CollectHomepage AdvertisementContact usMessage

Arid Land Geography ›› 2025, Vol. 48 ›› Issue (2): 179-189.doi: 10.12118/j.issn.1000-6060.2024.350

• Climatology and Hydrology • Previous Articles     Next Articles

Evaluation of hourly and daily precipitation forecasting performance of the CMA-MESO model in the warm season: A case of the Ili River Valley

MOU Huan1(), CHEN Chunyan1(), YANG Xia1, ZHAO Li2   

  1. 1. Xinjiang Uygur Autonomous Region Meteorological Observatory, Urumqi 830000, Xinjiang, China
    2. Xinjiang Information Engineering School, Urumqi 830013, Xinjiang, China
  • Received:2024-06-05 Revised:2024-07-15 Online:2025-02-25 Published:2025-02-25
  • Contact: CHEN Chunyan E-mail:mouh198209@foxmail.com;chency357@sina.com

Abstract:

The performance evaluation of quantitative precipitation forecasts can provide a scientific basis for the application and improvement of such forecasts. In this study, hourly site precipitation observation data and the CMA-MESO model’s quantitative precipitation forecast data from May to September (the warm season) of 2022—2023. Using evaluation indicators such as the probability of correct rainfall, threat score (TS), false alarm ratio, and missed alarm ratio, precipitation forecast performance over the Ili River Valley was analyzed. The results revealed the following: (1) The CMA-MESO model can reasonably depict the 1 h and 24 h precipitation characteristics in the Ili River Valley during the warm season. As the precipitation intensity increases, both the forecast and observed frequencies of precipitation show a downward trend. (2) The TS for the CMA-MESO model forecast of precipitation of different intensities is closely related to the forecast bias of the accumulated precipitation probability. The 24 h precipitation forecast TS score for the range of 6.1-12.0 mm is the lowest, with the highest cumulative probability forecast bias, exceeding a mean of 2.0%. The 1 h forecast TS score significantly decreases with the enhancement of precipitation intensity, reaching a peak bias of 1.7% at 0.1 mm. (3) The frequency of the forecasted and observed precipitation shows an increasing trend with altitude. However, the 24 h forecast frequency exhibits a negative bias across all altitudes, while the 1 h forecast frequency shows a positive bias in the low-altitude areas and a negative bias in the sub-high-altitude areas. (4) In terms of diurnal variation, the CMA-MESO model did not accurately simulate the characteristic of low precipitation frequency during the day and higher frequency during the night in the Ili River Valley. Specifically, the model tends to have more false alarms for daytime precipitation and more missed alarms for nighttime precipitation. A comparison of the frequency of precipitation observations with forecasts shows that the pattern of the forecast trend from early morning to afternoon is completely opposite to the observed frequency; the most significant forecast biases occurs between 13:00—14:00 and 02:00—05:00.

Key words: CMA-MESO, evaluation of precipitation forecast, the warm season, hourly and daily precipitation, the Ili River Valley