Welcome to SPACE

SPAtial Calibration and Evaluation in distributed hydrological modeling using satellite remote sensing data

The aim of this project is to develop a theoretical framework and new methodologies for spatial hydrological model evaluation that are required to enable a paradigm shift towards truly science based water management. The project output will contribute to the transition from an engineering practice based modelling towards a science based approach that balances the conceptual complexity and spatially distributed nature of current modelling schemes with the available data and the requirements for modern, spatially differentiated water management.

Currently, hydrological models remain focused on comparing simulations to a single spatially aggregated catchment scale observation in the form of river discharge, with the conviction that it provides some inherent insight into the internal hydrological behaviour of the river basin. This notion is outdated and limits the use of models for science based and differentiated water management. Therefore, a paradigm shift it required, moving away from the aggregated evaluation of hydrological models towards a spatially distributed approach. Recent advances in fully distributed and grid based model codes, computational power and spatial data availability have prepared the ground for bringing the science forward. However, hydrological model evaluation and calibration severely lack methodologies for incorporating spatial pattern information.

Therefore, the project will combine three lines of research:

1. Development of a new set of performance metrics that are specially designed for comparison of spatial patterns, e.g. by use of multiple-point geostatistics and pattern matching techniques
2. Generation of satellite remote sensing based datasets of hydrological states and variables such as soil moisture and evapotranspiration
3. Development of a new hydrological model evaluation and calibration framework based on the new spatial performance metrics and the satellite based observations. This framework must allow the simulated spatial pattern to adjust to an observed spatial pattern while considering the physical realism of the optimized parameter distribution. This is achieved through parameter regionalization schemes and combinations of joint inversion and super parameter optimization, where the á priori spatial parameter distribution is adjusted to match the spatial observations.