THESIS
2018
xii, 99 pages : illustrations (some color), maps (some color) ; 30 cm
Abstract
We refine the roughness length value in the Weather Research and Forecasting Model
(WRF). The land surface scheme in WRF assigns a single roughness length value to the
grid cells, in particular, the urban grid cells. Such implementation is insufficient to represent
the heterogeneous morphological properties in the Pearl River Delta (PRD) region.
As such, wind speed predicted by WRF is always overestimated, especially in the urban
cores. Although urban canopy models can be coupled with WRF (known as uWRF) to
remedy the wind speed prediction problem, both the exceptional amount of skyscrapers
in PRD region and a more expensive computational cost put uWRF to be infeasible for the
operational run.
Given the restrictions, we update the roughness length value as a compromised method...[
Read more ]
We refine the roughness length value in the Weather Research and Forecasting Model
(WRF). The land surface scheme in WRF assigns a single roughness length value to the
grid cells, in particular, the urban grid cells. Such implementation is insufficient to represent
the heterogeneous morphological properties in the Pearl River Delta (PRD) region.
As such, wind speed predicted by WRF is always overestimated, especially in the urban
cores. Although urban canopy models can be coupled with WRF (known as uWRF) to
remedy the wind speed prediction problem, both the exceptional amount of skyscrapers
in PRD region and a more expensive computational cost put uWRF to be infeasible for the
operational run.
Given the restrictions, we update the roughness length value as a compromised method
to improve wind speed prediction and maintain an efficient model simulation. By utilizing
the Local Climate Zones (LCZs) classified by the World Urban Database and Access
Portal Tools (WUDAPT) and its associated metadata, we calculate a new set of roughness length value and replace the model-provided data. Through the integration of the dataset
to the land surface model in WRF, there is a general improvement on the wind prediction.
Using January, May, and July as test cases, we found there is 10% improvement in terms
of root mean square error, and 20 – 25% improvement in terms of mean bias. We also
discovered the wind speed improvement is more significant in urban areas.
Post a Comment