Vegetation Classification and Mapping of the Da-an River Upstream Basin
| Autor: | Kun-Yi Li, 李坤益 |
|---|---|
| Rok vydání: | 2010 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 98 This study were according the literatures sampling data of Da-an river basin to arrangement the vegetation classification schedule of Da-an river basin, and then combined with the vegetation classification system that derived by the National Vegetation Mapping Project. We integrated an useful vegetation classification system which applied to vegetation mapping studies. The results of vegetation classification integrating were subdivided the vegetation into 12 natural vegetation types (including 9 formation types), 1 plantation forest type and 3 other types and established the identification key of Da-an river basin for field survey and mapping strategy. Vegetation mapping was based on alliance level and integrated with results of vegetation classification schedule to vegetation mapping. The results of vegetation mapping comprised within native vegetation units, in which the submontane evergreen broad-leaved forest type was the most abundant native formation types where occupied 7,795.40 ha (19.63％ of total area), and the Litsea acuminate alliance type occupied 6,771.98 ha (17.05％), which was the most abundant native alliance types. Moreover, the artificial forest type occupied 8,885.32 ha (22.37％). Within the other types, the native exposed type which occupied about 1,187.82 ha (2.99％) was the most abundant. After field surveying and data screening, we sifted out 1,017 observations and environmental data to applyed in prediction model in GAMs. The accuracy of prediction models were assessed by the area under the curve (AUC) of ROC and Kappa statistics, which fallows sifted the distribution of vegetation types of prediction models by maximum Kappa value (MaxKappa), and the results were compared with the existing vegetation map to estimate predictive accuracy. The results show the best accuracy in submontane evergreen broad-leaved forest type of formation types and Machilus thunbergii type in alliance types, but the other vegetation units were not achieved reasonable accuracy. The main reason was cause of the predictive error, which was subjected to different degrees of disturbance and inter-species competition. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|