The Impact of Weather and Slope Conditions on the Productivity, Cost, and GHG Emissions of a Ground-Based Harvesting Operation in Mountain Hardwoods

Abstract

Mountainous hardwood mixed stands offer challenges to timber harvesting operations in practice, including a harsh climate, variable topography, steep terrain, and large-sized timbers. This paper aims to develop productivity and cost models for a mountain-ground-based harvesting operation across the terrain (e.g., slope conditions), stand (e.g., tree volume) environmental (e.g., weather), and yard (e.g., winching distance) variables and to assess GHG emissions related to the equipment in use. This development was implemented in a timber harvesting practice under single-tree selection in mountainous forests of Iran where a motor-manual chainsaw is used for felling and a rubber-tired cable skidder is used for log extraction. The average delay-free productivity was 4.55 m³ for felling and 14.73 m³ h⁻¹ for skidding. Lower production costs and higher productivity rates were observed over the gentle slopes and in sunny conditions. The average production costs ranged between USD 4.27 m⁻³ for felling and USD 5.35 m⁻³ for skidding. The average emissions ranged between 0.96 kg m⁻³ for felling and 7.06 kg m⁻³ for skidding in snowy conditions over steep slopes. The study’s results confirm avoiding harvesting operations on steep slopes (greater than 35%) and in extreme weather conditions to obtain higher work efficiency and to minimize adverse effects of machinery on forest ecosystems. The results should be of use to harvest managers and forest planners considering the application of ground-based harvesting operations using a semi-mechanized system on a range of operating conditions in mountain hardwood stands.

1. Introduction

Timber harvesting in mountainous regions can be challenging and expensive due to the diverse and difficult nature of the terrain and/or excessive ground obstacles, extreme climate conditions, and large-sized hardwood timbers. Various factors and constraints contribute towards a successful harvest operation. These constraints often discourage deployment of the fully mechanized system (e.g., a combination of harvesters and forwarders), and harvesting contractors are forced to use the purpose-built steep terrain equipment to improve the productivity and reduce the harvesting costs [1] while minimizing the adverse environmental impact. Mechanization of forest operations in these regions often encourages the use of semi-mechanized equipment such as a combination of motor-manual felling and ground-based skidding vehicles including crawler skidders or rubber-tired cable skidders [2]. This is mainly due to the advantages of the productivity, low purchase cost, ease of service, and moderate skill requirements [3]. Currently, ground-based harvesting operations using semi-mechanized equipment represent 55% of today’s harvesting practices around the world [4], while the remaining 45% are carried out using mechanized systems. Different harvesting methods can be applied depending on a number of factors. It is, therefore, important to analyze and improve harvesting methodologies in the context of sustainable forest management [5].

Typically, the productivity of harvesting operations can be influenced not only by the choice of equipment (machine type, operator skill, functions), but also by many other variables that are also important. Among these variables are terrain (i.e., the slope gradient), stand (e.g., stump diameter), environmental (e.g., weather), and yard (e.g., winching distance) variables [5,6,7,8]. Indeed, the cost of a ground-based harvesting operation (felling and off-road transportation) constitutes more than 40% of the total wood supply chain cost [9,10]. Therefore, any savings in this practice as a result of accurate information will have a significant impact on the success of the operation. Taking all this into account, it is necessary for harvesting managers and/or contractors to assess the available equipment considering several factors, including terrain, stand, weather, and practice/logistics, to meet the management outcomes while improving the ergonomic and safety standards. This is even more important in mountain regions where the topography is often diverse, the climate is harsh, the work itself is physically demanding, and the machineries are heavy, expensive to purchase, and dangerous if not used properly. In addition, harvesting managers and/or contractors must be aware of the negative environmental impacts, such as greenhouse gas (GHG) emissions, of equipment in use during mechanical operations [11].

Since 1963, empirical-based time study models have frequently been used to analyze the performance of harvesting operations, such as modeling of productivity and costs, regarding stand, topography and yard, and/or logistic attributes and continue to be fundamental paradigms in the analysis of forest harvesting practices around the world [12]. However, a majority of the studies analyzed softwood stands on flat to medium terrain conditions using fully mechanized equipment, while few have been done in mountainous hardwood stands. The current study presents one of the first scientific attempts to evaluate the work productivity and costs of a ground-based skidding operation (felling and log extraction) using semi-mechanized equipment in mountainous hardwood stands, with the goal of developing predictive models of machine efficiency and GHG emissions in relation to several variables of different sources.

Overall, numerous studies have addressed the ground-based skidding system, mainly focused either on felling or log extraction, worldwide, which shows the importance of the subject (for example, see [10,13,14,15,16,17,18,19]). A large body of literature on this subject emphasizes the physical characteristics of harvesting operations (e.g., load volume per cycle, skidding distance, and tree diameter) as major factors influencing the performance of the ground-based operation [17,20]. For example, Wang et al. (2005) reported an 8% reduction in the unit skidding cost when the diameter of skidded logs increased over 26 cm [14]. However, some authors have pointed out that stand attributes and yard/logistical variables are the major factors influencing the productivity and harvesting cost of harvesting practices [15,21]. In a more recent study, Proto et al. (2018) developed statistical modeling approaches to evaluate a purposed-based skidding operation in mountain-ground-based extraction in southern Italy. They reported an increased productivity rate (1.5 times greater than that of the traditional skidding method) [20]. In addition to the stand, geophysical, and logistical factors, a very few recent studies have considered the significance of variable weather conditions on the performance of a ground-based harvesting operation. Grzywiński et al. (2020) found that the winter felling practice took about 13% longer compared with the summer felling practice [18].

