Eur J Forest Res (2008) 127:395–406 DOI 10.1007/s10342-008-0223-9
ORIGINAL PAPER
Comparing methods for determining forest sites: a case study in Gu¨mu¨s¸ hane-Karanlıkdere forest Lokman Altun Æ Emin Zeki Bas¸ kent Æ Murat Bakkalog˘lu Æ Alkan Gu¨nlu¨ Æ _ Ali Ihsan Kadiog˘ullari
Received: 14 March 2007 / Revised: 9 January 2008 / Accepted: 3 June 2008 / Published online: 23 August 2008 Ó Springer-Verlag 2008
Abstract Forest site classification has long been a problem for managers of Turkish forests. Forest management decisions and land use planning involving afforestation activities and silvicultural prescriptions are based on sound site information to formulate appropriate actions on the ground. In Turkey, two methods of forest site productivity are used; direct and indirect method. Indirect methods are usually reserved for practical applications as they are relatively simple, yet provide less accurate estimation of the real productivity. In this study direct, indirect and remote sensing (RS) methods were used to distinguish and map forest sites of Karanlikdere Forest District in Gumushane, Turkey. One hundred and twenty-two sample plots were established with 300 m 9 300 m grids in summer of 2003. In each sample plot, soil samples and the classical timber inventory measurements were taken. According to direct method, water-air economy method is preferred due to a water deficiency in the study area. Four different forest site classes; very dry, dry, moderate fresh and fresh were determined and mapped with geographic information system (GIS). In indirect method, the guiding curve was used to generate anamorphic site indexes (SI) of three classes; good (SI = I and II), medium (SI = III) and poor (SI = IV
and V). Furthermore, forest sites were estimated with Landsat 7 ETM (2000) data using supervised classification with a 0.843 kappa statistic value and 88% accuracy assessments. Some important differences between the methods were discovered. The indirect method indicate that site indices I and II are 298.5 ha, III 254.3 ha and IV and V 347.7 ha. In contrast, direct method related to very dry site of 107.7 ha, dry site of 484.6 ha, moderate fresh site of 304.7 ha and fresh site of 246.3 ha. Satellite image indicate that very dry site covers 291.2 ha, dry site 239.2 ha, moderate fresh site 287.4 and fresh site 325.5 ha. Approximately 242.8 ha area (open and degraded areas) were not determined by indirect method but were captured with RS method. The statistical analyses (ANOVA) showed no statistically significant (F = 0.720, p = 0.543) relationship between indirect method and direct methodRS method indicating clearly that indirect method is not an adequate measurement of forest site productivity. Forest sites, particularly in open and degraded areas should be determined with direct method.
Communicated by A. Merino.
Introduction
_ Kadiog˘ullari L. Altun E. Z. Bas¸ kent A. Gu¨nlu¨ (&) A. I. Faculty of Forestry, Karadeniz Technical University, 61080 Trabzon, Turkey e-mail:
[email protected]
Estimating forest productivity is necessary for both effective forest management decisions and evaluation of site conditions for ecological studies. Forest site classification is a concept developed by forest managers to project the expected production of trees on a given site Carmean (1975), Sturtevant and Seagle (2004). It is therefore influenced by factors such as available light, heat, moisture and nutrients, along with other soil characteristics such as soil
M. Bakkalog˘lu Deputy of General Directory, General Directorate of Forest-Village Relations, Republic of Ministry of Environment and Forestry, So¨g˘u¨to¨zu¨ Cad. No:14/E Kat:13, 06560 Bes¸ tepe/Ankara, Turkey
Keywords Forest site classification GIS Direct method Indirect method Remote sensing method
123
396
depth and aeration Wang and Klinka (1996). Site is generally viewed as the integrated complex of all environmental factors within a prescribed area. Grey (1980), described site as a natural unit, a spatial entity, which can be described, classified, recorded and mapped but cannot be further subdivided without the loss of some intrinsic characteristic. A site can also be defined as the function of the interplay between climate, topography, parent material and vegetation over a specific time Bailey et al. (1978). According to forest management regulations, the main goal is to supply national requirement for forest crops by producing the most quality timber in a site. To benefit the multiple values of forests, forest site classification must be carried out for forest management planning Seckin and Kahveci (1993). It is necessary to determine and restrict forest sites by classifying and mapping its attributes for sustainable forestry. Without the site classification of landscape and the generation of the associated site maps causes failure of any forest management applications Baskent et al. (2003). To develop accurate long-term forest management plans, it is necessary to ascertain the site quality of each individual stand within a forest management unit so that its future growth and development can be forecasted Baskent et al. (2000). For years, foresters and ecologists have tried to develop a reliable forest site classification system that can be widely applied over a range of species and regions. Two principal approaches, direct and indirect method, have been used. The direct method classifies site based on the ground vegetation as plant communities are the results of various climatic, edaphic, and topographic factors associated with the stand’s growth potential. The indirect method relies more on simpler approach such as the site-index concept Alemdag (1991). One of the most widely used measures of site productivity is site index (SI), defined as the average height of dominant trees at a specific reference age Wang (1998), McKenney and Pedlar (2003). The assessment of site productivity in a forest area is fundamental to a good resource management. Accurate and reliable evaluation of forest site classification is necessary for forest resource management, in making cost-effective decisions about silvicultural investment Corona et al. (1998), Kayahara et al. (1998). According to direct method, forest site classification process is highly time demanding, expensive and hard to conduct. Although SI is the most widely accepted method for estimating site productivity of even-aged forests, it doesn’t reflect the forest sites appropriately (especially, treeless and degraded areas). Thus it necessitates the use of powerful information technologies such as geographic information system (GIS) and remote sensing (RS). To help develop forest site maps, modern information technologies such as GIS and RS should be utilized effectively. Various factors such as landscape
123
Eur J Forest Res (2008) 127:395–406
structure, climatic profile, biotic factors and soil characteristics have to be determined in examining forest site characteristics. This study attempts to identify forest sites using direct, indirect and remote sensing methods in a research area of Karanlıkdere Forest District located in northeastern part of Turkey. The research also focuses on classifying and mapping the forest sites with the spatial analysis functions of GIS. Furthermore, site productivity values determined by direct, indirect and RS methods were compared and contrasted to evaluate the potential use of the methods.
