REVERSE THINKING THE NEGATIVE EFFECTS OF TOPOGRAPHIC SHELTERS ON TAIWAN RED CYPRESS DISTRIBUTION BY GEOSPATIAL INFORMATION TECHNOLOGY

摘要

Ecological niche modeling (ENM) coupled with 3S provides useful information on large spatial scales. Chamaecypahs formosensis (Taiwan red cypress, TRC) grows on Chilung Mountain and Paiku Mountain in fog-forest belt, but almost do not grow in most Huisun area, except Shou-Cheng Mt. Ultimately, we will attempt to find out northeastern season wind blocked by topographic shelters is a key factor by reverse casting from the case of Taiwan firs in Hohuan Mountains. We developed ENMs by generalized linear model (GLM), back propagation neural network (BPNN), and maximum likelihood (ML). The mean kappa of the models (0.96) developed by samples taken only from Shou-Cheng Mountain was significantly greater than that of models (0.54) samples taken from entire study area. Topographic variables currently used except elevation were almost useless. This result indicated that our ENMs could have left out critical predictor variables. We will incorporate surrogates of wind or humidity, topographic sheltering index, topographic sheltering index (wind or humidity), into ENMs to improve model accuracy.