Article Perception of the Harvester Operator’s Working Environment in Windthrow Stands

Grzegorz Szewczyk 1,*, Raffaele Spinelli 2 , Natascia Magagnotti 2, Bartosz Mitka 3 , Paweł Tylek 1 , Dariusz Kulak 1 and Kamil Adamski 4

1 Department of Forest Utilization, Engineering and Technology, Faculty of , University of Agriculture in Krakow, al. 29 Listopada 46, 31-425 Kraków, Poland; [email protected] (P.T.); [email protected] (D.K.) 2 Institute of Bioeconomy, CNR IBE National Council for Research, Via Madonna del Piano 10, I-50019 Sesto Fiorentino (FI), Italy; [email protected] (R.S.); [email protected] (N.M.) 3 Department of Agricultural Land Surveying, Cadastre and Photogrammetry, Faculty of Environmental Engineering and Land Surveying, University of Agriculture in Krakow, Balicka 253a, 30-198 Kraków, Poland; [email protected] 4 Krosno Forest District, Regional Directorate State Forests in Zielona Góra, Osiecznica, ul. Kro´snie´nska42, 66-600 Krosno Odrza´nskie,Poland; [email protected] * Correspondence: [email protected]; Tel.: +48-12-662-5086

Abstract: The aim of the study was to determine the mental workload of a harvester operator when working in late and in windthrown stands of the same type and age, using eye movement patterns as an indicator. Eyeball movement variability was analysed using the eye tracking method. The mean duration of eyesight fixations in windthrown stands was shorter than in the control undamaged stands by about 20% (444 ms and 534 ms, respectively). The mean time of eyesight movements (saccades) in the windthrown stands was shorter than in the control undamaged stands   by approx. 15%. The largest differences between the duration of saccades in the windthrown and control stands were observed between the cutting of and cutting logs off their root plates: Citation: Szewczyk, G.; Spinelli, R.; Magagnotti, N.; Mitka, B.; Tylek, P.; the saccades were longer by about 20% when working in the control stands (49 ms) as compared Kulak, D.; Adamski, K. Perception of to the windthrown stands (43 ms). Large differences in the duration of saccades between the the Harvester Operator’s Working windthrown area (42 ms) and the control area (47 ms) were also found when travelling between Environment in Windthrow Stands. successive operation sites. In both types of stands, the shortest saccades were observed during Forests 2021, 12, 168. https:// processing: 39 ms. Summary durations of saccades observed during the processing of successive doi.org/10.3390/f12020168 trees occurred in sequences showing repeated periods of variable eyeball activity, where longer saccades were followed by shorter ones. Documented more variability of eyesight activities of the Academic Editor: Angela Lo Monaco harvester operator performing the operations of processing and moving is new standard of eye balls Received: 8 January 2021 activities for the more taxing work conditions presented by windthrown stands. Accepted: 28 January 2021 Published: 31 January 2021 Keywords: timber harvesting; post-disaster; mental workload; eye-tracking; ergonomics

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- 1. Introduction iations. Catastrophic wind damage occurred long before the beginning of human interference in natural forest ecosystems. Windthrow events have always been an important process in the forest environment and often a necessary regenerating force that affected the vari- ability and diversity of forests [1]. In temperate regions, winds have a dominant role Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. among all weather phenomena. Together with insect outbreaks, floods, fires and droughts, This article is an open access article winds are among the most important natural phenomena that change ecosystems [2]. The distributed under the terms and first official records of catastrophic wind damage appeared in Europe in the nineteenth conditions of the Creative Commons century. In Poland, one can find reports about four hurricanes striking the Piska Forest at Attribution (CC BY) license (https:// that time. About 130 hurricanes have been recorded in Europe since the mid-20th century, 3 creativecommons.org/licenses/by/ with an average annual damage to forests estimated at about 35 million m of timber [3,4]. 4.0/). In mountain areas, a noticeable increase of wind damage was observed right after World

Forests 2021, 12, 168. https://doi.org/10.3390/f12020168 https://www.mdpi.com/journal/forests Forests 2021, 12, 168 2 of 16

