Assessing the relationship between soil moisture drought and cereal yield anomalies in Europe

Drought can arguably be considered the most important natural hazard affecting agricultural production worldwide. In rainfed crops, in particular, severe soil water deficit conditions can have direct impacts on crop yields, negatively affecting the local economy. Among rainfed crops, cereals are the most prominent production in Europe, accounting for about 20% of global production.

In this study, soil moisture drought conditions modelled following a global scale hydrological model (LISFLOOD) are used to explain cereal yield anomalies recorded over European regions (NUTS2) by Eurostat for the period 1991-2023. Due to the spatio-temporal mismatch between yield records (annual, over NUTS2 regions) and modelled soil moisture (daily, over a regular grid), different strategies are tested to assess the relationship between the two quantities. By focusing on the years affected by drought conditions, and the consequent expected reduction in yield, ranked zero-clustered correlation metrics are used to quantify the correspondence.

Over most of the regions, a positive and significant correlation between drought occurrence and yield reduction is observed, even if this is not the case for a few of the study regions. Overall, the temporal aggregation of soil moisture data over different seasons seems to play a major role in strengthening/weakening the relationship between soil moisture drought and yield reduction, with notable spatial patterns in the outcome. The typical European growing season, April-September, corresponds to the optimal case in most of the regions, but both earlier and later seasons (as well as shorter ones) are also observed in a non-negligible fraction of cases.  

A method to optimize the best aggregation strategy is proposed, by jointly minimizing the number of different solutions and maximizing the rank correlation. This optimization aims at providing a simple approach that can be used to infer the expected yield reductions given the antecedent modelled soil moisture status across European regions.

Acknowledgements: This work is partially funded by the European Union under the HORIZON-CL4-2023-SPACE-01-32 project “Strengthening Extreme Events Detection for Floods and Droughts” (SEED-FD), CUP: D43C23003660006 - 2023.