Study area Located in the Eastern Black Sea Region, the research area comprises 1,166 ha between 40°170 3900 –40°200 2800 north latitudes and 39°110 1400 –39°140 0600 east longitudes and about 40 km from Gu¨mu¨s¸ hane province, Turkey (Fig. 1). Elevation of the research area ranges from 1,545 to 2,270 m with an average of 1,908 m. The study area is situated on a steep topographic surface with a slope ranging from 33 to 70%, with an average of 50%. The area is located in the upper land of Hars¸ it River which carries the effects of the sea to the inland. But the climate of this area is moist, low temperature, highly water shortage during vegetation period and near the continental climate. Mean annual temperature is 5.5°C and the precipitation is 792.7 mm Erinc (1984), Anonymous (2001). Soils tend to be sand loam and loamy sand type. The study area covers mixed stands of pine ‘‘(Pinus sylvestris) L. and fir (Abies nordmanniona subsp. nordmannianna (Stev) Spach.’’ The forest stands are even-aged and high forest.
Materials and methods The research examines data for classifying and interpreting a specific forest site according to the degree of site productivity. In this assessment, a forest site delineated by topographic features such as elevation, aspect, slope and land surface was used. Forest site classification involved the soil, climate and physiographic factors, important to classify a site. Existing forest cover type map were digitized with Arc/Info ESRI (1999) and the sample plots were geographically distributed over the area on a 300 m 9 300 m grid. A total of 122 sample plots were generated, taking into consideration physiographic factors such as aspect, slope, land surface. Medium resolution satellite image (Landsat 7 ETM 2000) were used to estimate the site of the forest area (1143.3 ha). Direct, indirect and remote sensing methods were used to classify forest sites. The results were
Eur J Forest Res (2008) 127:395–406
397
Fig. 1 The research area with the location of the sample plots
compared to each other to evaluate the relative importance of each method. Direct methods Direct method states that the site productivity depends not only on the soil factors but also on topography (aspect, altitude, slope and landform) and climate data. Thus, forest site classification was conducted by combining edaphic, physiographic and climatic factors. As far as landform or physiographic classification is concerned, the research area was classified into six different elevation-climate zones (1,545–1,650, 1,650–1,750, 1,850–1,950, 1,950–2,050 and 2,050–2,270 m). Digital elevation models (DEM) contains the elevation of the terrain over a specified area, usually at
a fixed grid, displaying slope, aspect and landform. The source of the DEM data (at 10 m 9 10 m pixel resolution) comes from the contour line map with 10 m intervals digitized from digital topographic maps, registered with 6–8 m root mean square error with 3D modeling in GIS. The sample points in each of these zones were distributed according to physiographic structure of the area. The landform was stratified by slope and aspect subzones. East, north, north-west and north-east aspects were included in north aspect sub zone and the rest was grouped as south aspect subzone. The landscape was stratified into three slope groups; medium slope (17–32%), steep slope (33– 48%) and very steep slope ([48%) using GIS surface analyses function. The landscape surface was further stratified into six surface subzones; ridge, top hillside,
123
398
Eur J Forest Res (2008) 127:395–406
middle hillside, sub hillside, lowland and base land. The sample points were also grouped according to the physiographic stratification. Soil classification was carried out in each of the elevation-climate zone using the soil samples. Therefore, the soil was classified into bedrock types (Granit), soil deepness [medium deep (50–75 cm), deep (75–100 cm) and quite deep (100–125 cm)], rockiness (shallow, sparsely distributed rocks and densely packed rocks) and soil texture which describes the proportions of sand, silt and clay particles in the soil. The terms sand, silt and clay refer to different size fractions of the soil’s mineral content. Texture is quoted as the percentage of each of these components or, more usually, by internationally recognized shorthand of terms such as loam, sandy clay loam or silt clay. Soil samples were classified as sandy loam and loam sandy. Topographic parameters were used to delimit the forest sites while bedrock, soil depth, rockiness and soil texture were used to determine ecological units. Forest site maps such as soil depth, soil texture, rockiness, topographic surface or landform, slope groups and thus forest sites were produced using some of the spatial analysis functions of GIS. Two methods were used to classify and map the forest sites. The first method is based on soil nutrient regime where water deficit is nonexistent in summer months Altun (1995). The second method is applied according to soil moisture regime (SMR), where water deficit exists in summer months Kantarci (1980), Bakkaloglu (2003) and Gunlu (2003). There are not many meteorological stations in the study area to measure and gather the climate data specifically for each plot or stand. The meteorological stations in Gu¨mu¨s¸ hane, Torul ve S¸ iran are very close to the study area. Gu¨mu¨s¸ hane station is in the same watershed area with the average elevation of the study area and has long term average climate values. The climate related information of each sample area is harmonized according to its relative spatial location to the Gu¨mu¨s¸ hane meteorological station. The temperature decreases about 0.6°C in summer and 0.4°C in winter for each 100 m altitude, with an average decrease of 0.5°C Cepel (1966). The mean temperature change of 0.5°C was taken into consideration to determine the temperature for each sample plots in study area. Furthermore, we calculated average annual precipitation for each elevation-climate zone using Shreiber formula as suggested by Cepel (1988): Yh ¼ Yo 54 h