War I. For example, in 1920, a windthrow event in the Sudetes exceeded the five-year prescribed harvest. Today, wind damage is considered the consequence of the inadequate forest management practices introduced in the 19th century, which led to the establishment of even-aged monocultures, especially vulnerable to natural disturbances [5]. However, the large-scale catastrophic damage experienced in recent years is certainly related to the increasingly frequent and violent weather events, possibly derived from the growing global temperature [6–8]. Since 1990, several hurricanes have been recorded in Europe, including: Vivian and Wiebke (120 million m3 of windthrown timber), Lothar (200 million m3) and Kiryll (64 million m3). Most likely, the frequence and violence of such events will continue to increase [9–12]. In northern Poland, enormous damage was recorded in the summer of 2017, when heavy wind storms totally destroyed an area of 120,000 ha. The damage was es- timated at almost 10 million m3, and over 39,000 ha remained in the need of complete forest regeneration. That was the largest forest damage event in the history of Polish forestry. Salvage harvest of damaged timber is the primary task of forest services. This task is extremely challenging when it comes to work safety and productivity. High accident risk derives from the production of timber with characteristics that are nonstandard and difficult to evaluate [13]. Furthermore, timber harvesting in windthrown stands offers particularly difficult conditions when it comes to moving, setting a safe work sequence and minimizing additional value losses. There is a huge diversity in the forms of wind damage, all hiding some danger: lifted root plates that may fall on the operators once severed from the stems, bent trunks that may snap suddenly and unpredictably when crosscut, broken trees leaning over other standing trees. Such exceptionally dangerous working conditions discourage motor-manual work, since operators are fully exposed, with minimal protection [14]. Therefore, since the1990s, the use of harvesters has become the standard practice in windthrown stands [15–21]. By placing the operator inside a protected cab at a safe distance from the cut zone, the number and the severity of accidents recorded during timber harvesting in post-disaster areas has been reduced significantly [22–24]. Harvester work qualifies as partly-automated work, since travelling between work stations and trees require full operator participation, whereas wood processing is largely automated and the operator just plays a supervisory role. The StanForD control and measurement standard has been in operation since the 1980s, and allows easy exchange of automated working protocols [17]. However, partially automated work from a comfortable cab is only seemingly undemanding; in fact it generates a significant cognitive and psycho- logical load on operators [25,26]. The monotony and regularity of external stimuli generate a slowly developing condition of reduced awareness [27], while the complexity of some tasks may cause mental fatigue and psychological stress. That may lead to specific accident hazard and to a general negative effect on workers’ health [28,29]. Paradoxically, difficult working conditions, such as steep terrain in the mountains [30], or the heterogeneous nature of the salvage task in post-disaster areas, may mitigate the above-mentioned risks, at least to some extent. Therefore, the mental workload associated with work in post- disaster areas maybe peculiar, and somewhat different from that experienced under the conditions of regular cuts. The identification of the harvester operator’s mental workload under standard and post-disaster conditions should be the starting point for designing safe and efficient technological systems and work routines. It is particularly important to determine the degree of difficulty of each task and correlate it with work efficiency, mental workload and fatigue, as characteristic of automated production processes applied to forestry. Therefore, the goal of this study was to gauge the mental workload experienced by harvester operators working in post-disaster areas and compare its level and variability with those experienced during conventional planned operations in undamaged stands of the same type and age. Forests 2021, 12, x FOR PEER REVIEW 3 of 16

Forests 2021, 12, 168 3 of 16

2. Materials and Methods 2. MaterialsThe study and was Methods conducted in south-western Poland, in the Rudy Raciborskie Forest DistrictThe –the study Regional was conducted State Forest in south-western Directorate (RDSF) Poland, of in Katowice the Rudy Raciborskie(Figure 1). Forest District –the Regional State Forest Directorate (RDSF) of Katowice (Figure1).

FigureFigure 1. 1.ResearchResearch location: location: the RudyRudy Raciborskie Raciborskie Forest Forest District. District.

