Mapping Local Climate Zones for a Worldwide Database of the Form and Function of Cities

Abstract

Progress in urban climate science is severely restricted by the lack of useful information that describes aspects of the form and function of cities at a detailed spatial resolution. To overcome this shortcoming we are initiating an international effort to develop the World Urban Database and Access Portal Tools (WUDAPT) to gather and disseminate this information in a consistent manner for urban areas worldwide. The first step in developing WUDAPT is a description of cities based on the Local Climate Zone (LCZ) scheme, which classifies natural and urban landscapes into categories based on climate-relevant surface properties. This methodology provides a culturally-neutral framework for collecting information about the internal physical structure of cities. Moreover, studies have shown that remote sensing data can be used for supervised LCZ mapping. Mapping of LCZs is complicated because similar LCZs in different regions have dissimilar spectral properties due to differences in vegetation, building materials and other variations in cultural and physical environmental factors. The WUDAPT protocol developed here provides an easy to understand workflow; uses freely available data and software; and can be applied by someone without specialist knowledge in spatial analysis or urban climate science. The paper also provides an example use of the WUDAPT project results.

Contributing to WUDAPT: A Local Climate Zone Classification of Two Cities in Ukraine

“…However, it has yet to be fur-84 ther tested and validated on other cities than those previously 85 reported, i.e., Dublin, Houston, and Hamburg. Although build-86 ing heights and building densities differ between the urban 87 classes, it is possible to use spectral differences in urban 88 materials and cover to differentiate urban structure, negating 89 [12]. 160 The algorithm to create the LCZ classification requires train- 161 ing data.…”

“…239 An additional experiment was undertaken in which the mini-240 mum, mean, and maximum value of the resampled 120 m bands 241 were provided to the classifier, increasing the number of inputs 242 (or features) from 48 to 144 for Kyiv and from 60 to 180 for 243 Lviv. The idea was to determine whether providing additional 244 information about the spectral variation to the classifier, which 245 Two new steps were then added to the workflow of [12]. 248 The first was to undertake an independent validation using a 249 random stratified sample (Fig.…”

“…ware [23] from [12], the workflow was processed using R. Each were used in the classification since random forest is relatively 215 insensitive toward the number of features. All bands from the 216 five scenes were resampled using the area mean to a common 217 resolution of 120 m, which is within the range of 100-150 m 218 recommended by [12]. Therefore, 48 inputs were provided to 219 the random forest classifier for Kyiv (to include all four scenes) 220 and 60 inputs were used in total for Lviv.…”

“…The WUDAPT philosophy is based 74 on the use of data that are freely available and can be pro-75 cessed in a simple workflow using free software for any city 76 in the world. Numerous multispectral, thermal, and morpho-77 logical features as well as machine learning methods have 78 been tested for discrimination of LCZs [11] and subsequently 79 a workflow based on Landsat imagery and random forest has 80 been developed by [11] and [12] and implemented in SAGA. 81 Single studies have applied the method to cities with dif-82 ferent climatic and cultural backgrounds including Khartoum 83 in this Special Issue [13].…”

“…The user's accuracy was estimated 185 at around 87% for this class, whereas the overall accuracy for 186 all classes in this global product is around 80% [17]. ware [23] from [12], the workflow was processed using R. Each were used in the classification since random forest is relatively 215 insensitive toward the number of features. All bands from the 216 five scenes were resampled using the area mean to a common 217 resolution of 120 m, which is within the range of 100-150 m 218 recommended by [12].…”