Yh
ð1Þ
Annual mean precipitation value calculated for the area without meteorological station
123
Yo h
54
Annual mean precipitation value (mm) in meteorological station Difference (hectometer) between the altitude of meteorological station and the mean altitude of area whose precipitation value is to be calculated Coefficient
The research area was classified into six different elevation-climate zones. The mean altitude of each elevationclimate zone was calculated. The precipitation and temperature data were adopted for each elevation-climate zone and climate analyses conducted. Soil water deficit is based on the following algorithm where potential evapotranspiration (PET), actual evapotranspiration (AET), precipitation (P) and soil water capacity (SWC). Monthly potential evapotranspiration (PETm) is calculated using Thornthwaite. The climate analysis indicates soil water deficit for each elevation-climate zones in summer months. In this study, since Pm \ PETm then soil water reserve is used and the amount of water collected is a function of the water deficit accumulated over the previous months. Thus, AETm = Pm ? PSWCm, where SWC is the portion of the soil water capacity that is collected. When the period of water deficit is finished, the extra-water not transpired by the plant is used first to reconstitute the soil water reserve and then is flown out of the system. Therefore, the annual soil water deficit is computed as follows Berges et al. (2005): Rm ¼ 1 12ðAETm PETm Þ=PETm
ð2Þ
According to above formula, the SMR method is preferred due to water deficit in each elevation climate zone. Therefore, SMR is used to classify the forest sites in study area. For example, the water deficit of fourth elevation-climate zone was given in Fig. 2 and Thornthwaite analysis results were given in Table 1. Later, the following equation was used to calculate drought indices of each elevation climate zone in study area. Im ¼ 12 AET=Tom
ð3Þ
where Im drought indices, AET actual evapotranspiration (mm), Tom: total monthly max. temperature (°C), and 12 denotes annual coefficient. With Eq. 3, monthly indices are calculated and then annual indices are determined by averaging of monthly indices values. According to these index values, forest sites are determined (Table 2). In direct method, the available water holding capacity of the soils is mainly taken into account as it is a very important parameter in forest sites; but by itself is not adequate for mapping and distinguishing sites. Other abiotic factors have to be taken into account (altitude, exposure, slope, topography). Nevertheless, main rock, soil type, soil
Eur J Forest Res (2008) 127:395–406
399 Table 2 Forest site classification and corresponding site index values
Fig. 2 The water balance of fourth elevation climate zone over a year
skeleton with physiological soil depth as the most effective parameters for forest site classification of soils, are taken into consideration. The plots with similar properties were grouped together to develop ecological soil series for which average water holding capacities were calculated. Ecological soil series is general classification of the whole study area using main rock, soil type, soil skeleton and physiological soil depth of sample plots. They are visually identifiable areas that have similar soil and productivity conditions because of similar climatic and geologic processes (Fig. 3). The characteristics of the ecological soil series are provided in Table 3. The water holding capacity of a soil profile was then estimated as each horizon in soil profile was determined using quantity of thin soil and horizon length. Later, the water holding capacity of a soil profile was estimated using water holding capacity of each horizon by taking into consideration physiological soil
_ Index value
Forest site classification
Site index
\8
Very dry (VD)
IV–V
8–15
Dry (D)
15–23
Moderate fresh (MF)
23–40
Fresh (F)
40–55
Humid (H)
[55
Hygric (Hy)
III I–II
depth in 100 cm in a soil profile Kantarci (1980). The quantity of water holding capacity of each soil profile was calculated. Climate data in each elevation climate zones, mean water holding capacity of each an ecological soil series and topographic features were added to that classification to finalize the forest site classification and mapping. The drought index values calculated for each elevation climate zones are between 8 and 15, indicating the dry base site. Since this classification does not take aspect, slope and altitude (the spatial factors) into consideration, the drought index values, thus the base site, calculated according to ridge flat were adjusted by the spatial factors. Therefore, the adjustment method developed by Kantarci (2000) for various base sites was used to determine forest sites across the forest landscape accordingly (Fig. 4). For example, if the sample point is at the top of the hillside on a southfacing aspect, then the site is dry forest site according to the ridge-flat base site scheme in Fig. 4. The scheme in Fig. 4 has been developed based on ecological understanding of forest sites. Basically, the north and south aspect groups differ from each other in terms of water economics. Since the areas in south aspect group and exposure get more direct and longer solar radiation in northern hemisphere, the areas are drier than those