In summary, the previous literature mainly analyzed the work productivity and harvesting cost of ground-based harvesting operations considering three major themes: (i) geophysical (e.g., winching distance, winching slope), (ii) stand (e.g., tree diameter, distance between trees), and (iii) yard and/or logistics (e.g., level of mechanization, expertise of equipment operator) factors. However, few studies have considered the influence of weather conditions and terrain slopes, as main components of forest practices in steep mountainous regions, on the work productivity, costs, and GHG emissions of the equipment used in ground-based harvesting operations. Furthermore, the major focus of the literature on this subject has been softwood stands on flat to medium conditions considering the core revenue stream for forest companies when compared with hardwoods on steep terrains. As highlighted by recent studies, climate change scenarios predict a higher impact of global warming on hardwood species than softwood ones [22]. Therefore, the importance of hardwood timber will probably increase in the coming decades and there is a strong need to improve our scientific knowledge of these forests regarding all three pillars of sustainability (economy, environment, and society) in order to set up sustainable forest operations [23,24,25]. Taking into account that written above, the main objectives of this study were (i) to analyze how weather climate, topographic conditions, stand, and yard/logistical variables can affect the productivity and cost of the equipment used in ground-based harvesting operations (felling and skidding) in hardwood mixed stands, and (ii) to develop generic prediction models for estimating the productivity, cost, and GHG emissions of equipment used in each of the operations. The relationships between response variables (productivity and cost) and independent variables (slope, weather, stand, and yard) were analyzed in this study to deepen our knowledge of current harvesting practices in mountainous hardwood and mixed stands. Furthermore, the calculation of GHG emissions during harvesting operations (felling and skidding) and the analysis of the effect of weather and slope conditions on the performance of harvesting equipment used in the ground-based harvesting operations have not been studied previously. The results should be useful to forest managers, planners, and logging contractors considering the use of ground-based harvesting operations regarding such important variables. The model should also be useful to forest engineering practitioners specialized in modeling the productivity, cost, and emissions of equipment used in mountain harvesting practices with hardwood and mixed stands.

2. Materials and Methods

2.1. Study Area

The studied area was located in watershed no. 45 in the central part of the Hyrcanian hardwood stands in northern Iran (between 51°32′ E and 51°33′ E longitude and 36°27′ N and 36°40′ latitude). Three commercial harvesting sites were selected for the design of the experiments (No. 208, 209, and 221). The forest was mostly covered by natural uneven-aged forests consisting of mixed deciduous stands, including beech (Fagus orientalis Lipsky), hornbeam (Carpinus betulus L.), maple species (Acer velotinum Boiss., Acer cappadocicum Gled.), lime tree (Tilia begonifolia Stev.), Caucasian alder (Alnus subcordata C.A.Mey), Wych elm (Ulmus glabra Huds.), and chestnut leaved oak (Quercus castaneifolia C.A.Mey) that grow on semi-calcareous brown soils with a silt and silt–loamy texture. The terrain varied with slope gradients ranging from 20 to 60 percent. The altitude ranged from 800 m to 1610 m above sea level. Precipitation recorded at the nearest national weather stations located 20 km away from the research site ranged from 1420 mm to 1530 mm, with its peak in the spring and autumn. The forest was managed under a mixed, uneven-aged system with a single selective or group-cutting regime. The average tree density and standing timber volume before operations were 280 stems per ha and 401 m³ ha⁻¹, respectively. The average diameter of marked trees for harvesting was measured at 63.20 cm, 65.60 cm, and 60.50 cm in the low, medium, and high slope classes, respectively, ranging from 25 cm to 120 cm.

Motor-manual felling with a chainsaw was applied. This is currently the most widely applied method for felling and processing trees (delimbing, topping, and bucking) in the Iranian mountain forests, and will most likely remain the dominant method from an operational perspective due to the large trees in mixed hardwood stands. Harvesting contractors thereby used a rubber-tired cable skidder to collect harvested timbers from all over the harvesting sites. Felling of the marked trees was carried out during October and November 2016. Skidding operations were carried out between February and March of the following year. The felling team included a logger, a crew manager, and an assistant who were responsible for safety, production, and the quality felling of selected trees. The skidding crew consisted of three people including a choker man, a chainsaw operator, and a skidder operator. The crew had more than 15 years of experience with those machines. Felling was carried out by a Sthil MS880 chainsaw. Skidding operations were performed with a rubber-tired cable skidder (Timberjack 450C model).

Table 1. General characteristics of the equipment used at the studied sites.

Technical Characteristics
Timberjack 450C		Stihl MS 880
Overall height/width (mm)	3023/3175	Displacement (cm3)	121.60
Power (kW)	120.00	Power (kW)	6.40
Overall weight (kg)	10,270	Weight (kg)	10.00
Front axle weight (kg)	5682	Bar length (cm)	90.00
Rear axle weight (kg)	4588	Oil tank volume (L)	0.70
Displacement (cm3)	6800	Fuel tank volume (L)	1.30
Number of cylinders	6.00	Number of cylinders	1.00

presents the main technical characteristics of the chainsaw and the rubber-tired cable skidder used in the field experiment.

2.2. Experimental Design and Data Collection

The experiments studied the effect of slope and weather conditions on the performance indicators (i.e., cost, productivity, and GHG emissions) of chainsaw felling and ground-based skidding operations under ground-based harvesting operations. Because of variability in the performance of the harvesting crew and machines, three classes of slope gradients were considered: gentle slopes (G), in which the slope gradient of the skidding route was between 0 and 20%; moderate slopes (M), in which the gradient was between 21% and 30%; and steep slopes (S), in which the gradient exceeded 31% (up to 35% at short skidding distances). Weather conditions varied during the experiments, described as sunny, rainy, and snowy. Climatic conditions are one of the important factors affecting the machine productivity and operational cost in mountain forest harvesting. According to Iranian forest guidelines (FGI), the felling period usually ranges between December and February each year and harvesting operations are stopped in the case of extreme conditions such as strong winds, heavy rain, thick fog, and snowfall that covers more than 20 cm of ground. In this study, we explored the productivity rates of harvesting operations (felling and extraction) under three different weather conditions. Under sunny conditions, in which the weather is sunny and there was no rain in the previous week, the ground was not slippery and the temperature ranged between 10 and 20 °C. Under rainy conditions, in which the weather was rainy with continuous rainfall during harvesting, the ground was slippery and the temperature ranged between 10 and 15 °C. Under snowy conditions, the weather was snowy, the coverage of snow on the ground was between 10 and 20 cm, the ground was slippery, and the temperature was below 10 °C.

