氣象預測模式之測試分析

Applicable Measurement and Analysis of Climatic Data Models

陳琦玲(Chi-Ling Chen) ； 漆匡時(Kung-Shy Chi) ； 林正錺(Chen-Fang Lin)

《中華農業研究》 40卷3期 (1991/09) Pp. 351-363

https://doi.org/10.29951/JARC.199109.0012

摘要

日氣象資料爲許多應用與評估工作不可缺少的資訊。WGEN與WMAKER皆是由美國農業研究人員發展出來之日降雨量、日射量、最高溫與最低溫等四項氣象資料之預測模式。本報告以臺中縣霧峰鄉臺灣省農業試驗所1982-1990年九年與前五年資料爲例，進行模式木土化模擬測試與預測能力評估。以統計t-test進行模式模擬值與實測值月平均及年平均之比較，結果顯示WGEN對前述四項氣象變數之模擬相當具有代表性，預測能力亦優於WMAKER。WMAKER尤其對日射量的模擬準確度甚差。WGEN對日降雨量、日射量及溫度等氣象變數之模擬與預測的準確性，以溫度最佳，日射量居中，降雨量則因實際天氣的變異較大，模式較難掌握，但若以月或年平均降雨量爲測試單位，則模擬值仍具代表性。測試結果亦顯示以5年的資料爲統計依據，模式雖可模擬日射量與溫度之主要季節變化，但要確實的掌握每日氣象變數的變異，以提高預測的準確性，則需要更長的實測氣象資料爲統訓基礎。總言之，以九年的氣象資料爲統計依據，引用WGEN作爲氣象變數之預測模式，以應用於霧峰地區作物生產潛量之評估應是可行的。

關鍵字

WGEN ； WMAKER ； 氣象模式 ； 氣象預測

並列摘要

Daily weather data are required for many applications and assessments. Two computer simulation models, namely, WGEN and WMAKER, were used to generate the daily precipitation, solar radiation, maximum temperature, and minimum temperature. This study tests the simulation accuracy of the models, based on the daily weather data collected at Taiwan Agricultural Research Institute, Wu-feng, Taiwan from 1982 to 1990. This study also tests their predictability, based on the weather data from 1982 to 1987, on predicting the weather data of the next four years. The statistical t-tests of mean monthly and annual weather variables revealed that applying WGEN on simulating and predicting the daily weather data at Wu-feng, Taiwan are highly acceptable, although WMAKER does not promise well for the test especially in the simulation of solar radiation. In conclusion, it is suitable to apply WGEN as weather generator, as was proved by analyzing the nine years historical weather data at Taiwan Agricultural Research Institute, Wu-feng.

並列關鍵字

WGEN ； WMAKER ； weather model ； weather prediction