Improving the spatio-temporal accuracy of mesoscale hydrological modelling through multi-sensor remote sensing data fusion and assimilation

点击翻译按钮获取中文摘要

The anticipated future impacts of climate and land use change on the terrestrial water cycle demand modelling approaches and frameworks that are capable to simulate hydrologic processes at the appropriate scales for local and regional water resource management. At the mesoscale, physically-based hydrological models should reflect all relevant local conditions that control the spatially differentiated rainfall-runoff transformation. To establish a model that realistically mimics the behaviour of the hydrologic system, we will develop an ‘integrated’ multi-objective calibration framework that exploits, in addition to streamflow data, spatially detailed information on dominant runoff generation processes and daily rates of actual evapotranspiration (ETa), the latter through the integration of time series of high-resolution remote sensing data. For three selected catchments in the Nahe basin (SW-Germany), we will develop our approach within the highly customizable framework of the WaSiM-ETH model to construct an exemplary behavioural model that adequately represents the spatio-temporal dynamics of the rainfall-runoff transformation at hourly time steps. To this end, remotely-sensed land surface parameters and ETa maps derived from thermal imagery, both with a pixel size of ~30 – 100 m, will be incorporated. To obtain these maps with up to daily temporal resolution, we will implement multi-sensor data fusion and land surface temperature (LST) downscaling approaches that combine the fine spatial resolution of narrow-swath sensors like Landsat-8 and Sentinel-2 with high temporal resolution Earth observation systems such as MODIS and VIIRS. ETa estimation will then be carried out with a suite of well-established Surface Energy Balance models, including SEBAL, METRIC and SSEB. This processing chain will allow the integration of LST and ETa products at field scale, which is essential for a coupling with a spatially distributed modelling approach at the mesoscale. The derived remote sensing products can be used directly in the calibration process of WaSiM-ETH. We will establish manual and automatic calibration schemes, both based on a set of spatial metrics that capture the spatio-temporal dynamics of ETa patterns within the studied catchments. The combination of clear-sky ETa retrievals from remote sensing with continuous hydrological model simulations enables us to generate gap-free high-resolution ETa composite products for the studied catchments, which then will be upscaled to the entire Nahe basin. In addition, the properly calibrated behavioural model is considered to reflect changes in boundary conditions adequately and will thus allow us to gain insight into the response of the hydrological cycle under anticipated future global environmental change scenarios (e.g. for the RCP6.0 and RCP8.5 scenarios, which represent a strong change in temperature and precipitation conditions).