A continuous time study method using a handheld stopwatch technique was applied throughout the experiment with elemental times read and recorded at each element’s breaking point to measure the time lag of the start or end time for each element [26]. This provided the authors with more in-depth information on how felling and skidding operations were compiled for each work cycle. These observations were also used to establish relationships between cycle times (the time required to complete a specific task such as felling or skidding) and independent variables. Although a detailed time study was designed to measure effective and/or net productivity rates (excluding unproductive times), delay times were recorded to estimate the gross productivity [27]. Elemental time cycles for felling were Walk to Tree (WT), Prepare the Workplace (PWP), Choose felling Direction and anticipate the Escape route (CDE), Under-Cut (UC) and/or sink cut, Back-Cut (BC), Delimb the top of felled Trees (DFT), Cross-cutting (log processing) (LPr), and Delays (DT). The skidding elemental functions were Travel Unloaded (TUL), Cable Releasing (CR), Set Chockers (SC), Log Winching (LW), Travel Loaded (TL), Log Unhooked (LU), Log Piling (LPi), and Delays (DT).

A total of 214 felling cycles and 270 skidding cycles from three adjacent sites were recorded. The average tree density, DBH, and stand volume of stands before harvesting operations were 331 stem ha⁻¹, 25.90 cm, and 263.70 m³ ha⁻¹, respectively.

Felled trees were processed and bucked (at 7 cm from the top) into mill-specific lengths (4 or 6 m), such as logs, sawn timber, and pulpwood (processing phase), immediately at the stump. A rubber-tired cable skidder then extracted logs to the roadside landing and consolidated them into larger loads for further transportation by trucks. Skidding routes were predominantly uphill and designed before the operation. In this study, felling parameters included distance between trees, weather condition, slope gradient, and stump diameter, while the travel distance, payload per turn, stump diameter, tree length, winching distance, winching slope, and the slope of skid trails were measured for skidding operations. These independent parameters were used to assess and develop productivity and cost models. Before felling and skidding operations, the location of landings and the main bunching–extraction routes were marked. Felling directions were delineated on the tree stems so that the operator felled the marked trees either towards skidding routes or preferably at an oblique angle of 25–30° to the skidding direction. The DBH of marked trees was measured using a handheld caliper and the length of trees and travel distances were recorded with a laser distance meter. The information on the number of logs per loaded cycle was collected by visual observation during log winching at a safe distance. The volume of logs was determined according to Smalian’s formula by measuring the two log ends and stem length. Operational costs (chainsaw and skidder), as measured in USD per scheduled machine hour (SMH⁻¹), were estimated using forest practices codes in mountainous regions [28]. Skidding and felling costs were divided by productivity rates to determine the production cost per unit (USD m⁻³). The operational cost of each machine was calculated considering fixed costs (i.e., the sum of interest, depreciation, and insurance) and variable costs (i.e., the sum of oil, fuel, services, maintenance, repair, chains, and tires). The hourly cost values of the chainsaw and cable skidder are given in

Table 2. The operational hourly costs (USD SMH⁻¹) of the chainsaw and skidder.

Cost Element	Chainsaw (Stihl MS 880)	Skidder (Timberjack 450C)
Scheduled machine hours (SMHs) (h)	320.00	1200.00
Productive machine hours (PMHs) (h)	240.00	900.00
Total fixed cost (USD year−1)	560.00	25,644.00
Total hourly fixed cost (USD h−1)	1.75	21.37
Total variable cost (USD year−1)	516.00	27,468.00
Total hourly variable cost (USD h−1)	2.15	30.52
Total hourly labor cost (USD h−1)	15.32	17.05
System cost (USD h−1)	19.22	68.94

.

The pollutant emissions as the amount of GHG components (CO₂, CO, HC, and NOx) and PM₁₀ were assessed considering exhaust emissions from the consumption of one liter of fuel (GHG_L, g L⁻¹) that occurred during felling and skidding operations and taking into account both the emission factor related to the engine output power [26,29] and the thermal efficiency of the fuel combustion process [26,29,30]. The emissions per hour were calculated for each GHG component and the PM₁₀ produced during combustion of fuel (Efc) and produced during fuel production and logistics (Efp). This calculation was made using Equation (1).

Eh = F_c × E_t × C_v × T_e

(1)

where Eh is the emission for each single component of GHG and PM₁₀ related to fuel consumption per hour (g h⁻¹), F_c represents the fuel consumption per hour (L h⁻¹), E_t refers to the emission factor (g MJ⁻¹) of the engine output (for each single component of GHG and PM₁₀), C_v is the calorific value of the fuel, (MJ L⁻¹), and T_e represents the thermal efficiency. For further details regarding GHG and PM₁₀ assessment please see [30,31].

The fuel consumption was estimated by filling the tank to the maximum level at the roadside before starting the experiment and after completing the task at the same location and position [22]. This procedure was repeated for each time element of both felling and skidding operations. This allowed us to correctly estimate the emission for each single component of GHG and PM₁₀ as reported in Equation (1).

In order to have better information on the pollutant emissions strictly related to the GHGs and to give the possibility for further scientific comparisons, all the GHG components were converted to CO₂ equivalents. The conversion factors of GHG components (CO₂, CO, HC, and NOx) to CO₂ equivalents were used as follows [26]:

CO₂ = 1; CO = 2; HC = 10; NOx = 180.

Finally, the amount of GHG emissions per unit product (GHG_p, g m⁻³) was calculated using Equation (2).

$${\mathrm{GHG}}_p=\frac{{\mathrm{GHG}}_h}{P_n}$$

(2)

where $P_n$ is the net productivity rate (m³ h⁻¹), which is a function of $v$ and refers to the volume of logs (m³), and $t_n$ represents the effective time neglecting delay times (min) as shown in Equation (3).

$$P_n=\frac{60v}{t_n}$$

(3)

2.3. Data Analysis

To compute operation performance indicators, the effective and gross productivity rates were calculated using Equations (4) and (5).

$$P_n=\frac{60v}{t_n}$$

(4)

$$P_g=\frac{60v}{t_n+t_d}$$

(5)

where $P_n$ is the net productivity rate (m³ h⁻¹), $v$ refers to the volume of logs (m³), $t_n$ is the effective time neglecting delay times (min), p_g is the gross productivity rate including delays (m³ h⁻¹), and $t_d$ refers to delay times (min). Responses of relative changes in the productivity, cost, and GHG emissions to the combination of slope gradients and weather conditions were analyzed with a factorial one-way analysis of variance (ANOVA). For determining the statistical significance of main effects, means were separated using Duncan’s tests at α ≤ 0.05 level. Overall time-consumption models were developed by combining associated parameters for each operation, such as felling and skidding. When significant differences were determined, the associations were established by multivariate stepwise regression analysis. Pearson’s coefficient correlations were used to assess the strength of the linear association between the productivity, cost, and time-consumption element as response variables and independent variables, including harvested tree attributes (log volume, diameter, and length) and yard/logistical factors (slope, winching distance, winching slope, and skidding distance).

