應用AVHRR資料在HIRS/2視場雲量估算之研究

Applying AVHRR Data in Estimating the Cloud Cover of HIRS/2 FOV

劉振榮(Gin-Rang Liu) ； 陳萬金(Wann-Jin Chen)

《大氣科學》 21卷1期 (1993/03) Pp. 23-53

摘要

利用衛星觀測的輻射資料反演大氣垂直溫濕剖面是氣象衛星除了提供衛星雲圖以外，另一項主要的貢獻。而在溫濕剖面的反演過程中，反演單元視場內由於雲的存在所產生的問題是最難以解決的，而這也是造成衛星資料反演結果在和傳統探空比較上最主要的誤差來源之一。本研究的目的即是利用AVHRR高解析度的輻射觀測資料來估算HIRS/2視場內的雲參數，以供未來更準確反演雲區大氣垂直溫濕剖面之用。研究中使用Aoki（1980）所提出HIRS/2和AVHRR視場幾何位置對應的方法，並選用大量晴空和完全雲遮蓋的視場樣本，分別建立AVHRR第四頻道與HIRS/2第八頻道輻射強度的線性迴歸關係式，再配合HIRS/2的輻射強度觀測值以求得HIRS/2裂雲視場中的有效雲量。在研究結果的驗證上，本研究將估算得之HIRS/2視場的有效雲量值分為六類，並以相對應位置的AVHRR視場所偵測到的有效雲量為參考值，以誤差矩陣的形式來評估HIRS/2反演有效雲量分類的準確度。由十二個個案的資料顯示，不論冬季或夏季的個別個案以及綜合的個案，其誤差矩陣在主軸上顯示之準確率均在83%至89%之間，變化不大，代表本研究方法的結果極為穩定，足供未來衛星資料在裂雲區反演大氣垂直溫濕剖面時之應用。

關鍵字

衛星遙測 ； 雲量估算

並列摘要

With the exception of providing satellite cloud image, another remarkable achievement of weather satellite mission is the ability to inverse the atmospheric vertical profiles from the radiance observations. In the process of satellite sounding retrieval, it is found that retrieving accurate satellite sounding is more difficult if the HIRS/2 FOV is contaminated by cloud. In this situation, large difference will occur between satellite retrieval results and radiosonde data. For the purpose of solving the problem caused by cloud within the HIRS/2 FOV, the high resolution radiance observations, AVHRR data, are used to estimate the cloud parameters which will be applied to retrieve satellite sounding in the future. Aoki's position-matching method (1980) was applied to match HIRS/2 and AVHRR images, and lots of radiance observations of clear completely-cloudy FOV were selected to establish the linear relationship between the radiance of AVHRR channel 4 and that the HIRS/2 channel 8. From the relationship and the observed AVHRR radiance, accurate effective could amount within the HIRS/2 FOV can be estimated.In verification, the effective cloud amount within the FOV of HIRS/2 was classified into six classes and using the effective cloud amount of the AVHRR FOV located within the HIRS/2 FOV as reference to evaluate the accuracy of cloud amount estimation. The results were shown in a error-matrix. From the results of twelve cases, the accuracies showed on the principal diagonal of the error-matrix are about 83% to 89% for all of the cases. This small variation in accuracy represents the stability of this cloud cover estimation algorithm, and it is able to be applied in sounding retrieval in partly cloudy area in the future.

並列關鍵字

Satellite Remote Sensing ； Cloud Cover Estimation