Do we need better models or more local knowledge? Assessing the added value of using locally sourced data over larger-scale datasets in regional to local hydrological modelling

Future water security will be determined by climate change along with socio-economic changes, driving water availability, water demands and catchment conditions. Over recent decades, hydrological models have evolved to incorporate the effect of anthropic activities that allow them to explore the main challenges and opportunities regarding global water security. These advances have been underpinned by progress in high-resolution and large-scale data availability, as well as in computational and data storage capabilities. Hydrologists are currently capable of developing high-resolution large-scale hydrological models designed to represent and study the global hydrological cycle, and even to zoom-in on specific regions essentially removing the barriers between global and regional models. However, the use of these modelling approaches is often seen with suspicion by end-users, be it regional water managers or water users, who may consider that their personal knowledge and understanding of the catchments where they carry out their activities is disregarded in favour of novel technologies.

Indeed, despite their growing sophistication, the current generation of LHMs is not yet exempt of limitations in their ability to represent dynamic trade-offs in the water-food-energy-environment nexus, and water competition between upstream and downstream users in complex water resources systems. These limitations hinder the ability of LHMs to provide reliable insights at the regional or local levels, leaving the task of incorporating human water management activities within these models as one of the grand challenges for the hydrologic research community. The inclusion of this local knowledge into LHMs’ modelling process can, therefore, increase their capacity to support rigorous nexus analyses to inform water policy and management decisions. Unfortunately, the access to these data may be limited by several inconveniences such as overprotective water authorities, language access barriers, or simply not existing at all. This work explores to what extent the inclusion of local knowledge can improve the performance of globally formulated models as well as their reliability to support decision making on the ground. We discuss what type of data might be more relevant and what should be the priorities in data acquisition to maximise the output of modellers efforts.

To demonstrate this, we built a CWatM model of the Ebro River catchment in Spain using large-scale datasets to later substitute or enhance such datasets with data obtained from local and regional specific datasets available from local authorities and water users. The additional data/enhancements were included in separate and cumulative steps. The model improvements were assessed comparing the model results against gauged flows, reservoir storage, water demand and supply, and the system’s drought indicator. The findings of this study will assist in the transition of globally formulated models to being applied locally by identifying the priorities in data gathering and advances in modelling capabilities, ensuring that they provide reliable outputs to inform decision making.