753 Evaluation of Solar Radiation Forecasts of the Operational WRF-RTFDDA NWP System at CEPRI

Tuesday, 9 January 2018
Exhibit Hall 3 (ACC) (Austin, Texas)
Si Shen, NCAR, Boulder, CO; and Y. Liu, W. Y. Y. Cheng, Y. Liu, Y. Huang, P. Jimenez, W. Wang, C. Liu, S. Feng, and S. Jin

A Weather Research and Forecasting – Real-Time Four-Dimensional Data assimilation (WRF-RTFDDA) numerical weather prediction (NWP) system was developed at the Chinese Electric Power Research Institute to provide diverse weather services for the State Grid Corporation of China (SGCC) who operates over 75% electric power grids in China. Since May 2017, the system has been running four cycles a day, starting at 00, 06, 12 and 18 UTC, with 72h forecasts for each cycle. The model runs at grid intervals of 3km, covering all SGCC’s power grid operation regions in China. One of the major goals of the system is to produce wind and solar radiation forecasts to support grid integration of the renewable energy. In this study, we evaluate the surface solar radiation forecasts of the modeling system based on ~140 surface stations distributed over the China mainland. The verification period is from March 1, 2017 to December 31, 2017. The 3-km grid NWP system permits weather forecasting with explicit cloud physics representation. Among customization of several major model components, WRF-solar capabilities are configured for this application. Conventional statistical metrics including bias error, mean absolute error, root mean squared error, and correlation coefficients are computed for each station and for domain/sub-region averages. The dependencies of the forecast errors on geographical regions/locations, seasonal changes, and prevailing weather regimes are characterized. It is found that the effects of the cloud modulation to the solar radiation transfer are dramatically different between the higher latitude clouds and the lower latitude clouds, the winter clouds and the summer clouds, and the cloud over the lower terrain and the clouds over the high plateau. These metrics provide valuable guidance for correcting the forecast error through statistical post-processing and refining the model physics (in particular cloud and radiation) parameterization.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner