How Well Do Operational Numerical Weather Prediction Configurations Represent Hydrology?

Ervin Zsoter,Joaquin Muñoz-Sabater,Elisabeth Stephens,Hannah Cloke,Patricia De Rosnay,Florian Pappenberger,Christel Prudhomme

doi:10.1175/jhm-d-18-0086.1

Abstract

Abstract Land surface models (LSMs) have traditionally been designed to focus on providing lower-boundary conditions to the atmosphere with less focus on hydrological processes. State-of-the-art application of LSMs includes a land data assimilation system (LDAS), which incorporates available land surface observations to provide an improved realism of surface conditions. While improved representations of the surface variables (such as soil moisture and snow depth) make LDAS an essential component of any numerical weather prediction (NWP) system, the related increments remove or add water, potentially having a negative impact on the simulated hydrological cycle by opening the water budget. This paper focuses on evaluating how well global NWP configurations are able to support hydrological applications, in addition to the traditional weather forecasting. River discharge simulations from two climatological reanalyses are compared: one “online” set, which includes land–atmosphere coupling and LDAS with an open water budget, and an “offline” set with a closed water budget and no LDAS. It was found that while the online version of the model largely improves temperature and snow depth conditions, it causes poorer representation of peak river flow, particularly in snowmelt-dominated areas in the high latitudes. Without addressing such issues there will never be confidence in using LSMs for hydrological forecasting applications across the globe. This type of analysis should be used to diagnose where improvements need to be made; considering the whole Earth system in the data assimilation and coupling developments is critical for moving toward the goal of holistic Earth system approaches.

Highlights

This study focuses on the impact of the water budget closure on river discharge
The river discharge behavior provides a useful indication of the hydrological differences between the ONLINE and OFFLINE simulations
The impact on the hydrology could be demonstrated on two important surface variables: 2-m temperature and snow depth which are relatively well observed variables and can be used to analyze the impact of the land–atmosphere coupling and land data assimilation system (LDAS) on the surface globally in the two experiments

Summary

Introduction

Land surface models (LSMs) have traditionally been designed to focus on providing lower-boundary conditions to the atmosphere by describing the vertical fluxes. VOLUME 20 of energy and water between the land surface and the atmosphere, with less focus on predicting runoff (Mengelkamp et al 2001). LSMs maximize the quality of the atmospheric forecast, but do not necessarily bring the same benefits in the representation of the hydrological cycle (Kauffeldt et al 2015). There are significant limitations in the representation of hydrological fluxes and storages in LSMs, largely due to the large-scale focus of LSM applications, which has led to the neglect of some important processes for runoff generation (Overgaard et al 2006; Le Vine et al 2016), including inadequate snowmelt processes (Dutra et al 2012; Zaitchik and Rodell 2009)

Results

Discussion

Conclusion