resulting from the conventional strategies used by the farmers, without significant loss in crop yields, by almost always maintaining the soil moisture at or above the prescribed threshold. Considering that irrigation accounts for over 80% of the total water use worldwide, the value of such an approach as a decision‐support tool for irrigation optimization is self‐evident. We also posit that the co‐production of this tool with the farmers increases its usability and credibility.

中文翻译：

---

结合农场规模生态水文模型和天气预报的随机模拟优化框架进行中短期灌溉计划

尽管熟练的天气预报取得了显着改善，但它们在农场规模的灌溉水管理中的使用仍然受到限制。这归因于天气和水文模型之间规模的不匹配以及农场规模的生态水文过程的复杂性。在这里，我们开发了一种模拟优化算法，可通过确定降雨范围以及随后预报范围内各天的土壤湿度的条件概率密度函数，使用中短期预报来最大程度地减少灌溉用水。利用预测的土壤水分信息，该模型可确保农作物遭受水分胁迫的概率小于规定的阈值。该优化模型包括一个农场规模的生态水文模型，该模型可计算每日的蒸散量，径流量，通过蒙特卡洛模拟对降雨的概率描述得出的漏失和渗漏。在利用从农民那里获得的过去数据对生态水文模型进行校准和验证之后，采用了该拟议的优化框架来测试印度马哈拉施特拉邦纳西克两个试点的结果。我们发现，使用提议的框架，与农民采用的常规策略相比，灌溉水的使用量几乎可以始终保持在或高于规定的阈值。考虑到灌溉占全球总用水量的80％以上，这种作为灌溉优化决策支持工具的方法的价值是不言而喻的。