Table 1 Some results of Thornthwaite analysis for fourth elevation climate zone Balance components
Months I
Heat (°C) Heat indices PET
Annual II
III
IV
V
VI
VII
VIII
IX
X
XI
XII
-5.3
-4.2
-1.1
6.1
10.3
13.7
16.6
16.5
13.1
7.8
1.6
-2.9
6.0
0.0
0.0
0.0
1.4
3.0
4.6
6.2
6.1
4.3
2.0
0.2
0.0
27.6
0.0
0.0
0.0
37.5
68.7
90.8
109.5
102.2
72.0
40.9
7.9
0.0
529.4
Precipitation (mm) Store change
58.0 –
55.8 –
64.9 –
82.6 –
91.9 –
72.4 -18.4
37.1 -35.0
38.0 –
45.6 –
68.1 27.3
64.3 53.4
68.6 53.4
747.6
SWC
53.4
53.4
53.4
53.4
53.4
35.0
–
–
–
27.3
53.4
53.4
AET
–
–
–
37.5
68.7
90.8
72.1
38.0
45.6
40.9
7.9
–
401.5
Soil water deficit
–
–
–
–
–
–
37.4
64.1
26.3
–
–
–
127.9
More water
58.0
55.8
64.9
45.1
23.2
–
–
–
–
–
30.3
68.6
346.0
Surface flow
63.3
56.9
60.4
55.0
34.2
11.6
–
–
–
–
15.1
49.5
346.0
47.4
51.7
35.3
18.3
25.2
Drought indices (In = 12 9 AET/Tom)
14.8
PET potential evapotranspiration, SWC soil water capacity, AET actual evapotranspiration
123
400
Eur J Forest Res (2008) 127:395–406
fresh forest site, because soil is full of precipitation, water level at field capacity is held and any spare water has flown after rains. In fresh forest site, available water in soil is in adequate amount over an important part of growing period. Indirect method
Fig. 3 The map depicting the ecological soil units
with similar ecological attributes Cepel (1984). This fact can be clearly seen in steep, very steep and rugged terrain (Fig. 4). For example, while an area, which is located in the north aspect group, top hillside and steep, very steep and rugged location, is classified as dry forest site, the same area located in south aspect group would be classified as very dry site. All the sample points are evaluated likewise and recorded in a database to delineate the forest sites across the research area using GIS. The study resulted in four different forest sites; very dry, dry, moderate fresh and fresh (Fig. 5b). The very dry and dry sites are located in very steep slope and on south aspect hillside. Water is at a low level, not enough to meet the need of forest trees over the growing period. However, soils can store a part of seasonal rain due to deepness of the soil and less stony feature to provide the need of plants. In moderate fresh site, water deficit can occur or reach a level to cause droughtness in low rainy years. Even though this site is at a margin of doughtiness in water holding capacity, it is at steep slope on north aspect hillside. Because of local spatial factors, the site does not take water leakage from neighboring site. In
123
To quantify site productivity for different forest species, foresters usually evaluate the relationship between tree height and age. One way of doing this is by calculating the SI, defined as the dominant height of a population of trees at specific base or reference age. Although forest site productivity may not be fully represented by SI, this is the most widely used method for estimating the site productivity of even-aged forests (Dieguez-Aranda et al. 2006). In this study, forest site quality was estimated by the indirect method. It is commonly assumed that the influence of genetic and silvicultural factors on forest height growth remains low in single-species, even-aged, closed plantations (Wang, 1998). Trees in stands of a particular species or group of species, and with a full canopy of normally developed crowns were measured for age and height. In each sample plot, height and age were measured in freegrowing dominant and co-dominant trees (100 dominant and co-dominant highest trees per hectare, for example, four highest trees in a 0.04 ha plot) which are the strongest competitors for light, moisture, nourishment and growing space and showed no obvious signs of growth abnormalities or damage (Davis et al. 2001). Trees were bored with an increment borer to determine age at breast height (1.30 m above ground level), and total height was measured with Blume-Leiss Altimeter with 0.1 m precision. The breast height age were converted to total age by adding the number of years of reaching breast height level (1.30 m). SI was calculated for each stand at the reference age of 100 years by use of the model developed by Alemdag (1967) for Pinus sylvestris L. and the model developed by Asan (1985) for Abies nordmanniana (Stev) Spach. According to the indirect method, SI curves were generated by the guide-curve method. Then SI curves are developed by first fitting an average height-over-age guide curve to these data and then constructing a series of higher or lower curves with the same shape as the guide curve. Such a process is called anamorphic curving (Clutter et al. 1983). Guide curves are first used to generate anamorphic SI equations. We used to different anamorphic site index equations for each tree species [Pinus sylvestris L and Abies nordmanniana (Stev) Spach]. There were both pure and mixed stand types in study area. The SI of mixed stand types was determined according to the dominant tree species. Since different SI equations for each tree species exist, SI was determined for each sample. We then used SI values (not dominant height values directly as they are
Very deep-sparsely distributed rocks-granit-sand loam