The Spearman’s coefficient correlation was used to determine the relationship between elemental times of felling and environmental conditions (weather). In order to include different weather conditions in the regression models of effective time and effective productivity, we used two sets of dummy variables, i.e., Z₁ and Z₂, as follows: Z₁ = 0 and Z₂ = 0 for sunny, Z₁ = 1 and Z₂ = 0 for rainy, and Z₁ = 0 and Z₂ = 1 for snowy. SPSS version 19.0 software was used to perform statistical analyses, while Sigmaplot was also used to manipulate graphs and generate charts. Models were evaluated based on statistical indices, including goodness of fit (R²), the p-value, and the F-value, to give a more reliable model of chainsaw felling and skidding operations. A validation test was conducted to check the validity of the predictions using witness samples and confidence intervals of the predictions.

3. Results

A total of 214 work cycles were measured for manual felling with a total extractable volume of 762 m³ by a cable skidder at the studied logging sites.

Table 3. Summary statistics of harvested trees and skidding operations under the range of slopes.

Attributes	Slope Gradient (%)
	Gentle (<20)	Moderate (21–30)	Steep (>31)
Chainsaw felling
No. of felled trees	71.00	70.00	73.00
Average tree DBH (cm)	63.20 ± 22.14 a	65.60 ± 22.70 a	60.50 ± 25.38 a
Average tree height (m)	27.70 ± 5.09 a	28.60 ± 4.69 a	27.10 ± 4.51 a
Average volume (m3 tree−1)	2.85 ± 0.51 ab	3.05 ± 0.48 a	2.57 ± 0.40 b
Total volume (m3)	256.50	274.50	231.30
Total no. of logs	181.00	180.00	177.00
Extraction operation
No. of skidding cycles	88.00	92.00	90.00
Average skidding distance (m)	94.20 ± 20.6 a	86.70 ± 19.4 a	92.50 ± 20.2 a
Average slope of skid trail (%)	15.70 ± 4.46 a	13.90 ± 3.88 a	14.10 ± 4.11 a
Average No. logs per cycle (log cycle−1)	2.06 ± 0.48 a	1.96 ± 0.48 a	1.97 ± 0.40 a
Average volume per cycle (m3 cycle−1)	2.91 ± 0.76 a	2.98 ± 0.70 a	2.57 ± 0.50 b
Total volume (m3)	256.50	274.50	231.30

Note: Different letters in the rows indicate significant differences by Duncan’s test at α = 0.05.

presents descriptive statistics of the harvested trees under different slope gradients. Harvested log volumes and stump diameters were varied, which certainly affected the felling productivity rate and its relevant costs. The maneuverability of the skidder (driving loaded and traveling empty) was affected by slope gradients. Within the slope gradient classes, averaged over a gentle slope, the number of logs per turn, the load volumes per turn, and the total harvested volume were high compared with steep slopes.

Table 4. Elemental time averages and standard deviations for felling and skidding. WT, walk to the tree; PWP, prepare the workplace; CDE, choose the felling direction and prepare the escape route; UC, sink-cut; BC, back-cut; DFT, delimbing and topping; LPr, log processing; TUL, travel unloaded; CR, cable releasing; SC, set chokers; LW, log winching; TL, travel loaded; LU, log unhooking; LPi, log pilling; DT, delay time; TET, total effective time per cycle; TGT, total gross time per cycle.

Time Element	Felling and Processing		Time Element	Extraction
	Mean ± SD (min)	% of TGT		Mean ± SD (min)	% of TGT
WT	5.40 ± 2.10	13.40	TUL	2.10 ± 0.20	13.30
PWP	2.50 ± 0.90	6.20	CR	1.00 ± 0.10	6.30
CDE	1.80 ± 0.70	4.50	SC	1.60 ± 0.10	10.10
UC	1.10 ± 0.70	2.80	LW	1.20 ± 0.10	7.60
BC	0.90 ± 0.50	2.20	TL	4.40 ± 1.00	27.90
DFT	20.30± 4.00	50.50	LU	0.30 ± 0.10	1.90
LPr	4.60 ± 1.80	11.40	LPi	0.90 ± 0.10	5.70
DT	3.60 ± 1.20	9.00	DT	4.30 ± 0.90	27.20
TET	36.60 ± 3.00	-	TET	11.50 ± 1.90	-
TGT	40.20 ± 3.30	100.00	TGT	15.80 ± 1.90	100.00

presents breakdown events for the components of the chainsaw felling and skidding operations. The coincidence of felling and processing allowed the authors to consider these inter-related operations as a unique practice in preparation for the subsequent phase, i.e., skidding operations. The total effective felling time averaged 36.6 min, consisting of: 11.70 min (32%) for cutting (e.g., walking to the tree, preparing, choosing the felling direction, sink-cutting, and back-cutting), 20.30 min (55%) for delimbing and topping, and 4.60 min (13%) for the processing of logs. Except for delimbing, the processing of logs was the most time-consuming phase in felling, followed by walking to the tree. Out of the total time consumption for skidding (11.50 min), travel loaded accounted for a share of 27% (4.40 min), followed by travel empty to the landing with a share of 13% (2.1 min). Approximately 57% of the gross effective time was spent on the movement of the skidder, while the remaining 43% was related to the release of cables, set chokers, winching logs, etc. Unhooking of logs accounted for only 2% of the gross effective skidding time. The average delay times (mainly due to technical delays) ranged from 3.60 to 4.30 min, respectively, for felling and skidding operations. It is worth noting that the proportion of delay was three times higher for skidding compared with felling.

The relative change in net felling productivity (PMH) under weather conditions ranged from −22% in the rainy condition to −25% in the snowy condition compared with the sunny condition (

Table 5. Delay-free productivity (mean ± SD) of felling and skidding operations.

