ISPRS-Annals ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences ISPRS-Annals ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci. 2194-9050 Copernicus Publications Göttingen, Germany 10.5194/isprs-annals-V-4-2022-311-2022 URBAN FUNCTIONAL DISTRICT IDENTIFICATION AND ANALYSIS FROM MULTI-SOURCE DATA Zhang Y. 1 2 Qin K. 3 Liu W. 1 2 Zhu X. 1 2 Peng Y. 1 2 Wang X. 1 2 Zhai X. 1 2 Zhao T. 1 2 Li R. 1 2 National Geomatics Center of China, 28 Lianhuachi West Road, Haidian District, Beijing 100830, China Key Laboratory of Spatio-temporal Information and Intelligent Services (LSIIS), MNR, 28 Lianhuachi West Road, Haidian District, Beijing 100830, China School of Remote Sensing and Information Engineering, Wuhan University, 129 Luoyu Road, Hongshan District, Wuhan 430079, China 18 05 2022 V-4-2022 311 318 Copyright: © 2022 Y. Zhang et al. 2022 This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/4.0/ This article is available from https://isprs-annals.copernicus.org/articles/V-4-2022/311/2022/isprs-annals-V-4-2022-311-2022.html The full text article is available as a PDF file from https://isprs-annals.copernicus.org/articles/V-4-2022/311/2022/isprs-annals-V-4-2022-311-2022.pdf

Residents’ activities have a significant interaction with urban socioeconomic environment. Taxi trajectory data has been widely used to mine human activity patterns to identify urban functional districts. However, previous studies merely chose several spatiotemporal statistics of taxi pick-up and drop-off points. This paper compares seven time series statistics of taxi pick-up and drop-off points, and selects the best combination to identify urban functional districts. The basic analysis units are not only constructed based on the OpenStreetMap data, but also optimized with the fine-grained clean rasterized pixels, generated from preprocessed taxi trajectory data through the improved head/tail breaks method. The experiment conducted in Wuchang District, Wuhan, shows that the combination of the average statistics of pick-up points, the average statistics of drop-off points, and the ratio statistics of pick-up and drop-off difference achieves the best identification precision of 83.65%, the F1-score of 82.2%, and the recall score of 81.48%. The proposed approach has good scalability and can be transplant to other identification applications.