Very deep-sparsely distributed rocks granit-sand loam
Very deep-densely packed rocks granit-sand loam
Deep- sparsely distributed rocks granit-sand loam
Deep-densely packed rocks-granitsand loam
Medium deep-densely packed rocks-granit-sand loam
Very deep-sparsely distributed rocks-granit-loam sand
Deep-sparsely distributed rocksgranit-loam sand
Deep-densely packed rocks-granitloam sand
Medium deep-sparsely distributed rocks-granit-loam sand
Shallow-densely packed rocksgranit-loam sand
I
II
III
IV
V
VI
VII
VIII
IX
X
XI
WHC water holding capacity
Ecological site units
Number
Pure Pinus sylvestris
Pure Pinus sylvestris
PureAbiesnordmannianasubsp.nordmanniana and pure Pinus sylvestris
Pure Pinus sylvestris
PureAbiesnordmannianasubsp.nordmanniana and pure Pinus sylvestris
Abiesnordmanniana subsp.nordmanniana
Mixed stands of Abiesnordmanniana subsp.nordmanniana and Pinus sylvestris
Mixed stands of Abiesnordmanniana subsp.nordmanniana and Pinus sylvestris
Mixed stands of Abiesnordmanniana subsp.nordmanniana and Pinus sylvestris
Mixed stands of Abiesnordmanniana subsp.nordmanniana and Pinus sylvestris
Abiesnordmanniana subsp.nordmanniana
Tree species
Table 3 Some of the attributes of ecological soil series
7.4
19.0
13.7
21.0
32.3
16.0
19.0
26.1
29.8
36.5
53.4
Mean WCH (mm)
Ah-Cv or Ah-Bv-Cv
Ah-Bv-Cv or Ah-Cv
Ah-Bv-Cv
Ah-Bv-Cv and Ah-Al-Bt-BC-Cv
Ah-AB-B-Cv
Ah-Bv-Cv or Ah-Bv-Cn
Ah-Al-Bt-Cv or Ah-Al-Bt-Cn
Ah-Al-Bt-BC-Cv or Ah-Al-Bt-Cv
Ah-Bv-Cv or Ah-Bv-Cn
Ah-Al-Bt-BC-Cv or Ah-AB-BC
Ah-Al-Bt-BC-Cv
Horizon
0.8–6.3
0.7–4.9
0.8–7.4
0.8–2.6
0.8–6.0
0.9–4.8
0.9–7.9
0.7–3.8
1.3–5.3
0.6–8.9
0.6–3.5
Organic matter (%)
6.1–6.5
4.9–6.1
5.7–6.5
6.0–6.6
5.6–6.6
6.3–6.4
5.7–6.4
6.2–6.6
6.1–6.7
5.6–6.5
5.8–6.3
pH (pure water)
6.0–6.1
4.9–6.1
5.0–6.2
5.0–6.0
4.6–6.3
5.9–6.1
4.9–6.0
5.4–6.2
5.5–6.3
4.7–6.4
5.0–6.1
pH (KCL)
Eur J Forest Res (2008) 127:395–406 401
123
402
Eur J Forest Res (2008) 127:395–406
Fig. 4 Forest site units with steep, very steep and rugged terrain
±
a
±
b Forest Sites
Forest Sites
Very Dry
Good
Fresh
2.000 Meters
Moderate
Moderate Fresh
Fresh
0
Poor
Dry
Moderate Fresh
1.000
Site Index
Very Dry
Dry
2.000
±
c
1.000
0
Treeless areas
2.000
1.000
Meters
0 Meters
Fig. 5 Comparison of forest sites determined with a remote sensing, b direct method and c indirect method
different for different species) to determine the site class value and did not mix SI. This procedure was used to produce SI curves. The site productivity was classified as poor = IV and V; medium = III and good = I and II. The SI map was created with GIS functions (Fig. 5c). The SI value calculated for each sample plot was related to the location of the plot and the borders of the sites were then delineated by application of the spatial analysis functions (proximity, nearest neighborhood) of the GIS. The geographically adjacent (contiguous) sample plots with the same SI values were then combined by use of the reclassification functions (eliminate) of the GIS, to produce a map of forest SI values throughout the study area. Remote sensing method In this study, Landsat 7 ETM image with seven wavelength bands and 30 m spatial resolution were used. Subsets of satellite image was geometrically rectified using 1:25,000 scale topographical maps with UTM projection (ED 50 datum) using first order nearest neighborhood rules. A total of 24 ground points were used to register the ETM image subset with the rectification error of less than one pixel. The TM image were registered to the already registered ETM image through image-to-image registration technique
123
with rectification errors of less than 0.5 pixel with Erdas Imagine 8.6TM software. In this study, we used 5, 4, 3 spectral bands for Landsat 7 ETM image. Ground reference data were gathered from more than 120 points as learning signatures or ground truths for satellite image. The training points were equally distributed to each forest site type with at least ten points per forest site type. For the supervised classification of Landsat 7 ETM image, the forest site map was used to create ground signatures. In order to classify forest site types from the image, signatures were taken from these ground corrected forest site type map. These signatures were further controlled with image enhancement techniques such as transformed vegetation index and principle components analysis. These signatures were then used in a supervised maximum likelihood classification algorithm. Our classification was composed of four classes: very dry forest site, dry forest site, moderate fresh forest site and fresh forest site. We used equal number of control points with at least 30 points for each class Erdas (2002). The accuracy assessment of the image is checked and accepted if the accuracy was higher than 80%. After accuracy assessment, the image was clumped and eliminated 2 9 1 pixels and vectorized into polygon coverage. This polygon coverage was then preprocessed to eliminate areas less than 0.3 ha for spatial
Eur J Forest Res (2008) 127:395–406
403
landscape analysis. Landsat 7 ETM image (2000) was classified into four forest site classes successfully. This is generally acceptable as the overall classification accuracy is much higher (88%) with the Kappa statistics (Conditional Kappa for each Category) value of 0.843 (Table 4). Very dry forest site (93.55% of producer’s accuracy and kappa value 0.9549), dry forest site (92.00% of producer’s Table 4 Confusion matrix for the Landsat ETM (2000) image with supervised classification Classes
Reference data Very dry
Dry
Moderate fresh
Fresh
Total User acc.
