都市地表型態對行人尺度風場與熱環境之探討:以臺南市區為例

Abstract

隨著都市發展、熱島效應日益增強，從少數官方測站取得的觀測資料並不足以能夠代表都市的微氣候。本研究利用臺南市區與鄰近17個氣象局測站的風速資料以及高密度地面氣溫測量網(HiSAN)的溫溼度資料來計算行人尺度(2公尺高)的氣象資訊與體感溫度。本研究基於都市的建成環境資料，在考量平均地表粗糙度，使用改良的空間內插方法推算至行人高度的風速。然而在都市中，不只植物與建物的比率以及高度會影響地表粗糙度，大氣穩定度亦然。故本研究先利用交叉驗證法(Leave One Out Cross Validation)評估各氣象因子與條件下最好的參數設定，再選定參數進行空間內插。而在弱綜觀與無降雨的情況下，發現氣象因子的時空分布反映出有些區域有較高的體感溫度。為了了解都市地表型態與微氣候關係，利用主成分分析(Principal Component Analysis)選出較適當的參數，其中建物覆蓋率(BCR)、綠地覆蓋率(GCR)、建物高度(BH)與天空可視率(SVF)，皆被選出來用於分群演算法。局部氣候分區與分群演算法被用來區分都市型態並結合氣象條件，進而理解各型態的環境特徵。兩種都市型態分類方法都可以區分出合適的種類。這些資訊可以加值應用於即時呈現網頁，WebGIS，並提供都市規劃與決策使用。

As urban heat island effect intensifies, weather data produced by a mainly official weather station are not proper to represent and reflect the microclimate situation in a city. This study selected 17 weather stations in Tainan, Taiwan, to estimate the wind velocity on pedestrian-level and utilized 102 automatic stations from high-density street-level air temperature observation network (HiSAN) to measure air temperature and relative humidity at 2.5 meters height. Based on these observed weather data and urban environmental information provided from the government. This study established a method of generating high-resolution pedestrian-level weather information for urban areas. The method took urban morphological parameters, such as surface roughness, into consideration to be the factor of evaluating wind velocity. By interpolation and extrapolation, each grid obtained microclimate weather data on a pedestrian-level scale. The Leave One Out Cross Validation (LOOCV) tested the root-mean-squared errors for the interpolated data. Thus, the best set of parameters, height, methods, nearest neighbor numbers, and atmospheric stability, have been selected for conducting the calculation. The spatial and temporal distribution reflects the patterns of the microclimate conditions under the weak synoptic and no precipitation conditions. Furthermore, the results reflected the hot spots in the city and provided for the urban planners to improve this problem. Several parameters were examined to choose by principal component analysis. The results indicated four parameters, building coverage ratio, green area coverage ratio, building heights, and sky view factor, which should be used to the cluster analysis. The Local Climate Zone (LCZ) and the cluster analysis were used to classify the urban morphology, combining the microclimatic conditions for comprehensively understanding the pattern in each class. Both approaches showed the relationship within the appropriate categories. In addition, all data were integrated into the apparent temperature and presented by a useful tool, WebGIS. The application could provide a simple way to visualize an instantly environmental situation for urban planning and decision making.