Tree mapping using airborne, terrestrial and mobile laser scanning - A case study in a heterogeneous urban forest

: Holopainen M, Kankare V, Vastaranta M, Liang XL, Lin Y, Vaaja M, Yu XW, Hyyppä J, Hyyppä H, Kaartinen H, Kukko A, Tanhuanpää T, Alho P

Publisher: ELSEVIER GMBH, URBAN & FISCHER VERLAG

: 2013

: Urban Forestry and Urban Greening

: URBAN FORESTRY & URBAN GREENING

: URBAN FOR URBAN GREE

: 4

: 12

: 4

: 546

: 553

: 8

: 1618-8667

DOI: https://doi.org/10.1016/j.ufug.2013.06.002

We evaluated the accuracy and efficiency of airborne (ALS), terrestrial (TLS) and mobile laser-scanning (MLS) methods that can be utilized in urban tree mapping and monitoring. In the field, 438 urban trees located in park and forested environments were measured and mapped from our study area located in Seurasaari, Helsinki, Finland. A field reference was collected, using a tree map created manually from TLS data. The tree detection rate and location accuracy were evaluated, using automatic or semiautomatic ALS individual tree detection (ALS(ITDauto) or ALS(ITDvisual)) and manual or automatic measurements of TLS and MLS (TLSauto, MLSauto, MLSmanual, MLSsemi). Our'results showed that the best methods for tree detection were TLSauto and MLSmanual which detected 73.29% and 79.22% of the reference trees, respectively. The location accuracies (RMSE) varied between 0.44 m and 1.57 m; the methods listed from the most accurate to most inaccurate were MLSsemi, TLSauto, MLSmanual, MLSauto, ALS(ITDauto) and ALS(ITDvisual). We conclude that the accuracies of TLS and ALS were applicable for operational urban tree mapping in heterogeneous park forests. MLSmanual shows high potential but manual measurements are not feasible in operational tree mapping. Challenges that should be solved in further studies include ALS(ITDauto) oversegmentation as well as MLSauto processing methodologies and data collection for tree detection. (C) 2013 Elsevier GmbH. All rights reserved.