Benjamin Bechtel

Classification of Local Climate Zones Based on Multiple Earth Observation Data

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.

Robustness of Annual Cycle Parameters to Characterize the Urban Thermal Landscapes

Today, only few spaceborne sensors can deliver the thermal infrared information at medium scale required for urban heat island (UHI) modeling and assessment for applications in urban design. Despite recent advances in assessing urban thermal behavior, new methods for the estimation of UHI parameters are still needed. Relevant information about the thermal behavior is coded in the annual cycle of surface temperatures. In this letter, the robustness of parameters extracted from the annual cycle was tested. Therefore, a larger data set of Landsat surface temperatures was generated by cloud masking, and parameters from disjoint subsets were compared. Furthermore, a large number of virtual samples were generated to assess the accurateness of the extracted parameters. It was found that the mean and amplitude of the annual cycles in (black body) surface temperature can be estimated with an accuracy of approximately 1 K from 35 thermal infrared images. Hence, the multitemporal cycle parameters are an asset to characterize urban thermal landscapes.

A New Global Climatology of Annual Land Surface Temperature

Land surface temperature (LST) is an important parameter in various fields including hydrology, climatology, and geophysics. Its derivation by thermal infrared remote sensing has long tradition but despite substantial progress there remain limited data availability and challenges like emissivity estimation, atmospheric correction, and cloud contamination. The annual temperature cycle (ATC) is a promising approach to ease some of them. The basic idea to fit a model to the ATC and derive annual cycle parameters (ACP) has been proposed before but so far not been tested on larger scale. In this study, a new global climatology of annual LST based on daily 1 km MODIS/Terra observations was processed and evaluated. The derived global parameters were robust and free of missing data due to clouds. They allow estimating LST patterns under largely cloud-free conditions at different scales for every day of year and further deliver a measure for its accuracy respectively variability. The parameters generally showed low redundancy and mostly reflected real surface conditions. Important influencing factors included climate, land cover, vegetation phenology, anthropogenic effects, and geology which enable numerous potential applications. The datasets will be available at the CliSAP Integrated Climate Data Center pending additional processing.

Multitemporal Landsat data for urban heat island assessment and classification of local climate zones

The urban heat island is the most analyzed feature in urban climatology and an important application in urban remote sensing. A common approach is the application of thermal-infrared radiometers for inner urban temperature characterization, although several problems remain unsolved. Another approach is the direct determination of thermal surface properties and thematic classification using multitemporal thermal imagery. In this study multitemporal Landsat data is correlated with a long-term urban heat island pattern with R of up to 0.76. Furthermore, parameters of the annual cycle in surface temperatures at acquisition time are fitted and evaluated for different land use classes. The classes show significantly different probability density functions in the annual cycle parameters which therefore can be expected to be suitable features for their classification.

Estimation of Dense Time Series of Urban Air Temperatures from Multitemporal Geostationary Satellite Data

Monitoring and nowcasting of urban air temperatures are of high interest for prediction of heat stress in cities. Routine observation is so far limited by the complex coupling between atmosphere and land surface in urban areas, which makes estimation more difficult. In this study, we have investigated the capability of multitemporal land surface temperatures (LSTs) from the geostationary Spinning Enhanced Visible Infra-Red Imager instrument for estimation of urban air temperatures. The results are very promising with root-mean-square errors (RMSEs) of 1.5-1.8 K for six stations in Hamburg and explained variances of 97%-98%. Both the annual and diurnal cycles were well represented by the empirical models and the use of multitemporal data substantially increased the model performance. Further, the model was run in a forecast mode without actual LST information. Here, the best predictors reached RMSEs of 1.9-2.4 K and R2 of 95%-97% for a 2-h forecast.

