DTM MODELS TO ENHANCE PLANNING OF TIMBER HARVESTING

This paper studies the applicability of DTM with the resolution of 4 × 4 m for the analysis of macro-topographic factors (terrain slope, aspect, terrain ruggedness index) and one part of micro-topographic factors (occasional and constant streams) as features important for vehicle mobility during timber skidding. The analysis of directions of timber extraction in relation to the spatial position of primary forest traffic infrastructure of the study area was conducted in order to determine from which forest areas timber will be extracted up or down the slope (moving of loaded vehicle). Determination of water bodies (streams) and the surrounding sensitive areas was carried out using GIS tool TauDEM. Unevenness of the terrain was determined based on the Terrain Ruggedness Index (TRI) which showed moderately to very rugged terrain on 60.1% of the research area where vehicle mobility could be difficult (if not impossible) i.e. the necessity of a secondary forest road network is clear. DTM analysis of study area regarding vehicle (skidder) mobility and possible planning of timber extraction indicated different availability and quality of data. Digital terrain models present a good basis for the analysis of key constraints for forestry vehicles mobility or terrain trafficability (slope and direction of timber extraction). Using DTM of higher resolution (e.g. LiDAR images), will increase the accuracy of the results and the quality of the analysis.


INTRODUCTION
Planning in forestry is based on three different levels: strategic, tactical and operational (Sessions et al. 2007), but planning of timber harvesting together with forest opening (also planning of silviculture and forest protection activities) considers direct planning of specific filed activities and technologies (Čavlović 2013).At the strategic level of planning timber harvesting and forest opening, due to the terrain diversity, various stand conditions and different ways of forest management, primarily relate to: 1) rough descriptive terrain classifications based on terrain slope analysis (Löf-classifications based on terrain slope analysis (Löf-classifications based on terrain slope analysis (Löf fler 1984, Rowan 1977), 2) primary forest road density analysis (Pentek i Poršinsky 2012, Pentek i dr. 2014), 3) functional terrain classifications related to the selection of possible harvesting systems (Findeisen 2008, MacDonald 1999, McEwan et al. 2013, Pischedda 2004), 4) wood supply chain analysis concerning different relief areas (Epstein et al. 2007, D'Amours et al. 2008).
Timber harvesting is determined by procedures, methods of processing, machines and tools that are used during tending and regeneration as silvicultural process, where the impacts on the choice of mechanization have terrain factors (slope, surface obstacles, soil bearing capacity).Vehicle mobility is the ability of vehicle to move in the terrain while performing its primary mission, while terrain trafficability is its property to allow the passage of a vehicle, where terrain conditions show their influence.
From the standpoint of planning harvesting operations and forest opening at the strategic level, terrain slope is the most important field parameter that directly affects the selection of the harvesting systems.Terrain slope affects vehicle stability during its movement because all the wheels of the vehicle collide with equal macrotopographic values.Kühmaier and Stampfer (2010) consider that terrain slope of up to 30% (regardless of timber extraction direction) is the average limiting value for ground based (wheel) systems, which also depends on surface conditions, surface obstacles and soil bearing capacity.For terrain slopes between 30 and 50% Heinimann (2000) proposes the use of forest skylines, and for the slopes above 50% he recommends the use of helicopters for timber extraction.
A special form of micro-topographic terrain parameters that affect timber harvesting are natural or constructed water bodies.Watercourses of either occasional or permanent character, affect the performance of forestry activities not only as a surface barrier that vehicles cannot overcome (i.e.temporary passes must be built), but also acquire appropriate protective measures to avoid their contamination.During timber harvesting most common negative effects on water bodies are reflected in: 1) damage to the coastline and river beds, 2) the release of dangerous substances (oil, fuel, etc.), 3) soil erosion and 4) sedimentation of undesirable substances in aquatic ecosystems.The most common way to protect water bodies during forestry operations is by establishing protection zones around watercourses.Their width depends on: 1) the size (width) and the significance of the water body, 2) terrain slope of the surrounding area, 3) soil erodibility depending on its grain-size composition, 4) the type of forestry activities that will be performed (Adams and Storm 2011).Douglas (1986) describes the methods of determining water bodies and hydrological networks based on digital elevation models (DEM).Band (1986) on the basis of the Peuker and Douglas (1975) algorithm proposes the use of terrain depression as a possible field for watercourses.Peuker and Douglas (1975) algorithm defines a point (pixel) with the highest altitude in relation to its 4 adjacent points (pixels).Jenson and Domingue (1988) stated that the use of DEM to determine stream network, and its complexity, is associated with the quality and resolution of digital terrain models.Tarboton et al. (1991) define zones of watercourses (with the division of water streams into primary and secondary) on the basis of the variable decision threshold.The same author, in cooperation with the US Army Corps of Engineers in 1997, began to develop a computer program TauDEM (Terrain Analysis Using Digital Elevation Models) providing a watercourse network of watercourses and protection zones on the basis of DTM.

