Preprints
https://doi.org/10.5194/gmd-2016-94
https://doi.org/10.5194/gmd-2016-94
Submitted as: model evaluation paper
 | 
21 Jun 2016
Submitted as: model evaluation paper |  | 21 Jun 2016
Status: this preprint was under review for the journal GMD but the revision was not accepted.

Microphysics parameterization sensitivity of the WRF Model version 3.1.7 to extreme precipitation: evaluation of the 1997 New Year’s flood of California

Elcin Tan

Abstract. Providing high accuracy in quantitative extreme precipitation forecasting (QEPF) is still a challenge. California is vulnerable to extreme precipitation, which occurs due to atmospheric rivers and might be more intense with climate change. Accordingly, this study is an attempt to evaluate the extreme precipitation forecasting performance of a QPF model, the Weather Research and Forecast Model, version 3.1.7, for the extreme precipitation event that caused the 1997 New Year’s flood in California. Sensitivities of 19 microphysics schemes are tested by utilizing 18 various Goodness of Fit (GoF) tests for hourly and point-wise comparisons between 3-km horizontal domain resolution simulations of the WRF Model and observations. The results indicate that the coefficient of persistence (cp) is the first metric that needs to be evaluated because it determines whether simulation versus observation values are reasonable. Comparisons of 3 out of 8 stations in the American River Watershed passed this test. The results also show that Normalized Root Mean Square Errors (NRMSE) and Percent Bias (PBIAS) metrics are more representative than others due to their ability to discriminate model performances. Further, microphysics (MP) schemes are also significantly sensitive to location. Although 3 of the stations that passed the cp test are quite near to each other spatially, different MP schemes become prominent for different observation locations. For instance, for the ALP station, MP3, MP8, MP17, and MP28 indicate better performances, whereas the errors of MP3, MP8, MP9, and MP17 are less than other MPs for the BTA station. However, MP11 has the only reasonable results, according to cp values for the CAP station. The MPs are also evaluated for 72-hr and basin-averaged precipitation estimations of the WRF Model by means of true percent relative errors. The results show that the accuracy of the WRF Model is much higher for the 72-hr total basin-averaged evaluations than for the hourly and point-wise comparisons. Thus, the Thompson Scheme (MP8) indicates more trustworthy results than others, with a 3.1 % true percent relative error. Although WRF simulations overestimate the 72-hr basin-averaged precipitation for most of the MP schemes, this may not be pronounced for moderate, heavy, and extreme precipitation when hourly and point-wise evaluations are performed but is valid for light precipitation.

Elcin Tan
 
Status: closed
Status: closed
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
Printer-friendly Version - Printer-friendly version Supplement - Supplement
 
Status: closed
Status: closed
AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment
Printer-friendly Version - Printer-friendly version Supplement - Supplement
Elcin Tan
Elcin Tan

Viewed

Total article views: 2,000 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
1,019 876 105 2,000 86 101
  • HTML: 1,019
  • PDF: 876
  • XML: 105
  • Total: 2,000
  • BibTeX: 86
  • EndNote: 101
Views and downloads (calculated since 21 Jun 2016)
Cumulative views and downloads (calculated since 21 Jun 2016)

Cited

Latest update: 18 Mar 2024
Download
Short summary
California is vulnerable to extreme precipitation, which occurs due to atmospheric rivers. This study is an attempt to evaluate the performance of the WRF Model for the extreme precipitation event that caused the 1997 New Year’s flood in California. The results show that the accuracy of the WRF Model is much higher for the 72-hr total basin-averaged evaluations than for the hourly and point-wise comparisons. The Thompson Scheme indicates more trustworthy results than others, with a 3.1 % error.