LiDAR and Hyperspectral data for Landscape and Vegetation Classification and Monitoring
2014 (English)In: Proceedings of the 7th International Congress on Environmental Modelling and Software (iEMSs)June 15-19, San Diego, California, USA / [ed] Ames, D.P., Quinn, N.W.T., Rizzoli, A.E., Manno, Switzerland: International Environmental Modelling and Software Society , 2014, Vol. 4, 2172-2179 p.Conference paper (Refereed)
Mapping of forest areas and other landscapes as to combine information about ground structures, topography as well as other natural and man-made features can be made with help of LiDAR (Elmqvist, M. 2001). The result can be used for planning military and civil missions and analysis of the possibility to drive though areas with bad or no roads (Sivertun & Gumos 2006) as well as for management of natural recourses and for example in physical planning. By combining LiDAR and other remotely sensed data it is possible to make use of the different advantages the different sensors provides. In this article based on a test in Linköping municipality, Sweden, we have employed the LiDAR based SingleTree™ detection model (Ahlberg at al 2008) and hyper spectral image data as to improve the classification of the trees and the ground surface under the trees. This method differs from similar models like in Béland et al. (2014) and Côté et al (2011) that uses terrestrial TLiDAR sensors to identify the species of trees.
By detecting returns of laser beams that passed through the vegetation and are reflected back to the sensor, it is possible to detect ditches, stones, logs and other obstacles to passing through the area. The data from modern LiDAR sensors can have very high spatial resolution, in many cases 50 points/m2 or more. By filtering the LiDAR data it is also possible to detect vehicles and man-made objects that are hidden under the vegetation, especially if the LIDAR uptake is compared with an earlier registration, movements and differences can be detected.
LiDAR registrations are today made by the forest industry in order to obtain better and more accurate information about the vegetation and improve their activities. Observation of the health of plants or trees becomes more important as a consequence from global warming and increased pressure from insects and diseases. There is also an increasing demand on forests and crops as to fill the demands from a growing and partly wealthier world (Kamaruzaman J. and Kasawani I., 2009). In forestry the LiDAR data are used to plan for harvest, building forest roads and timber transports. Another important source of data is Hyper Spectral Scenes (HSS) where the reflected solar light is analysed to identify anomalies in the spectral response and get a hint about the health of the canopy (Hyperspectral Imaging 2011). The difference from using multispectral images in comparison with other remotely sensed data is that the hyper spectral sensor delivers response in several hundred small and well-defined spectral wavelength bands. Those are supposed to indicate the biomass and water content as well as the difference between the absorption and the reflectance band for chlorophyll and many other conditions. A system can be used to identify the spectral signature in a certain area in order to decide what material and colours that should be used for camouflage. The data can be combined with LiDAR and used in the classification of forests, soils and other landscape features in Geographic Information Systems (GIS). Modern development of sensors and platforms makes it possible to use for example Unmanned Air Vehicles (UAVs) like helicopters to collect LiDAR and HSS data.
Place, publisher, year, edition, pages
Manno, Switzerland: International Environmental Modelling and Software Society , 2014. Vol. 4, 2172-2179 p.
LiDAR, Hyper spectral data, SingleTree™ detection, Landscape classification and monitoring, GIS
Computer and Information Science
Research subject Militärteknik
IdentifiersURN: urn:nbn:se:fhs:diva-5405ISBN: 978-88-9035-744-2OAI: oai:DiVA.org:fhs-5405DiVA: diva2:817181
7th International Congress on Environmental Modelling and Software, June 15-19 2014, San Diego, California, USA