Parameterisation and scaling of the land surface model for use in a coupled climate-hydrological model

Abstract

Summary Climate impact studies in hydrology have traditionally neglected the land–atmosphere feedback. Hydrological models are forced with output from climate models but neglecting this feedback may lead to an inaccurate estimation of evapotranspiration (ET). Two-way coupling of a hydrological model and a climate model can overcome this problem by linking the two models through a shared land–atmosphere process description. In this study we analyse the hydrological model MIKE SHE using a two-layer energy-based ET model for use in a coupling with a regional climate model (RCM). The value of coupling to MIKE SHE is that it makes it possible to include lateral transport of surface water and groundwater not generally treated in RCM’s and to represent human interventions like groundwater pumping, irrigation schemes, etc. for adaptation studies. The hydrological model is applied to the FIFE site to investigate the effects of model resolution and parameter scales. The area of interest corresponds to a RCM grid cell. The hydrological model is parameterized with effective parameters assessed directly from field data at the site and literature. Using only these observed data and literature estimates to parameterise the model, it is able to reproduce the observed ET, sensible heat flux and to some extent surface soil moisture content; over a whole growing season. Hydrological simulations carried out over a range of spatial grids from 240 m to 15 km show that, for this case, the areal average ET appears to be insensitive to model resolution. The model is able to reproduce some of the spatial variability within the area, but not the exact pattern. By running the hydrological model at the highest resolution with uniform atmospheric input we examined the effect of using coarser resolution climate forcing, for example from a RCM. The areal mean ET and soil moisture (SM) temporal variations are reproduced quite well, but the spatial variability in the hydrological response is substantially underestimated; mainly because of the uniform precipitation. Our results are therefore encouraging for using this type of energy-based model in a coupling between a regional climate model and a distributed hydrological model. As the FIFE area is a relatively homogeneous site, additional tests are needed at heterogeneous sites to validate whether our findings are in general valid.