Effects of Land Use/Cover Changes and Urban Forest Configuration on Urban Heat Islands in a Loess Hilly Region: Case Study Based on Yan'an City, China

Abstract

In this study Yan'an City, a typical hilly valley city, was considered as the study area in order to explain the relationships between the surface urban heat island (SUHI) and land use/land cover (LULC) types, the landscape pattern metrics of LULC types and land surface temperature (LST) and remote sensing indexes were retrieved from Landsat data during 1990-2015, and to find factors contributed to the green space cool island intensity (GSCI) through field measurements of 34 green spaces. The results showed that during 1990-2015, because of local anthropogenic activities, SUHI was mainly located in lower vegetation cover areas. There was a significant suburban-urban gradient in the average LST, as well as its heterogeneity and fluctuations. Six landscape metrics comprising the fractal dimension index, percentage of landscape, aggregation index, division index, Shannon's diversity index, and expansion intensity of the classified LST spatiotemporal changes were paralleled to LULC changes, especially for construction land, during the past 25 years. In the urban area, an index-based built-up index was the key positive factor for explaining LST increases, whereas the normalized difference vegetation index and modified normalized difference water index were crucial factors for explaining LST decreases during the study periods. In terms of the heat mitigation performance of green spaces, mixed forest was better than pure forest, and the urban forest configuration had positive effects on GSCI. The results of this study provide insights into the importance of species choice and the spatial design of green spaces for cooling the environment.

Keywords: land surface temperature; landscape pattern analysis; single-channel algorithm; spatial random point analysis; urban green space cooling island.

Figure 1

Map of the study area showing: (a) its location in China; (b) the topographical status; and (c) the township administrative boundary (black polyline) and urban planning area (yellow circle) in 2015, where the circle at the center marks the intersection of three mountains (Baota mountain, Qingliang mountain, and Fenghuang mountain), as well as two rivers (Yanhe river and Duchuanhe river). Urban area (UA) is a kind of the construction lands within the urban planning area. The suburban area (SUA) is the area between the exterior of the urban area and the external outline of the study area, in the corresponding period, based on master planning and land use planning of Yan’an city from Baota District Branch of Yan’an Municipal Bureau of Land and Resources.

Figure 2

Map showing the distribution of sampling points in the region (plots).

Figure 3

Methods employed in this study, where the definitions of variables are given in Equations (1)–(19). ψ: atmospheric parameters; FV: vegetation fraction; NDVI: normalized difference vegetation index; MNDWI: modified normalized difference water index; IBI: an index-based built-up index; Tsen: at-sensor temperature; ε: land surface emissivity; SUHI: surface urban heat island.

Figure 4

Comparison of the average LST among different LULC types and different regions (urban area of Yan’an city, and suburban area of Yan’an city) during six periods from 1990 to 2015 (error bars represent the standard deviation in the corresponding average LST).

Figure 5

(a–f) Maps comparing the spatial distribution of the land surface temperature and land use/cover during the three periods from 1990 to 2000; (g–l) Maps comparing the spatial distribution of the land surface temperature and land use/cover during the three periods from 2005 to 2015.

Figure 5

(a–f) Maps comparing the spatial distribution of the land surface temperature and land use/cover during the three periods from 1990 to 2000; (g–l) Maps comparing the spatial distribution of the land surface temperature and land use/cover during the three periods from 2005 to 2015.

Figure 6

Changes in the landscape metrics for the LULC and thermal landscape in urban and suburb areas: (a) fractal dimension index (FRAC); (b) division index (DI); (c) aggregation index (AI); (d) Shannon’s diversity index (SHDI); (e) percentage of landscape (PLAND); and (f) expansion intensity (EI).

Figure 7

Linear correlation analysis of LST versus the remote sensing ground indexes (NDVI, FV, IBI and MNDWI) in urban and suburban areas: (a) Pearson’s correlation coefficients (r); (b) adjusted R-square values (R²_adj) of spatial linear regression.