Diagnosing hydrological limitations of a land surface model: application of JULES to a deep-groundwater chalk basin

N Le Vine,N Mcintyre,C Jackson,A Butler

doi:10.5194/hess-20-143-2016

Abstract

Abstract. Land surface models (LSMs) are prospective starting points to develop a global hyper-resolution model of the terrestrial water, energy, and biogeochemical cycles. However, there are some fundamental limitations of LSMs related to how meaningfully hydrological fluxes and stores are represented. A diagnostic approach to model evaluation and improvement is taken here that exploits hydrological expert knowledge to detect LSM inadequacies through consideration of the major behavioural functions of a hydrological system: overall water balance, vertical water redistribution in the unsaturated zone, temporal water redistribution, and spatial water redistribution over the catchment's groundwater and surface-water systems. Three types of information are utilized to improve the model's hydrology: (a) observations, (b) information about expected response from regionalized data, and (c) information from an independent physics-based model. The study considers the JULES (Joint UK Land Environmental Simulator) LSM applied to a deep-groundwater chalk catchment in the UK. The diagnosed hydrological limitations and the proposed ways to address them are indicative of the challenges faced while transitioning to a global high resolution model of the water cycle.

Highlights

Guidance to support adaptation to the changing water cycle is urgently required, yet the ability of water cycle models to represent the hydrological impacts of climate change is limited in several important respects
Observations of water fluxes, soil moisture and groundwater levels in the Kennet catchment are compared with the simulated values derived using the sequentially modified JULES model structure to represent the four hydrological functions of a catchment
Neither configuration is capable of producing any surface runoff. This is because the hydraulic conductivities of the catchment soils, even for relatively clayey soils, are sufficiently high to enable virtually all the instantaneous rainfall rates obtained using temporal disaggregation to infiltrate into the soil

Summary

Introduction

Guidance to support adaptation to the changing water cycle is urgently required, yet the ability of water cycle models to represent the hydrological impacts of climate change is limited in several important respects. Hydrology (as well as other soil–vegetation–atmosphere interactions) in climate models is represented via land surface models (LSMs) that partition water between evapotranspiration, surface runoff, drainage, and soil moisture storage. The deficiencies in hydrological processes representation lead to incorrect energy and water partitioning at the land surface (Oleson et al, 2008) that propagates into precipitation and near-surface air temperature biases in climate model predictions (Lawrence and Chase, 2008). Improving the representation of hydrology is a step towards the development of a global hyper-resolution model for monitoring the terrestrial water, energy, and biogeochemical cycles that is considered as one of the grand challenges to the community (Wood et al, 2011). The most recent third generation LSMs operate in a continuous time and distributed space mode, and simulate exchanges of energy, water, and carbon between the land surface and the atmosphere using physics-based process descriptions (Pitman, 2003). The physics-based nature of third generation LSMs allows widely available global data sets, such as soil properties, land use, weather states, etc., to be used as model parameters and inputs, making predictive modelling with LSMs very appealing

Methods

Results

Conclusion