Prediction of tree biomass in the forest–tundra ecotone using airborne laser scanning

Abstract

The effect of ongoing climate change on sub-arctic and alpine forests has led to increased interest in monitoring potential changes in the forest-tundra ecotone. In addition to climate change, insect damage, browsing pressure by herbivores such as moose and reindeer, as well as anthropogenic impacts will contribute to changes in the forest-tundra ecotone. These changes are difficult to monitor with manual methods because of the complex mosaic pattern of the ecotone. In this study, the possibility to predict maximum tree height, above ground tree biomass and canopy cover with airborne laser scanning (ALS) was therefore tested at a forest-tundra ecotone site near Abisko in northern Sweden (Lat. N 68°20', Long. E 19°01', 420-700 m a.s.l.). The forest in the area is dominated by mountain birch (Betula pubescens ssp. czerepanovii), which has highly irregular stem and canopy forms. Predictions from two different laser data acquisitions were compared. The first laser data set had 6.1 points m -2 and was obtained in 2008 with a TopEye MKII scanner carried by a helicopter flown at 500 m a.g.l. The second laser data set had 1.4 points m -2 and was obtained in 2010 with an Optech ALTM Gemini scanner carried by a fixed-wing aircraft flown at 1740 m a.g.l. Linear regression models were developed for the predictions using data from 73 sample plots with ten meter radius surveyed in 2009 and 2010. The relative RMSEs obtained for the TopEye and Optech data after leave-one-out cross-validation were, respectively, 8.8% and 9.5% for maximum tree height; 18.7% and 21.2% for above ground tree biomass; and, 16.8% and 18.7% for vertical canopy cover on plot level. The results were clearly improved by introducing a new procedure to compensate for unevenly distributed laser points. In conclusion, ALS has strong potential as a data source to map mountain birch biomass in the forest-tundra ecotone, even when using sparse point density ALS data.3

Change detection of mountain birch using multi-temporal ALS point clouds

“…The laser metrics created using nDSM values have lower for the reference plots and provide higher classification accuracy than metrics created from CH. Multi-temporal data requires measures to be spatially normalized at each time point to avoid problems with uneven distribution of laser points (Nyström et al 2012). There is a need for further research on spatial normalization methods when analysing multi-temporal data.…”

“…The 43 sample plots had six m radius and were subjectively selected to match the following criteria: (1) 90% of the trees in the range 1.5 2.5 m tall, (2) no trees taller than 3.5 m, and (3) no dominating bush layer (0-1.5 m) on the sample plot. In addition, 53 sample plots, inventoried in August 2009 and June 2010, from a previous study in the area (Nyström et al 2012) were used. These 53 existing sample plots had ten m radius and included some trees taller than 3.5 m. Only six m radius of the existing sample plots were used in the analysis and the sample plots were revisited in 2010 to assure they were unchanged.…”

“…[ Table 2] 2.4 Laser data processing TerraScan (Soinen 2010) was used to classify the point cloud into ground and non-ground in the same way as Nyström et al (2012). The statistical software R (R Development Core Team, 2011) and in house developed programs were used to further process the ALS point cloud.…”

“…Naesset and Gobakken (2008) using the height threshold as the lower limit and as the upper limit. A sum of squared canopy heights ( ) was calculated by taking the sum of squared canopy heights divided by the number of canopy height measurements above the height threshold (Nyström et al 2012). Vegetation ratio ( ) was calculated both by dividing the number of laser returns above the height threshold with the total number of returns inside the sample plot and alternatively by using a point-weighted approach (denoted ) according to Nyström et al (2012) using a radius of 0.59 m for TopEye and 1.00 m for Optech, where the radii are the same as in Nyström et al (2012).…”

“…Airborne laser scanning (ALS) can efficiently be used to estimate tree height, biomass, and canopy closure in the forest-tundra ecotone (Nyström et al 2012). In the future, series of ALS data might become available, such multi-temporal datasets will however likely be collected with various resolutions, scanning systems, system parameters, etc.…”