Factor	Chainsaw Felling and Processing (m3 h−1)	Cable Skidder (m3 h−1)
Weather condition
Sunny	5.39 ± 0.50 a	14.73 ± 1.95
Rainy	4.20 ± 0.50 b	-
Snowy	4.05 ± 0.30 b	-
F-Value	57.45 **	-
Slope gradient
Gentle	5.06 ± 0.50 a	18.80 ± 1.80 a
Moderate	4.60 ± 0.40 ab	13.37 ± 1.10 b
Steep	4.04 ± 0.30 b	9.55 ± 1.14 c
F-Value	89.05 **	102.42 **

** significant at α = 0.01. Different letters in the rows indicate significant differences by Duncan’s test at α = 0.05.

). Within the weather conditions, there were no statistically significant differences in felling production rates between rainy and snowy conditions, but these were significantly different compared with sunny conditions. The relative change in effective felling production among the slope gradients ranged from −9% on moderate slopes to −20% on steep slopes. There were no statistically significant differences in the mean values of felling production rates on gentle and moderate slopes but these were statistically significantly different between gentle and steep slopes. A similar trend was observed for skidding operations. The relative change in skidding production rates within slope gradients ranged from −29% on moderate slopes to −49% on steep slopes and was statistically affected by slope gradient classes (p < 0.001).

The results of the correlation analysis show that there were strong correlations between the work elements of felling and the independent variables. There were strong significantly positive correlations between felling elemental times and weather conditions, slope gradients, stump diameters, tree lengths, and log volumes; however, exceptions occurred. Correlations of CDE, UC, and BC with the weather condition and PWP with the slope gradient were positive, but not significantly different (

Table 6. Pearson’s correlation coefficients between elemental times of felling and statistically significant independent variables. For weather conditions, Spearman’s correlation coefficients were developed. WT, walk to the tree; PWP, prepare the workplace; CDE, choose felling direction and prepare the escape route; SC, sink-cutting; BC, back-cutting; DFT, delimbing of felled trees; LPr, cross-cutting (bucking).

Felling Work Phase	Weather Condition	Slope Gradient (%)	Tree DBH (cm)	Tree Length (m)	Tree Volume (m3)
WT	0.573 **	0.675 **	-	-	-
PWP	0.235 *	0.364	0.565 **	0.613 **	0.607 **
CDE	0.101	0.285 *	0.238	0.439 **	0.420 **
UC	0.156	0.379 *	0.631 **	0.627 **	0.609 **
BC	0.122	0.338 *	0.725 **	0.653 **	0.661 **
DFT	0.436 **	0.613 **	0.680 **	0.626 **	0.660 **
LPr	0.493 **	0.685 **	0.703 **	0.659 **	0.653 **

* Coefficient of determination is significant at the 0.05 level; ** Coefficient of determination is significant at the 0.01 level.

).

The felling production cost was significantly affected by slope gradients and weather conditions (

Table 7. Unit production cost (USD) of felling and extraction subject to slope gradients and weather conditions.

Item	Felling and Processing		Skidding
	Tree	m3	Tree	m3
Weather condition
Sunny	10.06 a	3.57 a	-	-
Rainy	12.90 b	4.58 b	-	-
Snowy	13.44 b	4.75 b	-	-
Slope gradient
Gentle	10.74 a	3.80 a	10.35 a	3.67 a
Moderate	11.79 ab	4.18 ab	14.55 b	5.16 b
Steep	13.44 b	4.76 b	20.37 c	7.22 c

Note: Different letters in the rows indicate significant differences by Duncan’s test at α = 0.05.

). The relative change in the unit felling cost under weather conditions varied from +28% in the rainy condition to +33% in the snowy condition compared with the sunny condition. There were no statistically significant differences in the mean values of the felling production cost in rainy and snowy conditions but these were significantly different from the values obtained in the sunny condition. Averaged over steeper slopes, the relative changes in the felling production cost ranged from +10% on moderate slopes to +25% on steep slopes compared with gentle slopes. Within the slope gradients, there were no statistically significant differences in the unit felling cost on gentle and moderate slopes but there were statistically significant differences between gentle and steep slopes. The relative change in skidding production rates within slope gradients ranged from +40% on moderate slopes to +96% on steep slopes and was statistically affected by slope gradients (p < 0.001).

Overall, there were strong significantly positive correlations between skidding work elements and independent variables (

Table 8. Pearson’s correlation coefficients between skidding work elements and independent variables. TUL, travel unloaded; CR, cable releasing; SC, set chokers; LW, log winching; TL, travel loaded; LU, log unhooking; LPi, log piling.

Extraction Work Element	Tree DBH (cm)	Tree Length (m)	Tree Volume (m3)	Slope Gradient (%)	Winching Distance (m)	Winching Slope (%)	Skid Trail Distance (m)	Skid Trail Slope (%)
TUL	-	-	-	-	-	-	0.667 **	0.237 *
CR	-	-	-	0.690 **	0.795 **	0.707 **	-	-
SC	0.503 **	0.306 *	0.410 **	0.311 *	-	-	-	-
LW	0.412 **	0.379 **	0.525 **	0.415 **	0.706 **	0.532 **	-	-
TL	0.706 **	0.700 **	0.739 **	-	-	-	0.783 **	0.554 **
LU	0.291 *	0.359 **	0.385 **	-	-	-	-	-
LPi	0.493 **	0.537 **	0.556 **	-	-	-	-	-

* Coefficients of determination are significant at the 0.05 level; ** Correlation of determination is significant at the 0.01 level.

). The coefficients of determination for the majority of variables were above 50%, which indicates that the model explains 50% of the variation in the time of one work cycle.

The performance models were developed for the chainsaw felling and skidding operations to estimate the overall time consumption and production rates as a function of statistically significant independent variables (

Table 9. Regression models developed for estimating the time and effective productivity for felling and skidding operations as a function of independent variables.

