The Accuracy Assessment of DTM Generated from LIDAR Data for Forest Area  a Case Study for Scots Pine Stands in Poland KRZYSZTOF STEREÑCZAK 1 *, MICHAÙ ZASADA 2 AND MICHAÙ BRACH 3 1* Department of Information Technology and Modelling, Forest Research Institute  IBL, 3 Braci Leúnej str., Sækocin Stary, 05-090 Raszyn, Poland, +48 22 7150 325, K.Sterenczak@ibles.waw.pl 2 Laboratory of Dendrometry and Forest Productivity, Faculty of Forestry, Warsaw University of Life Sciences  SGGW, 159 Nowoursynowska str., 02-776 Warszawa, Poland 3 Department of Forest Management, Geomatics and Forest Economics, Faculty of Forestry, Warsaw University of Life Sciences  SGGW, 159 Nowoursynowska str., 02-776 Warszawa, Poland Stereñczak, K., Zasada, M. and Brach, M. 2013. The Accuracy Assessment of DTM Generated from LIDAR Data for Forest Area  a Case Study for Scots Pine Stands in Poland. Baltic Forestry 19(2): 252262. Abstract The paper presents the analysis of the influence of various factors on the accuracy of the Digital Terrain Model (DTM) generated under forest conditions in pine stands. DTMs with 0.5, 1, 2, 3, 4, 5 and 10 meter pixel sizes have been evaluated. The accuracy assessment has been performed for 4 age classes (developmental stages) in flat terrain and on slopes. Additionally, the analysis of differences between values obtained from single pixels and from 3x3 pixels moving window for various DTM pixel size has been performed. The General Linear Model for repeated measures has been applied for the analyses. Only the overstocked young stands show different patterns of relationships between the DTM accuracy and resolution. On average, model error for slopes was 3 times larger than for the flat terrain. The height obtained from DTM created using direct pixel values of height is slightly more accurate than that obtained using values smoothed by 3x3 pixel windows, the difference, however, is not statistically significant. The use of percentiles is much more appropriate than mean-based measures, especially for not normally distributed and skewed data. Key words: digital terrain model, airborne laser scanning, repeated measures ANOVA, error analysis, percentile, earth surface, Scots pine, slope, survey. Introduction The digital model of the earth is one of the most frequently used spatial data. In most cases, the term Digital Surface Model (DSM) represents the earths surface and includes all objects on it. In contrast to a DSM, the Digital Terrain Model (DTM) represents the bare ground surface without any objects, like for ex- ample plants and buildings (Li et al. 2005). Accurate and reliable Digital Terrain Model is one of the most important sources of data for Geographic Information Systems (GIS) and photogrammetry workflows, anal- yses and studies. It can be used in a number of ways: to assess terrain parameters, to visualize and analyze the earths surface, to rectify airborne and satellite imagery, to perform spatial analyses and to build oth- er spatial models. The potential variation of this ap- plication allows it to be used as a base for numerous engineering and environmental analyses. One of the most commonly used digital models in forestry and in environmental study is the Crown Height Model (CHM). The CHM is created by subtracting the Digit- al Terrain Model Z (elevation) pixel values from the Digital Surface Model elevation pixel values. It can be used for the extraction of many forest parameters, for example stand density (N¿sset et al. 2004), tree height (N¿sset et al. 2004) and biomass or stand volume (Maltamo et al. 2004). The quality of DTM influences the excellence of CHM (Leckie et al. 2003, Maltamo et al. 2004, N¿sset et al. 2004, Straub et al. 2006, Stereñ- czak et al. 2008, Zawiùa-Niedêwiecki et al. 2008) and augments the dimension of topographic objects (Koch and Heipke 2006). There are a few basic data sources for the DTM generation. The most traditional method is a classical field survey. Today this is usually done with the use of total station or GNSS receivers. This is a very ac- curate method, but also expensive and time consum- ing (Grala and Brach 2009, Yakar 2009). A more effi- cient and cost effective method is based on the ISSN 2029-9230 252 2013, Vol. 19, No. 2 (37) K. STEREÑCZAK ET AL. BALTIC FORESTRY THE ACCURACY ASSESSMENT OF DTM GENERATED FROM LIDAR DATA /.../