Representing vegetation processes in hydrometeorological simulations using the WRF model

Abstract

For accurate predictions of weather and climate, it is important that the land surface and its processes are well represented. In a mesoscale model the land surface processes are calculated in a land surface model (LSM). These pro-cesses include exchanges of energy, water and momentum between the land surface components, such as vegetation and soil, and their interactions with the atmosphere. The land surface processes are complex and vary in time and space. Signi_cant e_ort by the land surface community has therefore been invested in improving the LSMs over the recent decades. However, improve-ments are still needed in the representation of the land surface variability and of some key land surface processes. This thesis explores two possibilities for improving the near-surface model predictions using the mesoscale Weather Research and Forecasting (WRF) model. In the _rst approach, data from satellite images were used to investi-gate the impact of accurate representation of vegetation fractions in the Noah LSM, which is the default LSM in WRF. In the second approach, advanced land surface parameterizations included in a new version of the Noah LSM with multiparameterization options (Noah-MP) were investigated. A novel method was used to derive high-resolution, high-quality vegetation information from Moderate Resolution Imaging Spectroradiometer (MODIS) vegetation indices. The indices were converted into a vegetation fraction, which is a key parameter for the representation of vegetation in WRF, us-ing both a linear and a quadratic approach. The quadratic approach was identi_ed as superior over the linear approach for use in the WRF model. In addition, it was noted that vegetation seasonality during 2006 deviated sig-ni_cantly from its climatology over some regions of Europe, possibly due to the occurrence of heat wave and drought conditions. The results of the sim-ulations over these regions obtained using the WRF model, at relatively low spatial resolution and using MODIS vegetation data sets, showed improved temperature predictions when using the quadratic approach. The quadratic MODIS vegetation data and the default vegetation data in WRF were further used in high-resolution simulations over Denmark down to cloud-resolving scale (3 km). Results from two spatial resolutions were compared to investigate the inuence of parametrized and resolved convec-tion. The simulations using the parametrized convection showed large overes-timations of precipitation during the summer, while the ones using resolved convection more accurately followed the observations. In general, equally good performance for precipitation, temperature and wind speed was found using both vegetation data sets. However, the MODIS data improved the simulation of the temporal evolution of the Bowen ratio during the summer. The work on the Noah-MP LSM included coupling, debugging and eval-uation of its performance in the WRF modeling framework. In addition, new options were added, such that several thousand combinations of parameterization settings are now available. In this thesis, the model is presented together with an evaluation of its performance during a summer simulation using the current recommended options. The simulations show that good results are obtained for temperature and precipitation.