Timber Harvesting Phase	Model Name	Performance Model	R2 (adj) %	SE	F-Value	p-Value
Felling and processing	Effective time (min)	FET = −72.862 + 1.498 (DBH) + 0.403 (GS) + 11.96 (Z1) + 18.626 (Z2)	0.85	16.84	367.76	<0.001
	Effective productivity (n h−1)	FEP = 2.786 + 0.0001 (DBH) − 0.021 (GS) − 0.726 (Z1) − 1.332 (Z2)	0.78	0.31	237.30	<0.001
	Effective productivity (m3 h−1)	FEP = 6.866 − 0.002 (DBH) − 0.047 (GS) − 1.448 (Z1) − 2.374 (Z2)	0.76	0.62	211.89	<0.001
Skidding	Effective time (min)	EET = −4.834 + 0.046 (DBH) + 0.247 (LV) + 0.067 (GS) + 0.006 (WD) + 0.110 (SD) + 0.287 (SS)	0.46	3.69	38.30	<0.001
	Effective productivity (n h−1)	EEP = 25.752 + 5.173 (LV) − 0.092 (WD) − 0.210 (SD) − 0.536 (SS)	0.63	6.91	204.10	<0.001
	Effective productivity (m3 h−1)	EEP = 28.498 + 5.172 (LV) − 0.007 (GS) − 0.091 (WD) − 0.210 (SD) − 0.535 (SS)	0.63	6.93	93.12	<0.001

Note: SE, standard error; FET, felling effective time; FEP, felling effective productivity; EET, extraction effective time; EEP, extraction effective time; GS, slope gradient (%); WD, winching distance (m), WS, winching slope (%); LV, load volume (m³); SD, skidding distance (m); SS, skidding slope (%). Z₁ and Z₂ are dummy variables that indicate weather conditions as follows: Z₁ = 0 and Z₂ = 0 for sunny, Z₁ = 1 and Z₂ = 0 for rainy, and Z₁ = 0 and Z₂ = 1 for snowy.

). Statistical indices (F-value and p-value) showed that the models were statistically significant (p < 0.001). The independent variables significantly influencing the total felling time and net productivity rate included the stump diameter, weather condition, and the ground slope, among others. Nevertheless, the skidding time model and the effective productivity model were significantly sensitive to variables such as the log diameter, load volume, ground slope, skid-trail slope, winching distance, and skidding distance, among others. The coefficients of determination (R²) for the effective productivity intercepted over 60% of the total variability, on average, which can be explained by the regression equations of the felling and skidding models.

Figure 1 presents the relationship between stump diameter and unit cost of felling. The unit cost of felling decreased exponentially with increasing stump diameter. The diameter class of less than 50 cm had a dramatic effect on the unit cost. The mean felling unit cost ranged from USD 2.65 m⁻³ for the 150 cm diameter class to USD 7.78 m⁻³ for the 25 cm diameter class.

The effect of travel distance on the unit cost of skidding is shown in Figure 2. The skidding unit cost increased linearly with increasing travel distance.

The skidding unit cost was inversely related to the load volume per cycle (Figure 3). The mean skidding unit cost ranged from USD 22.07 m⁻³ for the load volume of 0.51 m³ per turn to USD 1.96 m⁻³ for the load volume of 6.38 m³ per turn. The log volume per cycle was significant in explaining the total skidding unit cost as presented in Figure 3. These results suggest that during the winching phase, the skidder’s operator is typically concerned with a larger payload (i.e., close to the maximum load capacity) to reduce the skidding time and therefore the extraction cost.

Table 10. Productivity (mean ± SD) of chainsaw felling and cable skidding.

Attribute	Chainsaw Felling	Cable Skidding
Total number of harvested trees	214.00	-
Total harvested volume (m3)	762.30	-
Net time consumption (h)	167.70	51.75
Gross time consumption (h)	180.90	71.10
Net productivity (tree h−1)	1.28 ± 0.39	4.13 ± 0.75
Gross productivity (tree h−1)	1.19 ± 0.34	3.01 ± 0.11
Net productivity (m3 h−1)	4.55 ± 1.26	14.73 ± 2.06
Gross productivity (m3 h−1)	4.21 ± 1.05	10.72 ± 1.61

summarizes the gross effective production of felling and skidding per tree and cubic meter of wood for the entire experiment. The total harvested volume was 762 m³, which was extracted within 168 h. The net production of chainsaw felling was 1.28 trees (4.55 m³ h⁻¹), which was 8% higher than the gross production rate. The harvested volume almost took 52 h of the cable skidder traveling out to the landing. The net production of skidding was 4.13 trees (14.73 m³ h ⁻¹), which was 37% higher than the gross production rate.

Table 11. Pollutant emissions (g m⁻³) during felling and extraction practices.

	CO2	CO	HC	NOx	PM10
Felling and processing	125.80	6.67	0.12	3.79	0.43
Skidding	2228.25	25.99	0.40	26.05	3.77
Total	2354.05	32.66	0.52	29.84	4.20

presents components of the GHG emissions for felling and skidding practices. Skidding operations emitted large amounts of emissions, mainly CO₂ (kg 2.23 m⁻³; 95% of total emissions), into the atmosphere, while the use of motor-manual felling showed emissions of about kg 0.13 m⁻³ (5% of total emissions).

The relative change in the CO₂eq emissions under weather conditions ranged from +65% in the rainy condition to +97% in the snowy condition compared with the sunny condition (

Table 12. Greenhouse gas emissions (gCO₂eq m⁻³) produced during felling and skidding as a function of the weather and slope gradient classes.

Source of Variation	Felling and Processing	Skidding
Weather condition
Sunny	694.35 b	-
Rainy	891.08 a	-
Snowy	924.09 a	-
Slope gradient
Gentle	736.63 c	5463.60 c
Moderate	813.60 b	7682.55 b
Steep	926.38 a	10,755.57 a

Note: Different letters in the rows indicate significant differences by Duncan’s test at α = 0.05.

). The relative change in GHG emissions was significantly different under the different weather conditions. Averaged over steep slopes, the relative changes in CO₂eq emissions ranged from +40% on moderate slopes to +97% on steep slopes compared with gentle slopes. Within slope gradients, there were statistically significant differences in the CO₂eq emissions among gentle, moderate and steep slopes.

Delimbing and bucking processes accounted for 25% and 64% of the total emissions released during the felling operation (Figure 4). Within the skidding time elements, travel loaded and travel empty produced the highest amounts of emissions (52% and 23%, respectively).

4. Discussion

Harvesting operations using a motor-manual chainsaw for felling and a cable skidder for bunching–extracting logs have been carried out for more than 50 years in some countries, but much remains unknown about the performance of this practice on mountainous terrain, mostly regarding mixed hardwood stands. This is caused by the variable topographic conditions, the harsh weather, and the use of purpose-built equipment, which hamper the success of the operation compared with flat terrain.