Very dry
29
1
0
0
30
96.67
Dry
1
23
5
1
30
76.67
Moderate fresh
1
0
28
1
30
93.33
Fresh
0
2
2
26
30
86.66
Total
31
26
35
28
120
Prod acc. Kappa
93.55 0.9549
92.00 77.78 0.7046 0.9044
92.59 0.8216
106
Overall classification accuracy is 88% and kappa statistics value is 0.84
accuracy and kappa value 0.7046), moderate fresh forest site (77.78% of producer’s accuracy and kappa value 0.9044) and fresh forest site (92.59% of producer’s accuracy and kappa value 0.8216). A map depicting the classification is shown in Figs. 5a. A flow chart in which the methods are compared is shown in Fig. 6.
Results and discussion In this study, we classified forest sites according to direct, indirect and RS methods and compared them to each other. The forest sites were identified with landscape attributes such as altitude, aspect, slope, hillside; climate attributes such as length of vegetation period, temperature and precipitations; soil attributes such as soil depth, soil skeleton and water holding capacity of soil that all were found to be effective factors to distinguish forest sites. The direct method identified forest sites as very dry, dry, moderate fresh and fresh; indirect method as a good, medium and poor sites and RS method as very dry, dry, moderate fresh and fresh sites. Some important differences between the methods were identified. The forest sites determined by indirect method indicate that area for SI I
Sample Plots of Study area
A. Direct Method
1. Topographic factor
2. Soil samples
Digital Elevation Model (DEM)
B. Indirect Method
3. Climatic factor
Temperature, Rainfall
Laboratory Analyses
Anamorphic Method Site Index
Sand, Silt, Clay Field capacity Wilting point Water holding capacity Physiological soil depth Main-rock Amount of tiny soil Amount of soil skeleton
Ecological Soil Units Map
Supervised Classification of Landsat Image (ETM-2000) with Erdas Image
Climatic analyze Thornthwaite Method
Soil type Main rock Soil Skeletion Physiological soil depth Water holding capacity
+
Image Enhancement Techniques
Accuracy Assessment of Supervised Classification
Aspect&Slope
Landform (Ridge,Hillside
Classical inventory (Stand age, Tree height)
C. Remote Sensing Method
Vectorizing of Classification data
+
Soil Moisture Regime Method
Forest Site Classification map with GIS
Site index map with GIS Comparison of Site Index and Forest Site Classification Maps with GIS functions
Forest Site Classification Map with Remote Sensing
Fig. 6 The flow chart of forest site classification process
123
404
Eur J Forest Res (2008) 127:395–406
site and 50.4 (15%) ha area of fresh forest site were not determined by indirect method but RS method. While 30.5 ha (28%) of 107.7 ha poor site was determined by indirect method, 51.7 ha (48%) was in very dry site by remote sensing method. Similarly, 37.8 ha (15%) of 246.3 ha moderate site was determined by indirect method, 137.4 ha (56%) of it was in fresh site by RS method. Similar results were observed in other forest sites (Table 5). The results show that most of the areas can be evaluated and their productivity can be determined with RS method whose ground control points or signatures are based on the sites determined by direct method. To a certain extent, RS method reflects forest sites better than indirect method does. However, due to similar reflectance values of different forest sites on satellite image, erroneous classifications have occurred when compared to direct method. The low representation of sites by RS can greatly be attributed to low radiometric and spatial resolution of satellite data. The relationships between forest sites determined by direct method and SI were statistically analyzed with analysis of variance. No significant (F-ratio = 0.720; P-value = 0.543) relationship was observed between the SI and the four forest site groups (very dry, dry, moderate fresh and fresh forest site) and RS method (very dry, dry, moderate fresh and fresh forest site) as depicted in Fig. 7. The statistical analyses showed clearly that SI is not an adequate measure of forest site productivity as also indicated by Pojar et al. (1987), Sims and Uhling (1992), Pothier et al. (1995) and Johansson (1999). However, the study carried out (Gu¨nlu¨ et al. 2008) in unmanaged forest, the relationship between direct and indirect methods was statistically analyzed using chisquare test. The test indicated a statistically significant relationships between forest sites determined by direct method and Quicbird satellite image (v2 = 36.794; df = 16; p = 0.002), but no significant relationships with Landsat 7 ETM satellite image (v2 = 22.291; df = 16; p = 0.134). Moderate association was found between
