Abstract Droughts pose a major risk to agricultural production. By comparing the outputs from an ecophysiological crop model (Sirius) with four drought severity indicators (DSI), a comparative assessment of the impacts of drought risk on wheat yield losses has been evaluated under current (baseline) and two future climate scenarios. The rationale was to better understand the relative merits and limitations of each approach from the perspective of quantifying agricultural drought impacts on crop productivity. Modelled yield losses were regressed against the highest correlated variant for each DSI. A cumulative distribution function of yield loss for each scenario (baseline, near and far future) was calculated as a function of the best fitting DSI (SPEI-5July) and with the equivalent outputs from the Sirius model. Comparative analysis between the two approaches highlighted large differences in estimated yield loss attributed to drought, both in terms of magnitude and direction of change, for both the baseline and future scenario. For the baseline, the average year differences were large (0.25 t ha−1 and 1.4 t ha−1 for the DSI and Sirius approaches, respectively). However, for the dry year, baseline differences were substantial (0.7 t ha−1 and 2.7 t ha−1). For the DSI approach, future yield losses increased up to 1.25 t ha−1 and 2.8 t ha−1 (for average and dry years, respectively). In contrast, the Sirius modelling showed a reduction in future average yield loss, down from a baseline 1.4 t ha−1 to 1.0 t ha−1, and a marginal reduction for a future dry year from a baseline of 2.7 t ha−1 down to 2.6 t ha−1. The comparison highlighted the risks in adopting a DSI response function approach, particularly for estimating future drought related yield losses, where changing crop calendars and the impacts of CO2 fertilisation on yield are not incorporated. The challenge lies in integrating knowledge from DSIs to understand the onset, extent and severity of an agricultural drought with ecophysiological crop modelling to understand the yield responses and water use relations with respect to changing soil moisture conditions.

中文翻译：

---

评估英格兰未来的干旱风险和小麦产量损失

摘要 干旱给农业生产带来了重大风险。通过将生态生理作物模型 (Sirius) 的输出与四个干旱严重程度指标 (DSI) 进行比较，在当前（基线）和两个未来气候情景下评估了干旱风险对小麦产量损失的影响的比较评估。其基本原理是从量化农业干旱对作物生产力的影响的角度更好地了解每种方法的相对优点和局限性。针对每个 DSI 的最高相关变量对建模的产量损失进行回归。每个情景（基线、近期和远期）的产量损失累积分布函数计算为最佳拟合 DSI (SPEI-5July) 的函数，并使用 Sirius 模型的等效输出。两种方法之间的比较分析突出显示，对于基线和未来情景，干旱造成的估计产量损失在幅度和变化方向方面存在巨大差异。对于基线，平均年份差异很大（DSI 和 Sirius 方法分别为 0.25 t ha-1 和 1.4 t ha-1）。然而，对于干旱年份，基线差异很大（0.7 t ha-1 和 2.7 t ha-1）。对于 DSI 方法，未来产量损失增加至 1.25 t ha-1 和 2.8 t ha-1（分别为平均和干旱年份）。相比之下，Sirius 模型显示未来平均产量损失减少，从基线 1.4 t ha-1 降至 1.0 t ha-1，未来干旱年从基线 2.7 t ha-1 略有减少到 2.6 吨公顷-1。比较突显了采用 DSI 响应函数方法的风险，特别是在估算未来与干旱相关的产量损失时，其中未考虑改变作物日历和 CO2 施肥对产量的影响。挑战在于整合来自 DSI 的知识以了解农业干旱的发生、程度和严重程度，以及生态生理作物模型，以了解与土壤水分条件变化有关的产量响应和用水关系。