適當氣候平均法（Optimal Climate Normals，簡稱OCN）是以一個用一段時間（K年）的平均值作為預測結果的簡易預報方法。隨著全球暖化現象愈趨明顯，美國氣象局在1994年將OCN列入預報作業模式，一直延用至今。本文將介紹OCN方法對台灣月平均氣溫的預測能力。研究發現，不論在那一個季節，OCN對台灣東部測站溫度的預報技術都不高。對台北、台中、高雄等西部測站而言，OCN以11月的預報技術最好，K=7是適當的選擇。OCN的預報技術主要來自於溫度有明顯的線性上升趨勢，而且沒有週期5-7年的振盪現象。以10月來說，雖然平均溫度和11月差別不大，但因5-7年的振盪現象比較明顯，以致於OCN預報技術不高，和11月的差距甚大。本研究可作為挑選臺灣氣候預測模式的參考，亦即如果一個模式預測11月氣溫的技術得分不能高於OCN模式，就表示該模式不具有預測臺灣溫度的能力。

The Optimal Climate Normals (OCN) is a simple statistical forecast model that requires data only from the station where the forecast is needed. The OCN has been used as a formal method for US seasonal forecasts at the Climate Prediction Center/NOAA since late 1994. Until the present, this method is still used as a bench mark for forecast skill measurement mainly due to strong warming trend after late 1970s. In this study, the forecast skill of OCN used for predicting Taiwan temperatures is analyzed. Four stations: Taipei, Taichung, Kaoshiung and Hualien are used to represent temperature characteristics in north, central, south and east Taiwan, respectively. Our results suggest that the forecast skill of the OCN method is generally poor over east Taiwan. However, over other regions, OCN shows robust high forecast skill in November, when the number of years (K) used in making the climate normal is 7. The November high skill can be explained by strong warming trend in this particular month. The trend is less clear in other months. An interesting contrast appears in October, of which the temperature shows clear 5-7 years variations that results in lower forecast skill. It is suggested that the OCN for November temperature over western Taiwan can be used as a baseline for evaluating the monthly or seasonal climate forecast models.