How landscape patterns impact Land Surface Temperature
- 作者：Chen AilianChen LidingSun Ranhao
- 会议时间：2011-08-01
- 关键词：Landscape PatternLandscape Metrics ; Urban Heat Island ; Land Surface Temperature ; Connectivity
- 作者单位：State Key Lab of Urban Regional Ecology, Research Center for Eco-Environmental Sciences, CAS, Beijing 100085,China
- 母体文献：第五届海峡两岸遥感遥测会议论文集
- 会议名称：第五届海峡两岸遥感遥测会议
- 会议地点：哈尔滨
- 主办单位：中国地理学会
- 语种：chi
- 分类号：U21;TK1
摘要
Urban landscape patterns, as the spatial composition and construction of landscape patches, have been claimed to greatly impact the urban heat island (UHI) effects. As for a sustainable urban is concerned, it is better to arrange more vegetation or wetlands in the urbanized area, which then bring a tradeoff that extensive cities cost more in facilities. Is it possible to balance this tradeoff by maximizing the cooling effect of vegetation or wetlands by rearranging landscape, since landscape pattern could affect landscape processes? This paper aims to inspect how landscape patterns correlate to LST and thus serve to mitigate UHI effects by landscape design or planning. We took some part of Beijing as a case study, used fine resolution QUIKBIRD image (pixel resolution: 2.5m) to map the landscape types, and retrieved land surface temperature (LST, pixel resolution: 60m) from 2 Landsat 7抯 ETM+ images in almost the same season. Then landscape pattern, including percentage, fractional character and connectivity was calculated at both class and landscape level, and correlated to the LST map. For the calculation of Class metrics, we resampled LST maps to a pixel resolution of 120m, thus within tolerate geometric error, one LST pixel corresponds to 48 by 48 landscape pixels having a resolution of 2.5m.Using a 48 by 48 window, we clipped the landscape map into small quadrate pieces and counted class metrics of each pieces. The result shows that the mean temperature of landscape types differs from each other and the selected class metrics of forests, low and medium albedo buildings and lakes show significant correlations to LST and could be probably utilized to help further landscape planning. Class metrics of other landscape types however hardly show significant correlations to LST. Some metrics at landscape level show significant correlations with LST, but it is difficult to explain the implication.