MATERIALS AND METHODS
The research was conducted on 562.06 hectares (forest covered area) of selective beech and fir forests of two management units ("Belevine" and "Kupjački vrh") in the Forest and Research Training Centre Zalesina which is managed by the Faculty of Forestry University of Zagreb.
Digital elevation model was created based on two sets of elevation data obtained by different methods, techniques and procedures of data collection: aerial-photogrammetric survey and stereo-restitution processing and digitalization of contour lines and angles of CBM (Croatian Base Map, 1:5000).Several layers of data were used for the creation of the digital elevation model (raster data) and with the computer program Global Mapper 13 converted into a triangulation irregular network (TIN), and then to grid network.The data were further processed and analyzed in the ArcGIS 9.3 computer programme.Errors (sink and peaks) resulting from the new DEM were corrected with the use of Raster Calculator and Focal Statistics.
Terrain slope, as one of the most important macro-topographic features, was determined on the basis of digital terrain model analysis with Spatial Analyst tools (application of ArcGIS program).This application uses the Burrough and McDonnell (1998) algorithm for calculating slope of the terrain.
Based on the recommendations for the terrain slope categorization for forest operations (Mellgren 1980, Löffler 1984Berg 1992and Rowan 1977), the analysis of terrain slope gave five different categories: 1) 0-10%, 2) 11-20%, 3) 21-33%, 4) 34-50 % and 5) > 51%.The analysis was also conducted by using tools of Spatial Analyst in ArcGIS program.Terrain aspect was determined with SAGA GIS tools (Terrain Analysis application) as well as Terrain Ruggedness Index (TRI).TRI was determined on the base of change in elevation and slope direction (aspect) according to the guidelines of Riley et al. (1999).Calculating TRI is based on total change (sum) in terrain elevation and slope direction between central cell (0.0) and its eight surrounding cells.
The direction of timber extraction (Krč andKošir 2008, Lubello 2008) was determined in the ArcGIS program on the basis of the assumption that cut timber is uniformly distributed over the entire forest area, with respect to the main macrotopographic terrain feature -slope and with respect to the shortest distance to the primary forest road network as well as distance to the closest public roads on which loading of timber is possible.
Water courses and surrounding areas were determined by tools within TauDEM application.TauDEM gave each cell of DEM (according to its elevation and aspect) a so called contributing area which showed the possibility of forming watercourses in the study area (Fig. 1).

RESULTS
The relief of MU "Belevine" is mildly undulating and fan-like striated with many beds of water streams (occasional and permanent character), many of which are not shown in the official topographic maps (for example on CBM).MU "Belevine" is divided to 18 compartments (18th being dislocated) and major part of its area (36.73%) is on terrain with slopes between 11-22% i.e. 105.49ha, follows slope class from 21 to 33% where 27.05% of MU area (79.13 ha) is located, then slope class 0-10% for 19% of MU (54.57ha) area.Slopes from 34 to 50% can be found on 12.31% of MU surface (35.36 ha) and the last slope class (also the most unfavourable one) -class 5 (slopes higher than 51%) can be found on 4.40% of the MU i.e. on 12.64 ha.
Relief of MU "Kupjački vrh" is filled with karst phenomennons and steep cliffs.Slope analysis showed a clear difference in the proportion of certain slope classes with respect to MU "Belevine".MU "Kupjački vrh" is divided into 16 compartments and major part of its area (32.16%) is also on terrain with slopes between 11-22% i.e. 88.41 ha.The similarity between this MU and MU "Belevine" now stops, because second biggest class (in terms of area coverage) is class 4 (slopes 34-50%) where 28.19% (77.49ha) of MU can be found, follows class 5 (slopes higher than 51%) with 23.69% (65.12 ha) of the MU.Favourable slope classes 1 and 2 together take only 15.96% (43.86 ha) of the MU area.Terrain slope aspect showed to which side of the world the sloping ground is exposed.The measured direction of 0° or 360° shows the orientation to the North, 90° to the East, 180° to the South and 270° to the West.Figure 4 shows the direction of the slope of the researched area.Changes in slope direction (aspect) are especially visible in the MU "Belevine" even though slope of this management unit is much more »mild« and definitely more favorable for executing forestry operations than those that can be found in the MU "Kupjački vrh".
Terrain Ruggedness Index was determined using data on terrain slope and aspect (Riley et al. 1999, Murkherjee et al. 2013) which gave four different TRI classes for the researched area (Fig. 5): 1. Undulating terrain on 39.88% of the researched area (TRI between 0-0.59), 2. Moderately rugged terrain on 36.56% of the researched area (TRI between 0.60-1.11),3. Rugged terrain on 17.75% of the researched area (TRI between 1.12-1.85),4. Very rugged terrain on 5.81% of the researched area (TRI between 1.86-4.32).The analysis showed that 60.10 % consists of unfavourable terrain conditions on 60.1% of the research area where vehicle mobility could be difficult i.e. the necessity of a secondary forest road network is clear.Terrain of MU "Kupjački vrh" is more difficult and unfavourable for performing forestry operations, but certain "difficulties" can also be found in MU "Belevine" where the change in slope and its direction resulted in an increase of TRI (for example compartments in 1, 2, 17 and 18).
The analysis of timber extraction direction was carried out on the basis of DTM and spatial position of the primary forest transport infrastructure (Fig. 6).It was conducted only in the ArcGIS programme by converting raster image and line layer (roads) to vector (point) layer.Several commands in Spatial Analyst tool were used (for example Spatial Join, Calculate Field etc.) which then gave the result as shown on the map (Fig. 6).Timber extraction direction represents parts of the forest area from which timber can be extracted up or down the slope of terrain i.e. movement of a It should be noted that this analysis refers to the movement of timber in theory, because neither the size of terrain slope nor the existence of secondary forest infrastructure were taken into account.
The conducted analysis showed that on 106.09 ha of MU "Belevine" cut timber would be extracted up the slope and 183.76 ha would be extracted down the slope of terrain.Similarly in MU "Kupjački vrh" 98.30 ha of timber would be extracted up the slope and 180.50 ha would be extracted down the slope of terrain.
The determination of water stream network and sensitive (protective) surrounding areas was done using TauDEM GIS tools: Flow Direction, Contributing Area and Stream Definition by Treshold.The results were then compared with stream networks from three different maps (Rauš 1975, CBM 1:5000, BSM 1:25,000), but only for MU "Belevine" (Fig. 7) because there was no information on the existing water network in MU "Kupjački vrh" and validation of TauDEM generated network could not be preformed for both management units.Figure 7 clearly shows that TauDEM tools can also be used in forestry for determining water stream network and surrounding sensitive areas, which definitely should be considered while planning timber harvesting operations.TauDEM tool confirmed that even though slope is more favorable in MU »Belevine«, changes in aspect should also be considered because these two terrain characteristics together give a more accurate picture of terrain conditions for forestry operations without performing field surveys.