Classification of Local Climate Zones Using SAR and Multispectral Data in an Arid Environment

There is an urgent need for more detailed spatial information on cities globally that has been acquired using a standard method to facilitate comparison and the transfer of scientific and practical knowledge between places. As part of the world urban database and access portal tools (WUDAPT) initiative, a simple workflow has been developed to perform this task. Using freely available satellite imagery (Landsat) and software (SAGA), WUDAPT characterizes settlements using the local climate zone (LCZ) scheme, which decomposes the city into distinctive neighborhoods (>1 km2) based on typical properties (e.g., green proportion and built fraction). In this paper, the methodology is extended to examine the effect of adding synthetic aperture radar (SAR) data, which is now freely available from Sentinel 1, for generating LCZs. Using the city of Khartoum as a case study, the results show that combining multispectral and SAR data improves the overall performance of several classifiers, with random forest (RF) performing the best overall.

Assessing the Capability of a Downscaled Urban Land Surface Temperature Time Series to Reproduce the Spatiotemporal Features of the Original Data

The downscaling of frequently-acquired geostationary Land Surface Temperature (LST) data can compensate the lack of high spatiotemporal LST data for urban climate studies. In order to be usable, the generated datasets must accurately reproduce the spatiotemporal features of the coarse-scale LST time series with greater spatial detail. This work concerns this issue and exploits the high temporal resolution of the data to address it. Specifically, it assesses the accuracy, correct pattern formation and the spatiotemporal inter-relationships of an urban three-month-long downscaled geostationary LST time series. The results suggest that the downscaling process operated in a consistent manner and preserved the radiometry of the original data. The exploitation of the data inter-relationships for evaluation purposes revealed that the downscaled time series reproduced the smooth diurnal cycle, but the autocorrelation of the downscaled data was higher than the original coarse-scale data. Overall, the evaluation process showed that the generation of high spatiotemporal LST data for urban areas is very challenging, and to deem it successful, it is mandatory to assess the temporal evolution of the urban thermal patterns. The results suggest that the proposed tests can facilitate the evaluation process.

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

Local climate zones (LCZs) divide the urban landscape into homogeneous types based on urban structure (i.e., morphology of streets and buildings), urban cover (i.e., permeability of surfaces), construction materials, and human activities (i.e., anthropogenic heat). This classification scheme represents a standardized way of capturing the basic urban form of cities and is currently being applied globally as part of the world urban database and portal tools (WUDAPT) initiative. This paper assesses the transferability of the LCZ concept to two Ukrainian cities, i.e., Kyiv and Lviv, which differ in urban form and topography, and considers three ways to validate and verify this classification scheme. An accuracy of 64% was achieved for Kyiv using an independent validation dataset while a comparison of the LCZ maps with the GlobeLand30 land cover map resulted in a match that was greater than 75% for both cities. There was also good correspondence between the urban classes in the LCZ maps and the urban points of interest in OpenStreetMap (OSM). However, further research is still required to produce a standardized validation protocol that could be used on a regular basis by contributors to WUDAPT to help produce more accurate LCZ maps in the future.

Developing a community-based worldwide urban morphology and materials database (WUDAPT) using remote sensing and crowdsourcing for improved urban climate modelling

This paper outlines the WUDAPT (World Urban Database and Access Portal Tools) concept and highlights progress to date in developing this database for cities around the world. The next steps in the WUDAPT project are outlined, both in the immediate and longer term. Ultimately the goal is to provide an open access resource on urban morphology, materials and metabolism for all major cities that can be used for many different applications, in particular for climate and weather modelling, and urban climate change studies.

Towards an urban roughness parameterisation using interferometric SAR data taking the Metropolitan Region of Hamburg as an example

Increased surface roughness and the enhanced drag effect of the urban surface are important alterations in urban boundary layer dynamics. An enhanced roughness mapping would be beneficial for several modelling applications since state of the art roughness parameters derived from land use data do not represent the heterogeneity of urban surfaces at higher resolutions. Today most morphometric methods of urban roughness estimation are based on empirically derived functions of obstacle geometry parameters, which have conceptual and practical shortcomings. In this study we propose a new approach based on the topology, statistics and texture of overall roughness elements as represented by a Digital Height Model generated from side-looking Interferometric Synthetic Aperture Radar. These state of the art remotely sensed high resolution data are readily available for Western Europe, the US, as well as parts of Asia and the Caribbean. A set of new angular dependent urban morphology parameters is developed and tested for suitability for the mapping of roughness characteristics by way of an improved methodology. The results are very convincing and consistent over a wide range of surface characteristics.