Classification of Local Climate Zones Based on Multiple Earth Observation Data

Abstract

Considerable progress was recently made in the determination of urban morphologies or structural types from different Earth observation (EO) datasets. A relevant field of application for such methods is urban climatology, since specific urban morphologies produce distinct microclimates. However, application and comparability are so far limited by the variety of typologies used for the description of urban surfaces in EO. In this study Local Climate Zones (LCZ), a system of thermally homogenous urban structures introduced by Stewart and Oke, was used in a pixel-based classification approach. Further, different EO datasets (including satellite multitemporal thermal and multispectral data as well as a normalized digital surface model (NDSM) from airborne Interferometric Synthetic Aperture Radar) and different classifiers (including Support Vector Machines, Neural Networks and Random Forest) were evaluated for their performance in a common framework. Especially the multitemporal thermal and spectral features yielded high potential for the discrimination of LCZ, but morphological profiles from the NDSM also performed well. Further, sets of 10-100 features were selected with the Minimum Redundancy Maximal Relevance approach from multiple EO data. Overall classification accuracies of up to 97.4% and 95.3% were obtained with a Neural Network and a Random Forest classifier respectively. This provides some evidence that LCZ can be derived from multiple EO data. Hence, we propose the typology and the method for the application of automated extraction of urban structures in urban climatology. Further the chosen multiple EO data and classifiers seemed to yield considerable potential for an automated classification of LCZ.

GIS-based delineation of local climate zones: The case of medium-sized Central European cities

“…Our analysis shows that the decision-making algorithm for defining the percentage coverage for individual LCZs showed good agreement (in 79-89% of cases) with areas defined on the basis of expert knowledge. When the distribution of LCZs on the basis of our method and the method of Bechtel and Daneke (2012) was compared, the results were broadly similar; however, considerable differences occurred for LCZs 3, 5, 10, D, and E. It …”

“…for modelling Bokwa et al, 2015;Geletič et al, 2016): and 3. for mapping intra-urban land cover (Bechtel and Daneke, 2012;Lelovics et al, 2014;Danylo et al, 2016). Bechtel and Daneke (2012) and Lelovics et al (2014) created the first LCZ mapping methods and moved the LCZ concept toward a generally recognised regional typology.…”

“…for modelling Bokwa et al, 2015;Geletič et al, 2016): and 3. for mapping intra-urban land cover (Bechtel and Daneke, 2012;Lelovics et al, 2014;Danylo et al, 2016). Bechtel and Daneke (2012) and Lelovics et al (2014) created the first LCZ mapping methods and moved the LCZ concept toward a generally recognised regional typology. With such a radical shift in the LCZ concept some new methodological problems appeared: the size of a spatial mapping unit (pixel size); the method used for generalisation; the temporal variability of the physical properties of the environment; the objectification and standardisation of the classification procedure; and other issues (Bechtel and Daneke, 2012;Lelovics et al, 2014;Lehnert et al, 2015).…”

Spatial and temporal variability of urban soil water dynamics observed by a soil monitoring network

“…These urban land use types represent 7.8 % (green space) and 21.5 % (residential buildings) of the total area of the city of Hamburg (Statistisches Amt für Hamburg und Schleswig-Holstein 2014). Categorizing them into the scheme of local climate zones (LCZ) by Stewart and Oke (2012), compared with results from Bechtel and Daneke (2012), they match the LCZ field/low plants with 0 to 5 % surface sealing ratio, and regular housing/open low rise with about 50 % surface sealing ratio, respectively. The sites are named according to the district with BS^(shallow groundwater table district) and BD^(deep groundwater table district) and by reflecting land use with the suffix B_G^(green space) and B_H^(housing areas).…”

Correction of approximation errors with Random Forests applied to modelling of cloud droplet formation

“…Recently, the RF models have been applied to classification and regression problems including classification of climate zones (Bechtel and Daneke, 2012), earthquake induced damages (Tesfamariam and Liu, 2010), remote-sensing data (Pal, 2005) and disease prediction (Munro et al, 2006;Yao et al, 2013). In papers by Bechtel and Daneke (2012), Tesfamariam andLiu (2010), Pal (2005), a comparison between different algorithms were carried out. Despite its simplicity, the RF was observed to perform at least equally well as the more complicated algorithms in classification and regression problems.…”