Identifying Local Scale Climate Zones of Urban Heat Island From Hj-1b Satellite Data Using Self-Organizing Maps

Abstract

ABSTRACT:With the increasing acceleration of urbanization, the degeneration of the environment and the Urban Heat Island (UHI) has attracted more and more attention. Quantitative delineation of UHI has become crucial for a better understanding of the interregional interaction between urbanization processes and the urban environment system. First of all, our study used medium resolution Chinese satellite data-HJ-1B as the Earth Observation data source to derive parameters, including the percentage of Impervious Surface Areas, Land Surface Temperature, Land Surface Albedo, Normalized Differential Vegetation Index, and object edge detector indicators (Mean of Inner Border, Mean of Outer border) in the city of Guangzhou, China. Secondly, in order to establish a model to delineate the local climate zones of UHI, we used the Principal Component Analysis to explore the correlations between all these parameters, and estimate their contributions to the principal components of UHI zones. Finally, depending on the results of the PCA, we chose the most suitable parameters to classify the urban climate zones based on a Self-Organization Map (SOM). The results show that all six parameters are closely correlated with each other and have a high percentage of cumulative (95%) in the first two principal components. Therefore, the SOM algorithm automatically categorized the city of Guangzhou into five classes of UHI zones using these six spectral, structural and climate parameters as inputs. UHI zones have distinguishable physical characteristics, and could potentially help to provide the basis and decision support for further sustainable urban planning.

Mapping Local Climate Zones: A Bibliometric Meta-Analysis and Systematic Review

“…Germany, whileXu et al (2017b) reported that RF and SVM performed slightly better than NN in Guangzhou and Wuhan, China.The object-based LCZ mapping methods generally use image segmentation algorithms to obtain objects, and then use machine learning algorithms to classify objects into LCZs. For example,Wei & Blaschke (2016) used the multi-resolution segmentation and self-organization map (SOM) algorithms to classify LCZs. dosAnjos et al (2017) used the multiresolution segmentation and random forest (RF) algorithms for LCZ classification Simanjuntak et al (2019).…”

Influence of the choice of geographic unit on the response of urban thermal environment:Taking Beijing as an example

“…地 理 科 学 进 展 第 40 卷 分异， 过程则是地理现象的时空演变 [5] 。 随着城市化的快速发展， 城市地域的景观格局 趋于复杂， 因此， 有必要在城市地域进行跨单元研 究。城市热环境效应作为城市地域地理学研究的 重要部分， 地理单元划分是研究其作用机理的重要 基础。现阶段国内外学者对于地理单元划分的选 择多基于数据的可获得性， 对自然要素和人文要素 作用机制进行探讨。他们多基于土地利用分类单 元探究城市地表温度与归一化差值植被指数(normalized difference vegetation index, NDVI)、 归一化 差 值 建 筑 指 数 (normalized difference build-up index, NDBI)、 归一化差值水体指数(normalized difference water index, NDWI) [6][7][8][9] 或 斑 块 密 度 (patch density, PD)、 香 农 多 样 性 指 数 (Shannon's diversity index, SHDI)、 最大斑块面积指数(largest patch index, LPI) [10][11][12] 等因素的关系。人口普查街区作为研 究单元则多关注社区社会经济、 人口热脆弱性与热 岛的关联 [13][14] 。网格则是探究城市热岛(urban heat island, UHI)尺度效应常用的单元 [15] 。局地气候区 在近几年热岛研究中也得到学者青睐， 多用来探 究 UHI 的时空分布特征 [16][17][18] 。学者同样也基于城 市群 [19][20] 、 气候带 [21][22] 等大尺度单元对多城市进行对 比研究。在景观组分与热岛关联的研究中， 植被指 数 NDVI、 不 透 水 面 比 例 (impervious surface area, ISA)、 水体指数 NDWI 被广泛使用 [23][24][25][26] 露岩石和水体 5 种自然下垫面(图 1) [28] 。…”