Unveiling environmental dimensions of hydrological drought in Southern Spain using open-source data and machine learning techniques

Drought is an increasing hydroclimatic threat in the Mediterranean, profoundly impacting water resources and ecosystems. Andalusia (Spain) is highly vulnerable due to climatic variability and prolonged dry periods. Effective drought management requires methods to assess impacts on groundwater and surface water systems, which in turn threaten ecological and socio-economic resilience. While socio-economic impacts are more analysed, environmental effects are overlooked due to delayed onset or unclear links to drought. However, drought-induced degradation of natural resources and hydrology-linked ecosystem services can exacerbate challenges in agroforestry, livestock, and tourism. Examining the environmental dimensions of hydrological drought risk is therefore essential.

This research takes a first step in analysing the impacts of drought on water-related ecosystem services. It specifically investigates hydrological and hydrogeological anomalies and examines their spatial and temporal dynamics across varying levels of drought severity. This study defines hydrological anomalies by leveraging high-resolution, open-access data from Copernicus and other datasets available on Google Earth Engine. These include estimates of soil moisture, groundwater storage, terrestrial water storage, flows and evapotranspiration that can be obtained from GLDAS 2.2, FLDAS, CERRA-Land, etc. In situ measurements, such as piezometric and streamflow records, are also integrated to validate findings and provide a robust basis for analysis of the impacts on water systems. Machine learning algorithms are then used to model the complex linkages between the identified hydrological anomalies and the climatic conditions, measured with well-known drought indices like the Standardized Precipitation-Evapotranspiration Index (SPEI) at different scales.

A pilot study in an Andalusian sub-basin with minimal anthropogenic influence serves as a testbed for developing a scalable methodology to evaluate the impacts of short and long-term drought conditions on groundwater and surface water. In line with related relevant research, correlation analyses run for this pilot highlight strong associations between hydrological variables and drought indices. A rapid response of surface water systems to short-term droughts is observed, while groundwater displays delayed, yet significant changes linked to drought, reflecting its buffering capacity and resilience.

This research highlights the potential of tested datasets for assessing drought impacts on water systems and demonstrates the value of open-source hydrological data for improving drought risk assessment and predictive tools. However, the study also reveals limitations regarding spatial resolution, which constrain detailed-scale assessments. On the one hand, the follow-up research will expand the performed analysis to additional sub-basins across Andalusia to compare results. On the other hand, similar modelling methodologies will be applied to understand how the identified droughts and associated anomalies in surface and groundwater systems propagate, leading to a reduction in the provision of ecosystem services. This will include exploring ecological impacts such as failures to maintain ecological flows, declines in extension of wetlands, or anomalies in primary productivity and ecosystem functioning in natural areas.