and II is 298.5 ha, III 254.3 ha and for sites IV and V 347.7 ha, and other areas 242.8 ha. However, forest sites determined by direct method relate to very dry site of 107.7 ha, dry site of 484.6 ha, moderate fresh site of 304.7 ha and fresh site of 246.3 ha. Forest sites determined by RS method are very dry site of 291.2 ha, dry site of 239.2 ha, moderate fresh site of 287.4 and fresh site of 325.5 ha (Table 5). Non-forest areas such as treeless and degraded areas (242.8 ha) were not determined by indirect method. However, most of those areas can be evaluated and their productivity can be determined with direct method and RS method as opposed to the indirect method that was unable to determine the site class for those open areas. Table 5 shows that the study area was classified into good, moderate and poor sites with indirect method. The results of direct method were distributed over the study area for compassion to other methods using GIS functions. First of all, the comparison shows that direct method classified 592.3 ha (107.7 ha ? 484.6 ha) of area as very dry–dry site, 304.7 ha moderate fresh site and 246.3 ha fresh site; indirect method classified 347.7 ha poor site, 254.3 ha moderate site, 298.5 ha as good site and 242.8 ha areas other areas; RS method classified 530.4 ha (291.2 ha ? 239.2 ha) of area as very dry-dry site, 287.4 ha moderate fresh site and 325.5 ha fresh site. Obviously, the indirect method failed to identify the site productivity of 242.8 ha open areas. Second, of the 347.7 ha poor site determined by indirect method, nearly 199.0 ha (57%) was in fact in freshmoderate fresh site which shows a significant misrepresentation of sites. Similarly, of the 254.3 ha moderate sites, nearly 125.9 ha (50%) was in fact in very dry–dry site that indicates erroneous interpretation of sites. Furthermore, of 298.5 ha good sites, nearly 170.1 ha (57%) was in fact in very dry–dry site according to direct method, that also shows a significant misrepresentation of sites. Comparison of indirect and RS methods reveals some important differences between them. For example, 60.9 (56%) ha area of very dry forest site, 86.0 (18%) ha area of dry forest site, 45.5 (15%) ha area of moderate fresh forest
Table 5 Distribution of sites determined by three methods of direct, indirect and remote sensing Direct method
Total (ha)
Forest site
Remote sensing method
Indirect method
Forest Site
Site index (SI)
VD
D
MF
F
I
II
III
IV
V
ND
VD
107.7
21.5
30.2
29.1
26.8
–
3.9
12.2
10.8
19.7
D
484.6
145.4
87.2
87.2
164.8
19.3
146.9
113.7
62.5
55.7
86.0
MF F
304.7 246.3
70.1 54.2
67.0 54.8
73.1 98.0
94.5 39.4
21.3 38.6
44.1 24.4
90.6 37.8
56.5 48.5
47.2 46.8
45.5 50.4
1143.3
291.2
239.2
287.4
325.5
79.2
219.3
254.3
178.3
169.4
242.8
Total (ha)
60.9
VD very dry, D dry, MF moderate fresh, F fresh, SI I and II good site, SI III moderate site and SI IV and V poor site, ND not determined
123
Eur J Forest Res (2008) 127:395–406
405
when SI values are used. Furthermore, the growth of tree species depends on the characteristics of the growing sites (ground formation, climate, main rock, soil) not easily picked by SI either. Since the direct method is based on edafic, climatic, physiographic and abiotic factors, it becomes a reference method in determining forest sites. Although the direct method appears to be an appropriate method for classifying forest sites, particularly in degraded and open areas, it is obviously very time-consuming, difficult and expensive to carry out in larger areas. In this case, RS method with appropriate spatial, temporal, spectral and radiometric resolution based images may be a good option in accordance with field survey data.
Fig. 7 The statically relationship between site index and forest sites and remote sensing classification
indirect method and direct method (v2 = 16.724; df = 8; p = 0.033) In summary, this study clearly indicates that height of dominant trees (SI) is not an adequate measure of site productivity. The trees in a site must be free of damage, suppression and any other factor that could influence its growth. These characteristics ensure that the growth of the site trees reflects the productivity of the site. However, approximately half of forest areas in Turkey are degraded to some extent, which creates a serious problem for accurate determination of forest SI values. As the target trees (dominant and co-dominant) in degraded areas have been cut down either with management plan or irregular disturbances, it is almost impossible to find suitable (or target) trees to determine site with indirect method (Altun et al. 2008). The direct method and RS method based on the results of direct method as ground control points are therefore preferable methods for classifying forest sites, particularly in the degraded and open areas.
Conclusions Forest site classification is a major problem in Turkish forest industry, particularly in preparing forest management plans and silvicultural activities. The results clearly showed that the indirect method was not suitable for estimating productivity in all areas. There are several reasons for this. First, accurate estimations were not obtained in forest sites including treeless areas and afforested areas. Secondly, it is almost impossible to find suitable areas for target trees in degraded areas that have already been cut down either deliberately or damaged during irregular disturbances. Large portion of degraded areas in Turkey creates a serious problem in forest management decisions
Acknowledgments We would like to extend our appreciation and thanks to Ayhan Usta and Murat Yilmaz who assisted greatly in field works.
