Investigation of the sensitivity of WRF radiation physics options to hourly and daily temperature prediction accuracy

Date

2024-11-28

Author

Çatal, Ali Cem

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

174
views

152
downloads

Cite This

The physical configuration options in the Weather Research and Forecasting (WRF) model affect the simulation’s outputs. Therefore, improving the accuracy of hourly and daily temperature predictions requires a deep understanding of the sensitivity of WRF radiation physics options. This thesis examines the impact of various radiation physics schemes within WRF on temperature forecasts at different temporal resolutions. By analyzing 40 different LW and SW physics options, the study aims to uncover nuances in the model’s temperature prediction performance. Utilizing ERA5 reanalysis data and station-based observational datasets for the year 2020 within the Türkiye domain, comprehensive sensitivity analyses are conducted at a 9 km spatial resolution. Also, the thesis examines the implications of these findings to highlight the potential for enhancing the reliability of temperature predictions. It is observed that the radiation physics could affect the hourly temperature correlations up to 0.22 and affect the bias up to 15 ºC. Also, using even the basic machine learning models on the best five models, the hourly bias can be decreased to 0.0002ºC, and correlations can be improved from 0.965 to 0.967, which may be further improved by using more complex models or applying hyperparameter tuning.

Subject Keywords

WRF, Sensitivity analysis, Radiation physics, Temperature predictions, Machine learning

URI

https://hdl.handle.net/11511/112912

Collections

Graduate School of Natural and Applied Sciences, Thesis