InIn the the summer summer of 2017,2017, thethe area area was was hit hit by by a hurricane a hurricane that that completely completely destroyed destroyed overover 800 800 ha ha of of forest forest and severelyseverely damaged damaged an an additional additional 1500 1500 ha. Theha. The estimated estimated wind- wind- damaged timber volume amounted to 260,000 m3, which was almost three times the size of damaged timber volume amounted to 260,000 m3, which was almost three times the size the planned annual felling. In the Solarnia Forest Range, where the research was carried of out,the planned the annual annual felling felling. increased In bythe 60,000 Solarnia m3 (FigureForest2 Range,). For safety where reasons the research and to obtain was carried out,adequate the annual work felling efficiency, increased harvesters by and60,000 m3 (Figure were 2). used For to safety clear thereasons disaster and areas. to obtain adequateOverall, work 20 such efficiency, sets operated harvesters in the area. and forwarders were used to clear the disaster areas. Overall,In 20 particular, such sets study operated data was in collectedthe area. in forest compartments 411d and 410c (Table1). In these compartments, 95% of the total damage consisted of windthrows, and only 5% of windsnaps. The analyzed stands grew in a flat area. Before the work commenced, corridors were designated in typical parts of the anal- ysed areas. Corridors had a total length of approx. 200 m each, were spaced 15 m apart and were placed both in the damaged areas and in the adjacent forest areas that had been left undamaged. In the damaged areas, corridors were set perpendicular to the wind direction, so as to allow for processing trees on the skid trails in front of the machine and on its sides. Each research plot corresponded to the area covered by a single corridor. Therefore, the study included two types of research plots: damaged and undamaged (i.e., control). A 770 harvester travelled along the corridors and processed trees into 2.5 m and 5 m long logs. In the undamaged areas, the harvester also performed felling. The technical characteristics of the machine are presented in Table2. The machine was 7 years old, had clocked 14,000 machine-hours and it was in good technical condi- tions (Figure3 ). Before starting work, the operator was properly instructed about work requirements, and in particular about safety procedures and log quality specifications. All work covered in this study was carried out by the same forty-five-year-old operator with 10 years’ experience in mechanized harvesting jobs and 4 years’ experience on this particular machine. The operator had no diagnosed vision disorders.

Figure 2. Disaster areas in the Solarnia Forest Range in 2017, Rudy Raciborskie Forest District, forest compartments 411d and 410c.

Forests 2021, 12, x FOR PEER REVIEW 3 of 16

2. Materials and Methods The study was conducted in south-western Poland, in the Rudy Raciborskie Forest District –the Regional State Forest Directorate (RDSF) of Katowice (Figure 1).

Figure 1. Research location: the Rudy Raciborskie Forest District.

In the summer of 2017, the area was hit by a hurricane that completely destroyed over 800 ha of forest and severely damaged an additional 1500 ha. The estimated wind- damaged timber volume amounted to 260,000 m3, which was almost three times the size of the planned annual felling. In the Solarnia Forest Range, where the research was carried out, the annual felling increased by 60,000 m3 (Figure 2). For safety reasons and to obtain Forests 2021, 12, 168 4 of 16 adequate work efficiency, harvesters and forwarders were used to clear the disaster areas. Overall, 20 such sets operated in the area.

FigureFigure 2. Disaster2. Disaster areas areas in the in Solarnia the Solarnia Forest Range Forest in 2017, Rang Rudye in Raciborskie 2017, Rudy Forest Raciborskie District, forest Forest District, compartmentsforest compartments 411d and 410c. 411d and 410c.

Table 1. Forest appraisal features in forest compartments 411d and 410c under analysis.

Species Merchantable Age Disaster Location Compartment Area [ha] Site Type Composi- DBH [cm] Height [m] Timber [Years] [%] tion [m3·ha−1] Humid Pine 40% 22 10 9 32 Solarnia 411d 3.65 mixed Pine 30% 28 15 15 42 50 F.R. forest Alder 30% 28 12 12 28 Humid Pine 60% 33 17 15 77 Solarnia 410c 3.15 mixed Pine 20% 33 17 16 12 50 F.R. forest Pine 20% 33 16 15 19

Table 2. General technical data of the John Deere 770 D harvester.

Parameter Value Weight [t] 11.55 Dimensions: length/width/height [m] 5.91/2.45/3.62 Crane: type/range [m] John Deere 140H/7.9m Cutting head Waratah H412 Engine power (kW/KM) 140/190.35 Forests 2021, 12, 168 5 of 16 Forests 2021, 12, x FOR PEER REVIEW 5 of 16