In this study, we developed a generic framework for the analysis of upstream activities of the supply-chain network (felling and skidding) under the assumptions of ground-based harvesting operations in mountainous terrain conditions. Felling and skidding operations are the most expensive tasks in forest practices, and they are highly sensitive to variable weather conditions, stand characteristics, and geophysical factors of the terrain. Nevertheless, these attributes also offer significant advantages for cost reductions and productivity improvements through careful planning and application.

Weather and slopes are typical phenomena of mountainous timber harvesting, which has received little attention from forest engineering sectors to date. Knowledge of the application of harvesting operations obtained from time studies is a key component of the evaluation of various planning scenarios, budgeting, and the allocation of available resources to different parts of forested areas to meet management objectives while improving safety standards and adverse environmental impacts [1].

The results reveal that the ‘walking to the tree’ element accounted for 13.4% of the total felling time in snowy conditions. This was more likely due to the difficulty of traversing over snow and therefore finding marked trees, which increased the total gross effective felling time. In addition to the snow, the longer time of walking between trees may be associated with the slope gradient, which is a common obstacle in mountainous ecosystems. This could be more evident in the case in which uphill movement dominates over downhill movement. Traveling between trees can be significantly reduced using a modern navigation tool, such as a Global Position System (GPS), which allows for quickly finding the location of marked trees, especially in the winter season, while reducing fatigue during the work time [32].

The log processing time (delimbing and bucking) accounted for 67% of the effective gross felling time, which is a significant amount of time compared with other components. Indeed, the felling and processing of logs are the most labor-intensive phases of mountain timber operations, especially in mixed stands. Arguably, this component (e.g., felling and log processing) represents a bottleneck in the production process of the wood supply chain network, determining the success of subsequent practices and generating high profitability values for wood sellers and buyers. Wang et al. [33] found a 20% reduction in timber values due to poor processing practices. The longer log processing time can be attributed to the large volume of hardwood stems, the variable topographic conditions, and the higher density of understory vegetation. Similar patterns were noted by Mousavi [34] and Grzywiński et al. [18], who found that 58% of the labor time was spent during felling and processing in a ground-based harvesting operation.

The felling production rates in our study are lower compared with previously reported results [35] and [10] in the Hyrcanian forest region. The only possible reason for this difference is that, in the present study, felling and processing were carried out as a common practice simultaneously at the stump location. This is quite consistent with the practice of mountainous ground-based harvesting operations to immediately prepare the logs for further transportation [32,34]. In the majority of previous studies, they were treated as two separate practices, mainly due to unfavorable weather conditions and the short duration of the felling period [36]. Generally, harvesting contractors prefer to apply hot systems in which felling and processing operations are synchronized. Under this strategy, the skidder can easily access the bucked assortments and quickly move them to the roadside landing while avoiding a reduction in the timber’s value and therefore reducing damage to the residual trees and forest soil profiles [37]. In addition, this strategy allows contractors to respond quickly to the mill’s demand without further disrupting the mill’s production line.

Our study indicates that increasing slope gradients and the emerging harsh climate significantly hampered the productivity of harvesting operations. The greatest reductions in felling productivity were recorded over the steep slopes in snowy conditions, ranging from 20% to 25%. Analogous to the findings for the felling productivity, the unit costs were increased by 25% to 33% over the steep slopes in the snowy condition. This result is consistent with the findings of Carey et al. [38], where a slope gradient of over 20% was found to reduce the felling productivity by up to 39%. The presence of snow and working on steep terrain resulted in an increase in the felling time and the unit production cost. This could be attributed to the multiple uses of wedges and the deployment of hydraulic jacks to fell trees in a direction opposite to their lean. These situations are common in Hyrcanian forests where stands are mainly composed of mixed deciduous trees with a high timber volume and a massive slanted crown on uneven terrain that necessarily requires more careful consideration than usual [39].

Although the current study confirms that both slope gradients and weather conditions affected the felling productivity and the cost, this is not always straightforwardly the case. This result is not consistent with the findings of Grzywiński et al. [18], in which felling productivity rates were similar in winter (snowy) and summer (sunny) in ground-based harvesting operations.

Harvesting of large-sized trees resulted in increased felling and processing times and showed an inverse relation with the unit production cost as reported previously in a number of productivity studies [40]. In this study, the unit cost of felling production dropped by about 66% when large-sized trees (up to 150 cm DBH) were harvested.

Delays are an inevitable part of harvesting operations and have a significant impact on the productivity and cost. In our case study, technical delays (e.g., the use of obsolete equipment, chain malfunctions, or engine breakdowns) constituted a larger proportion than organizational and personal delays. The delay accounted for 9% of the felling gross effective time, resulting in an average machine utilization of 91%. This means that the productivity of the manual felling operation can the increased by ±9% of the gross effective time by further training the felling crew and investing in new equipment. However, in some cases delays are unavoidable.

Delimbing and bucking processes accounted for 89% of the total CO₂eq emissions produced during the felling operation. The higher emission rate is associated with the large volume of hardwood timbers, the difficulty of operating safely on the terrain, and the use of obsolete tools (more than 10 years of use), which consumed more fuel than usual (e.g., flat terrain, small-sized trees, and mechanized equipment).

Elemental skidding variables, including total cycle time, travel loaded, travel empty, productivity rate, and unit cost, were sensitive to the travel distance and the slope of the skid-trail. The skidding round trip makes up 40% of the total gross effective time and is of a similar dynamic to that reported previously in various skidding productivity studies [2,13,34]. The time required for loaded travel was two times longer than that required for empty travel. Differences in the time consumption between driving loaded and traveling empty can be explained by the uphill movement of a skidder when loaded, slope gradients, and the size of logs, which in turn increase the amount of wheel slippage and reduce the machine’s traction compared with downhill movement when travelling empty. The performance of machines is mainly influenced by the conditions of the terrain on which they are deployed. Operators therefore have a limited opportunity to reduce the time consumption of the skidding component on steeper slopes [41].

