311 / 2019-02-18 14:39:00

Analyzing the Spatial and Temporal Variation of the Potential Hotspots in Wuhan from a Local Scale

land surface temperature,Local Scale,,hotspots,,Latent Pattern,surface morphology

全体主题 > Spatial statistics

The surface thermal environment has been impacted by the rapid urbanization with the replacement of natural land cover by artificial impervious surface. Urban Heat Island effect (UHI), convinced to influence the region climate and socioeconomic development, is crucial to understand the variation of Land Surface Temperature (LST). Current studies are mostly at city scale where cities are treated holistically, and the temperature heterogeneity within urban areas are ignored due to the constraint of scale. Moreover, current studies mainly employ raw LST data retrieved from remotely sensed satellite images, which contains noises interfering the extraction and analysis of UHI. This study, on one hand, follows the cutting-edge method to identify the potential hotspots based on the Latent pattern of LST (LLST), a smooth and noise-free surface representing the spatial heterogeneity in a typical time quantum within a city, and the morphology of LLST. On the other hand, this study analyzes the spatially and temporally heterogeneous distribution of the potential hotspots using Standard Deviational Ellipse (SDE) at local scale, which is consistent with the strategies in urban planning.
Specifically, by focusing on the LST of Wuhan, China, the 8-day MODerate-resolution Imaging Spectroradiometer (MODIS) Aqua LST 1km Grid products (MYD11A2), from 2005 to 2017 with a 6-year time interval are utilized. The non-parameteric Multi-Task Gaussian Process model (MTGP) is adopted to generate the LLST. The Multi-Scale Shape Index (MSSI) is applied to numerically characterize the morphological features of LLST subsequently. k-means is employed to investigate the potential number of class and provide Back-Propagation Neural Network (BPN) with heuristic knowledge, and BPN is then used to classify the surface regions into various classes. The potential hotspots are identified as the regions with higher LLST and more critical morphological deformations. SDE is then utilized on every snapshot to analyze the temporal dynamics of directional characteristics and concentration degree of spatial distribution of hotspots. The LST sensitive empirical indexes, specifically Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), Impervious Surface Index (ISI), are used to explore the relationship between land cover and the potential hotspots. This research can serve both the fields of temperature and urban planning by two approaches, 1) the spatial and temporal heterogeneity of LST within a city can be characterized from the process-based perspective of phenomena; 2) the identification and analysis of the potential hotspots at the local scale assists better for optimization measures of surface thermal environment to be practical.