Integrating isotopes and remote sensing into large-scale ecohydrological modelling of heavily managed water resource systems in drought sensitive areas

Whilst stable water isotopes have enhanced process understanding and tracer-aided ecohydrological modelling in contrasting landscapes around the World, most studies have been undertaken in relatively small catchments with limited anthropogenic management and disturbance. There are clear knowledge gaps in terms of how such approaches can be applied in larger, more complex landscapes with more intrusive management impacts from agriculture, industry and urban areas. Under more dominant influence of human management decisions, the ecohydrological coupling between land use, water storage and water fluxes become even more complicated. Understanding and quantifying these couplings requires improved, integrated modelling of the more natural and more managed components of catchment systems. As water stable isotope (ẟ¹⁸O and ẟ²H) are effective in identifying water sources, flow paths and transit times, and have been increasingly applied in tracer-aided hydrological modelling, we use them here to better understand the hydrology of the Spree catchment in Germany. This is a major strategic water resource which provides Berlin’s main drinking water supply, maintains significant agricultural irrigation and sustains local industrial needs. We focus on a 2800km² sub-catchment; the ET-dominated Spreewald region that has a heterogenous mixed land use (croplands, pastures, forests and urban) but is heavily influenced by water resource management interventions (regulated and unregulated abstractions, inter-basin transfers etc.). We used the spatially distributed tracer-aided model STARR to simulate the effects of natural water storage-flux dynamics and monitored management intervention on stream flow over a 6 year period. We found that conventional spatially-distributed streamflow-based calibration resulted in unrealistic isotope simulations, with large uncertainty in rainfall partitioning, overestimation of soil evaporation and ambiguity of runoff sources. Re-parameterization of the model provided a better constraint on isotope simulations across the model domain with no deterioration of streamflow estimates and a much stronger apportionment of runoff to groundwater and upstream sources. This was further validated against local flux tower data and satellite derived ET products (PML) for the region. However, some sub-catchments within the model domain under-predicted summer stream flows and these were consistent with areas where un-monitored irrigation in riparian croplands likely increases ET and reduces stream flow. The modelling framework used shows promising potential for wider use of isotopes in large scale tracer-aided modelling of complex, heavily managed catchments. Isotopes can help reduce equifinality in traditional water resource modelling and help identify the influence of unregulated human managements effects such as irrigation. However, given the inevitable logistical constraints in isotope sampling over extensive areas, integration with satellite products, such as ET or soil moisture estimates, can help leverage maximum value from available isotope-based insights in large-scale modelling.