References Alemdag S (1967) Forest structure, site index and management principals of Scots pine forests in Turkey. Forestry Research Institute, Technical Bulletin Series, Number 20, Ankara Alemdag IS (1991) National site index and height-growth curves for white spruce growing in natural stands in Canada. NFI, Forestry Canada, Chalk River Altun L (1995) The role of site factors in distinguishing forest sites in Trabzon-Mac¸ka Ormanustu forest, PhD thesis, Karadeniz Technical University, The Graduate School of Natural and Applied Sciences, Trabzon Altun L, Baskent EZ, Gunlu A, Kadiogullari AI (2008) Classification and mapping forest sites using geographic information system (GIS): a case study in Artvin Province. Environ Monit Assess 137:149–161 Anonymous (2001) Gumushane meteorology station climate data Asan U (1985) A study on forest site index of fir (Abies nordmanniana Spach.) forests in Artvin. Faculty of Forestry Journal, Istanbul University, Istanbul Bailey RG, Pfister RD, Henderson JA (1978) Nature of land and resource classification—a review. J For 76:650–655 Bakkaloglu M (2003) Classification and mapping of forest sites in Gumushane-Karanlıkdere Forest District. PhD thesis, Karadeniz Technical University, The Graduate School of Natural and Applied Sciences, Trabzon Baskent EZ, Jordan GA, Nurullah AMM (2000) Designing forest landscape (ecosystems) management. For Chron 76(5):739–742 Baskent EZ, Barli O, Ayaz H, Bilgili E, Turna I, Ipek A, Altun L (2003) A new approach at reconfigure of Turkish (I). Forest and Prey, number 6, vol 80, Ankara Berges L, Chevalier R, Dumas Y, Franc A, Gılbert JM (2005) Sessile oak (Quercus petraea Liebl.) site index variations in relation to climate, topography and soil in even-aged high-forest stands in northern France. Ann For Sci 62:391–402 Carmean WH (1975) Forest site quality evaluation in the United States. Adv Agron 27:209–269 Cepel N (1966) Practical base of forest sites introduction and forest site map. Kutulmus¸ press, Istanbul Cepel (1984) Forest ecology. Faculty of Forestry, Istanbul University, Istanbul Cepel N (1988) Forest ecology. Faculty of Forestry, Istanbul University, Istanbul
123
406 Clutter JL, Fortson JC, Pienaar LV, Brister GH, Bailey RL (1983) Timber management: a quantitative approach. Wiley, New York Corona P, Scotti R, Tarchiani N (1998) Relationship between environmental factors and site index in Douglas-fir plantations in central Italy. For Ecol Manage 1001:195–207 Davis LS, Johnson KN, Bettinger PS, Howard TE (2001) Forest management, to sustain ecological, economic and social values, 4th edn Dieguez-Aranda U, Grandas-Arias JA, Alvarez-Gonzalez JG, von Gadow K (2006) Site quality curves for birch stands in NorthWestern Spain. Silva Fenn 40(4):631–644 Erdas (2002) Erdas field guide, 6th edition. Erdas LLC, Atlanta Erinc S (1984) Climatology and methods. Istanbul University, Oceanography and Geography Institute, Istanbul ESRI (1999) Using ArcMap, ISBN-1-879102-69-2. Environmental Systems Research Inc, Redlans Grey DC (1980) On the concept of site in forestry. S Afr For J 113:81–83 Gunlu A (2003) A study about distinguishing and mapping forest site in Artvin-Genya mountain, Turkey. MSc thesis, Karadeniz Technical University, The Graduate School of Natural and Applied Sciences, Trabzon _ Ercanlı I_ (2008) Classifying Gu¨nlu¨ A, Bas¸ kent EZ, Kadıog˘ulları, AI, oriental beech (Fagus orientalis Lipsky.) forest sites using direct, indirect and remote sensing methods: a case study from Turkey. Sens J 8:2541–2550 Johansson T (1999) Site index curves for common Alder and Grey Alder growing on different types of forest soil in Sweden. Scand J For Res 14:441–453
123
Eur J Forest Res (2008) 127:395–406 Kantarcı MD (1980) Investigating forest soil types and forest site classification mapping in Belgrad forests. Faculty of Forestry Journal, Istanbul University, Istanbul Kantarcı MD (2000) Soil science, Faculty of Forestry. Istanbul University, Istanbul Kayahara GJ, Klinka K, Marshall PL (1998) Testing site index-sitefactor relationship for predicting Pinus contorta and Picea engelmannii? Picea glauca productivity in central British Columbia, Canada. For Ecol Manage 110:141–150 McKenney DW, Pedlar JH (2003) Spatial models of site index based on climate and soil properties for two boreal tree species in Ontorio. For Ecol Manage 175:497–507 Pojar J, Klinka K, Meidinger DV (1987) Biogeoclimatic ecosystem classification in British Columbia. For Ecol Manage 22:119–154 Pothier D, Doucet R, Boiley J (1995) The effect of advance regeneration height on future yield of black spruce stands. Can J For Res 25:536–544 Seckin B, Kahveci O (1993) The develop problems and solution proposals of silvicultural applications in Turkey forestry, vol. III. First Forestry Council, Ankara, pp 296–304 Sims RA, Uhling P (1992) The current status of forest site classification in Ontario. For Chron 68:64–77 Sturtevant BR, Seagle SW (2004) Comparing estimates of forest site quality in old second growth oak forest. For Ecol Manage 191:311–328 Wang GG (1998) Is height of dominant trees at a reference diameter an adequate measure of site quality? For Ecol Manage 80:95–105 Wang GG, Klinka K (1996) Use of synoptic variables in predicting white spruce site index. For Ecol Manage 80:95–10