The results reveal that the skidding productivity rates decreased by 49% on steeper slopes when compared with more gentle slopes. This was partly due to the dominance of uphill skidding over downhill movement and the difficulty of traveling over steeper slopes. The longer loaded travel time can be attributed to the reduced machine traction and the slower machine speed (44.76 vs. 21.36 m min⁻¹ in unloaded and loaded cycles, respectively), particularly in the case where a large load must be taken uphill. Thus, the machine slips more and can adversely impact the skidding time and productivity, therefore increasing unit skidding costs. Nevertheless, this could be more evident when the skidding distance is increased. The higher load size per turn resulted in an increase in the winching time and the number of skidding cycles, which can positively influence the total skidding time and the overall productivity rates while reducing the unit production costs of skidding operations. This could be attributed to the fact that, during steep terrain operation, due to the variable topographic conditions, the large size of logs (between 6 m and 8 m), and, thus, the heavy weight of the load, the skidder’s operator is not capable of carrying logs at the maximum load capacity of the machine at each turn. Therefore, under these conditions, due to the risk of the machine rolling over and the traction imbalance, the skidding load capacity was always 15%–30% lower than on gentle slopes. In the present study, the average number of logs per cycle was 2.06 (2.94 m³ per cycle on average) on slopes with a gradient of less than 30% and 1.97 (2.57 m³ per cycle) on slopes with a gradient >31%. The unit skidding cost decreased drastically to 38% of the average value (USD 5.25 m⁻³) when the machine load capacity went over 3 m³ per turn. Nevertheless, the unit production cost was at a minimum rate (USD 1.9 m⁻³) when the skidder load capacity reached over 6 m³ per turn. Therefore, it is necessary to improve the machine performance by optimizing the load capacity for the available equipment on steep mountainous terrain. In addition, the choice of the right technology increases laborers’ safety while reducing unfavorable impacts on the environment. Averaged over steep slopes, the unit production costs of skidding increased by 96% when compared with gentle slopes. In addition to the slope gradient, increasing the travel distance positively affected the production skidding unit cost. In this study, the cost ranged from USD 0.82 m⁻³ to USD 10.2 m⁻³, respectively, for travel distances between 15 m and 186 m. The extraction distance was found to be the primary factor affecting machine productivity, consistent with results stated in some earlier studies [19,35,42].

The 14.73 m³ h⁻¹ production rate of skidding obtained from the results is comparable to the productivity of 11.10 m³ h⁻¹ reported by Mousavi [34] and the productivity of 18.51 m³ h⁻¹ reported by Ghaffariyan et al. [10] in other studies in Hyrcanian forest regions. Similarly to the felling operations, the time consumption of the skidding practice involves delay times. Approximately 27% of the gross effective total time was identified as delay time during the skidding operations, which resulted in an average machine utilization rate of 73%. According to Mousavi [34], operational delays and technical delays accounted for almost 85% of the delay time. It is interesting to note that the delay time in skidding operations was three times higher than that in felling operations as skidders require more time for routine maintenance (engine breakdowns, access to technicians, etc.) than chainsaws.

Harvesting operations in the forest ecosystem contribute towards emissions into the atmosphere, showing a negative influence regarding the issue of climate change. A higher amount of emissions was observed in the skidding operations, mainly in loaded travel and empty travel, compared with felling operations with the highest amount in snowy conditions on steep slopes. The increase in GHG emissions could be attributed to the longer skidding time on steep slopes than on gentle slopes. When logs are pulled, the speed of the machine is reduced, considering the lower grip and the higher wheel slip, which obviously results in high fuel consumption and consequently emits more CO₂ emissions compared with gentle slopes. Furthermore, in uphill skidding the rear axle is subject to more load than the front wheels and, as a consequence of the longer time, the machine requires more torque to perform skidding activities. This is in agreement with Tavankar et al. [26], who observed a higher amount of emissions under moist soil conditions over steep slope gradients. In addition to slope gradients, operations in a harsh climate, mainly snow, led to an increase in the amount of CO₂ emissions. This was particularly true when the skid trail was moist. Operators must be more cautious during operations in this type of condition due to the fact that a high level of moisture on the skid trail could make the route slippery and therefore prone to rollover [37]. This causes higher fuel consumption as a consequence of the longer time and thus emits more emissions than in sunny conditions. Moreover, the use of outdated equipment is another reason for high amounts of emissions [17]. The skidders used in our case study are rather obsolete (more than 25 years old) with low quality standards. Continuous deployment of these outdated machines in mountain timber harvesting induces time delays, which not only limit the operational capacity but also lead to a higher environmental footprint.

5. Conclusions

This study provided detailed information about the time elements, productivity, cost, and GHG emissions associated with upstream activities of forest operations (i.e., felling and skidding) on steep slopes under a ground-based harvesting operation. The results demonstrate that the effects of slope gradients and weather conditions may create particularly challenging conditions for mountain harvesting operations. The productivity rates of felling and skidding decreased substantially with a consistent increase in the production cost on both steep slopes and snowy conditions. In addition to the productivity and cost, CO₂eq emissions caused by the harvesting equipment increased as a result of the harsh climate over steep slopes. As a consequence, we recommended that harvesting operations be scheduled to occur during sunny weather (if the timber demand would allow it) and that skidding operations on steep slopes (more than 35%) be avoided in order to obtain higher safety and work efficiency and to minimize the adverse effects of machinery on forest ecosystems. Mechanized felling (e.g., purpose-built tracked feller-bunchers) or a small-scale cable yarding system could be applied on steep terrain where ground-based harvesting equipment cannot effectively and safely operate. Directional felling toward skidding routes can decrease the time required for felling and skidding operations while reducing unit production costs. Given the large-sized trees per turn (more than 3 m³) in downhill skidding, extending the length of skidding routes may allow for a reduction in the unit production cost of operations (up to 38%). This may avoid traction on steeper slopes over long distances and limit adverse impacts caused by wood harvesting and extraction machines on the forest ecosystem (e.g., damage to soil profiles and residual stands); however, we have not addressed these issues in the present study. Given the increasing environmental awareness coupled with increased harvesting costs, it is therefore necessary to quantify and compare the environmental impacts of machines used throughout the entire supply-chain network from a life cycle assessment perspective. Integrating this information with traditional optimization models can help harvesting managers to make better operational decisions using the information on machine productivity, cost, and GHG emissions.