DISCUSSION
It is known that terrain slope, its direction (aspect) and terrain forms affect vehicle mobility (Löffler 1984, Poršinsky 2005).An analysis of terrain slope indicated that in MU GJ "Belevine" terrain slope varies from 1% to 98%, with an average value of 22 ± 14%.Compared to MU "Kupjački vrh" values were a bit different, ranging from 1% to 152%, with a higher average value (39 ± 21%).However, ground based systems for timber extraction (the most common in Croatia according to Beuk et al. 2007) are also affected by ground obstacles and soil bearing capacity which were not visible on this DTM (Đuka 2014).
Terrain Ruggedness Index has, in the past, mostly been used for ecological monitoring and not for the planning of operations in timber harvesting.Division of TRI to 4 classes by Jensk optimization, Riley at all (1999) and Mukhrjee et al. (2013) highlighted that number of classes depends on a research goal.
Stream network was established by TauDEM application (Wechsler et al. 2009, Fan et al. 2014, Rampi et al. 2014).This tool is also helpful for discovering sensitive areas around streams (and their amount) where timber harvesting operations should be restricted/changed/supervised etc. Analysis showed that 44,69ha of MU "Belevine" can be considered as sensitive areas for timber harvesting operations, but also gave an insight into what a secondary forest network should look like.

CONCLUSION
Digital elevation models are a good basis for the analysis of key constraints regarding forestry vehicles mobility and terrain trafficability (terrain slope and timber extraction direction).The use of DTM's of higher resolution (e.g.LiDAR images) will surely increase accuracy of the results and quality of the analysis.By having information on terrain slope, aspect, Terrain Rouggedness Index, timber extraction direction and water protection areas without "leaving the office" one will have the opportunity for an easier forest operation planning regarding: • Secondary forest openness (choice between skid roads and trails i.e. choosing the right equipment and machinery for the construction phase; choosing the form of secondary forest network -regarding slope, primary forest roads, timber extraction direction and water protection areas), • Defining protection areas -regarding slope, aspect and water protection areas, • Marking areas for possible woodlots, • Harvesting systems -choosing vehicles based on terrain conditions i.e. slope class, TRI and protection areas.
Terrain Ruggedness Index should be definitely more explored, especially in terms of timber harvesting operations, because it combines terrain slope and aspect, which is very important for vehicle mobility.TauDEM tools should also be used more in the planning of timber harvesting, because that will not only show the possible water stream network, but also the surrounding sensitive areas which should be considered while performing forestry operations in an environmentally sound manner.Planning of timber harvesting should not be done without any field surveys (especially if DEM is of poor resolution), but GIS tools shown in this article give a good starting point in the determination of terrain characteristics.The best example for that are these two neighboring management units, "Belevine" and "Kupjački vrh", close to each other, both covered with selective forests of beech and fir, but completely different in terms of terrain characteristics.Knowing that before making any field surveys is definitely valuable information in terms of timber harvesting planning.

Figure 2 .
Figure 2. Share of slope classes in study area

Figure 3 .Figure 4 .
Figure 3. Slope classes of the researched area