Date of Award

Spring 5-1-2005

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Geography, Geology, and Anthropology

First Advisor

Qihao Weng

Second Advisor

Gregory D. Bierly

Third Advisor

Brian Ceh

Abstract

Mapping urban temperatures are one of the major foci in urban climatology studies due to its close relationships with environmental, economic, and social issues. Most previous urban temperature studies were based on a single scale. This study utilized remote sensing and GIS data to perform a multiple scale analysis of urban temperatures. Land surface temperatures (LST) data derived from a Landsat Enhanced Thematic Mapper Plus (ETM+) image were used to examine census-based variations and to model their numerical relationships with urban morphology in Marion County, Indiana. Selected variables for the urban morphology include Normalized Difference Vegetation Index (NOVI), buildings, roads, and water. Analyses of Pearson correlation and stepwise regression at each census level (i.e., block, block group, and tract) were performed. Two different modeling schemes (the complete model with non-sampled data and two sub models with sampled data) were developed and compared. The sensitivity of the relationship to aggregation and thus the scale effect of Modifiable Areal Unit Problem were examined. Results show that, when four urban morphology variables were considered, LST had the strongest positive correlation with buildings, but was negatively correlated with water at three census levels. In contrast, when water was removed from the modeling, LST correlated the most with NOVI, but the least with roads. All LST models accounted for the most of the variance in LST, with the best performance at the census tract level and the poorest performance at the census block level. Generally speaking, the larger the analytical unit or the higher aggregation level, the stronger the correlation between LST and the urban morphology variables. It is suggested that census tract level was the most appropriate scale of the three census levels to model LST as far as the current study is concerned. Overall, the complete LST model is better than any of the two LST sub models. Among the four urban morphology variables, water was found to be a least significant factor in modeling LST, while vegetation, buildings, and roads were all of higher significance. It is concluded that LST studies should be conducted at multiple scales to systematically examine how the underlying processes change over space.

Download

Share

COinS