年流量爲水資源分析與規劃常被引用之重要參考指標。流量特性之變化深受流域特性之影響，有相似流域特性反應之流域，常具有類似之流量特性。未量測流域之流量訊息，可以區域分析法估計獲得，而區域分析首需藉助於均質性區域之劃分，流域特性乃爲劃分流域均質區最具代表性之指標之一。因此，本文以台灣東部區域設有水文站之28個流域爲樣本，選取年雨量及17項流域特性指標爲說明變數，以多變量統計分析方法，分析其流域特性之基本結構與特徵，據以劃分流域均質區，探究其區域特性與空間差異，以建立與探討各流域均質區年流量與流域特性之相關與特性，最後，建立各流域均質區之判別函數，以做爲未量測流域之歸屬判別準則，俾利其水資源分析與規劃之進行。本文研究結果顯示，18項流域特性指標，可簡化綜合成6個重要特徵，其總解釋量達87.21%；根據流域特性之特徵，28個流域樣本可劃分成3個流域均質區，每一流域均質區皆有特殊之流域特性特徵；各流域均質區之年流量與流域特性之多元迴歸與相關；僅第一區通過顯著差異性檢定，其預測能力極高而且有效；所建立之3個流域均質區之判別函數，經檢定結果皆具顯著差異性，且其總判中率達96.43%，爲本區未量測流域歸屬之優良判別準則，可供水文地理學區域分析與研究之參考。
The annual discharge is an important index with regard to water resource analysis and planning. The characteristics of drainage basins affect the variation of the characteristics of discharge. There is a resemblance of the characteristics of discharge among the similar characteristics of drainage basins. The information of discharge in ungauged area can be estimated by the method of regional analysis based on the identification of homogeneous regions. One of significant indicators to identify homogeneous region is the characteristic of drainage basins.
The purposes of this paper therefore are both to establish the discriminant functions of drainage basins to classify the ungauged drainage basins and to study the correlation between the annual discharge and the characteristics of drainage basins of hydro logic homogeneous regions (HHR) from the annual discharge and 18 characteristic indicators of 28 samp1ing gauged watersheds by the multivariate analysis for the identification of HHR in the watersheds of eastern Taiwan.
Based on the resu1ts of analysis, it is possible to obtain 6 significant factors from 18 indicators, which could account for over 87% of the common variance, and to find 3 HHRs with significant characteristics of drainage basins in 28 sampling watersheds. The discriminant functions are adequate to completely separate the three HHRs, and the results are statistical significance the multiple regression function between the annual discharge and the characteristics of drainage basins in the first HHR is also statistical significance, and its ability of prediction is valid.
