Analysis of soil moisture memory from observations in Europe

Soil moisture is known to show distinctive persistence characteristics compared to other quantities in the climate system. As soil moisture is governing land‐atmosphere feedbacks to a large extent, its persistence can provide potential to improve seasonal climate predictions. So far, many modeling studies have investigated the nature of soil moisture memory, with consistent, but model‐dependent results. This study investigates soil moisture memory in long‐term observational records based on data from five stations across Europe. We investigate spatial and seasonal variations in soil moisture memory and identify their main climatic drivers. Also, we test an existing framework and introduce an extension thereof to approximate soil moisture memory and evaluate the contributions of its driving processes. At the analyzed five sites, we identify the variability of initial soil moisture divided by that of the accumulated forcing over the considered time frame as a main driver of soil moisture memory that reflects the impact of the precipitation regime and of soil and vegetation characteristics. Another important driver is found to be the correlation of initial soil moisture with subsequent forcing that captures forcing memory as it propagates to the soil and also land‐atmosphere interactions. Thereby, the role of precipitation is found to be dominant for the forcing. In contrast to results from previous modeling studies, the runoff and evapotranspiration sensitivities to soil moisture are found to have only a minor influence on soil moisture persistence at the analyzed sites. For the central European sites, the seasonal cycles of soil moisture memory display a maximum in late summer and a minimum in spring. An opposite seasonal cycle is found at the analyzed site in Italy. High soil moisture memory is shown to last up to 40 days in some seasons at most sites. Extremely dry or wet states of the soil tend to increase soil moisture memory, suggesting enhanced prediction potential during extreme events at most sites considered in this study.