COS 147-1
Quantifying spatiotemporal patters of urban heat island in Beijing: An improved assessment using time-series LANDSAT LST data and GaoFen-1 data

Friday, August 14, 2015: 8:00 AM
341, Baltimore Convention Center
Kai Liu, Key Laboratory of Water Cycle & Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources, Chinese Academy of Sciences, China
Weimin Wang, Shenzhen Environmental Monitoring Center, Shenzhen, China
Lijun Yang, Shenzhen Environmental Monitoring Station, Shenzhen, China
Hong Liang, Shenzhen Environmental Monitoring Station
Hongbo Su, Key Laboratory of Water Cycle & Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources, Chinese Academy of Sciences

Assessing the spatial pattern and temporal characteristic of surface urban heat island (SUHI) is essential for better understanding the urban thermal environments and associated urban ecological planning. The purpose of this study was to comprehensively analyze a short-time spatial variation of urban land surface temperature (LST) in relation to land cover features and associated landscapes components. We delineated the SUHI of Beijing, China, using LST data set from the MODIS/Terra and LANDSAT 8 TIRS over the period of warm season from May to November in 2013. The Spatial and Temporal Adaptive Fusion Model (the STARFM model) developed by Gao was employed to yield LST time series of high spatial resolution and reliable temporal distribution over urban areas in an interval of approximate 10 days. The effects of four main landscapes components on SUHI were also investigated on the basis of two scenes of high spatial resolutions GoFen (GF) images acquired in the summer days (June. 10 and August. 13 in 2013). 


LST differences were found among various land cover types, with LST for the impervious surfaces about 3-8 K higher than for the urban greens through most of the study period and peak in July/August. Our results clearly show that SUHI does exist in Beijing in the months from May to Oct, yet the temporal variation still needs further investigation. Inspections of the statistical relationships between LST and landscape components reveal that LST was correlated with landscape metrics used, as well the percentage of land cover was the most important correlated variable. However, these statistical relationships do not seem to be very significant, and are sensitive to local meteorological conditions and varying with temporal trends regarding on impervious landscapes components. The STARFM model is shown to have the potential in achieving reasonable simulation of LST results and facilitating the delineation of landscape factors for SUHI study. Our study also suggests that new launched optical sensors with high spatial resolutions (2 m/ 8 m for visible-NIR and Panchromatic band), such as GF, can be used to reveal the relationship between urban thermal environments and the associated landscapes ecology at a fine local scale. In addition, to minimize the effects of urban heat island, urban planners should dedicate to urban design, mainly involving of reasonably allocating rational landscape components and increasing green space fractions.