FigureFigure 3. 3.The The John John Deere Deere 770d 770d harvester harvester during during operation operation on on a researcha research plot plot [photo [photo by by K. K. Adamski]. Ad- amski]. During work, the harvester operator’s eyeball activity was recorded with the use of theDuring Tobii Pro work, Glasses the harvester 2 reflective operator’s eye-tracker eyebal (Tobiil activity AB, CO)was recorded (Table3) andwith the the Tobii use of Prothe GlassesTobii Pro Controller Glasses 2 software reflective installed eye-tracker on a(Tobii portable AB, computerCO) (Table [31 3)]. and Before the Tobii starting Pro theGlasses work, Controller the operator software was informed installed about on a the portable purpose computer of testing [31]. and Before the measurement starting the method.work, the The operator measurements was informed started atabout 9 am, th aftere purpose a 60-min of warm-up. testing and After the calibrating measurement the glasses,method. the The recording measurements session started began at and 9 am, the af operatorter a 60-min performed warm-up. routine After tree calibrating felling and the buckingglasses, tasks.the recording session began and the operator performed routine tree felling and bucking tasks. Table 3. Technical data of the Tobii Pro Glasses 2 eye-tracker. Table 3. Technical data of the Tobii Pro Glasses 2 eye-tracker. Eye-tracker Eye-tracker 50–100 Hz (in respect to eye-trackers the sampling 50-100 Hz (in respect to eye-trackers the frequency means the number of identified locations of sampling frequency means the number of Sampling frequency fixation points per one second. This frequency determines Sampling frequency the qualityidentified of results locations and the accuracy of fixation of measurementspoints per one taken) second. This frequency determines the quality of Cameras 4 results and the accuracy of measurements taken) CamerasScene camera FOV 82◦ horizontally,4 52◦ vertically SceneScene camera camera FOV parameters h.264; 1920 82°× 1080horizontally, pixels; @25 52° fpsvertically ◦ SceneField ofcamera view parameters 160 h.264; 1920 × 1080 pixels; @25 fps ◦ FieldDiagonal of view of scene camera FOV 90 ; 16:9 160° Sound recording Yes Diagonal of scene camera FOV 90°; 16:9 Weight 45 g Sound recording Yes Battery 120 min. Weight 45 g BatteryRecording Station 120 min. RecordingConnection Station HDMI, Micro USB, 3.5 mm Jack ConnectionFrequency 2.4 GHz & HDMI, 5 GHz bandMicro USB, 3.5 mm Jack FrequencyDimensions 130 × 85 × 2.427 GHz mm & 5 GHz band DimensionsWeight 312 g 130 × 85 × 27 mm Weight 312g The recording sessions lasted approx. 40 min each. After each session, the data was automaticallyThe recording saved, sessions the eye-tracker lasted approx. was calibrated 40 min againeach. After and a each new sessionsession, wasthe started.data was automaticallyThe measurements saved, the made eye-tracker it possible was to calibrated obtain film again material and a that new was session processed was started. using the TobiiThe Promeasurements Lab Analyzer made version it possible 1.102 to software obtain film (Tobii material AB, CO). that The was analysed processed videos using allowedthe Tobii for Pro distinguishing Lab Analyzer four version work activities1.102 soft forware which (Tobii suitable AB, CO). measurement The analysed boundary videos pointsallowed were for defined.distinguishing These were:four work activities for which suitable measurement boundary points were defined. These were:

Forests 2021, 12, 168 6 of 16

Activity 1: felling—it started the moment the harvester head was positioned at the base of the tree stem and ended with the first passage of the cut stem through the feed rollers in the head (this activity was conducted only on the standard, undamaged research plots); Activity 2: cutting off the tree stump—it started when the harvester head was posi- tioned at the base of the stem to be cut and ended when cross-cutting was completed (this activity was conducted only on the damaged research plots); Activity 3: delimbing and cross-cutting—it started with the first passage of the stem through the feed rollers in the head and finished when the tree top was cut off (on all research plots); Activity 4: moving—it started the moment the machine moved from the current work station after discarding the top of the previously processed tree and ended when the machine had reached the next work station and the harvester head had been positioned at the base of the next tree to be cut (beginning of Activity 1 or Activity 2, depending on plot type). The work recorded was smooth, with no breaks related to repairs, servicing or rest. All recordings were scanned using the standard Tobii Pro Labi-VT filters [32]. The At- tention Filter was applied in the configuration of snapshots and heatmaps. We used the Attention Filter because the recordings were performed in dynamic situations, where the operator was in constant motion. In this situation, a large variability of eyeball movements can be observed: fixations, saccades, smooth pursuits and VOR (vestibulo-ocular reflex). The Attention Filter in Pro Lab is essentially the Tobii Pro IV-T Filter, with the velocity threshold parameter set at 100 degrees/second instead of the default 30 degrees/second. Based on the analysed film material, snapshots (individual film frames) were selected for the purpose of preparing heatmaps. Heatmaps were generated for the processing of randomly selected work batches containing five subsequent trees in both of the analysed stand types (i.e., damaged and control). For greater clarity, the heatmaps were generated in a simplified form. The original images were calibrated to snapshots with overlaid heatmaps using the Bentley Descartes software. The calibration was done by means of projective transformation with the use of 4 common points. The next step was to suppress the heatmaps and to vectorise the head with the accompanying elements in the photos. After vectorisation, the snapshots were suppressed and the heatmaps were brought back, obtaining the effect of a vector model being superimposed on the heatmap. The conversion of the coloured image to gray scale was done in Adobe Photoshop. On the snapshots described above, we designated five areas of interest (AOIs) as fol- lows: AOI 1 = the harvester crane and head; AOI 2 = the stem being processed; AOI 3 = the log storage area; AOI 4 = the felling site; AOI 5 = stand and corridor. We scanned the films with the fixation filter configured as “standard” in Tobii Pro Lab, with the following settings: gapfill-in (interpolation) = disabled; eye selection = average; noise reduction = moving median; I-VT fixation classifier = threshold 30 degrees/second; max. time between fixations = 75 ms; max. angle between fixations = 0.5 degrees; minimum fixation duration = 60 ms. Fixations were noted individually in relevant AOIs for the whole footage. The research into the operator’s eye activity during a work shift (the length of fixations and saccades), considering AOIs and the analysed stands, was conducted based on data generated in the Tobii Pro Lab software: metrics and raw data. In view of the oblique distribution of fixation durations, the significance of differences of the median of the analysed variable was determined based on the Kruskal-Wallis non parametric test. The variability of the duration of fixations observed during the harvester’s operation was determined by presenting the measurement data as a time series, i.e., a sequence of observations of a given variable as a function of time [18,33]. The harmonic structure of the series (i.e., observed sequence of saccades) was determined using the autocorrelation function, after prior logarithm transformation of the original data. The variability of successive, consecutive sums of saccade durations during work on one tree was defined as Forests 2021, 12, x FOR PEER REVIEW 7 of 16

Forests 2021, 12, 168 7 of 16 the mean value of the indicators of variation of saccade durations observed during work on the previous and the next tree (1). the mean value of the indicators of variation of saccade durations observed during work on the previous and the next tree (1). 𝑎 × 100 (1) 𝑆 = 𝑎 ax−1 × 100 where: Snm = (1) S—indicator of variation of saccade durationsax [%] where:n—activity (p—processing, m—moving, c—cutting), S—indicatorm—stand category of variation (s—standard of saccade durations stand, p—post-disaster [%] stand), n—activityax—sum of (p—processing, saccade durationsm—moving, duringc—cutting), work with a tree [ms] m—stand category (s—standard stand, p—post-disaster stand), 3. Resultsax—sum of saccade durations during work with a tree [ms]

3.The Results analysed study material included over 5000 sec. of recordings in damaged stands and approx. 30,000 sec. of recordings in undamaged, control stands. Heatmaps were gen- The analysed study material included over 5000 s of recordings in damaged stands and eratedapprox. on the 30,000 basis s of of recordings over 1500 in fixation undamaged, points. control Analyses stands. of Heatmaps the duration were generatedof saccades on and fixationsthe basis were of overperformed 1500 fixation for the points. entire Analyses database, of thewithin duration the selected of saccades AOIs: and 219 fixations complete cycleswere in performedthe damaged for the stands entire and database, 469 in withinthe undamaged the selected stands. AOIs: 219 In total, complete about cycles 3500 in fixa- tionsthe and damaged saccades stands in the and damaged 469 in the stands undamaged and almost stands. 25,000 In total, fixations about 3500 and fixations saccades and in the controlsaccades stands in thewere damaged obtained. stands and almost 25,000 fixations and saccades in the control standsThe mean were obtained.time of saccades in the damaged stands was significantly shorter than in the controlThe stands mean time (Figure of saccades 4, Table in 4). the Significant damaged standsdifferences was significantly in the duration shorter of thansaccades in in the thecompared control stands stands (Figure were 4also, Table found4). Significant between differences felling and in cutting the duration off the of saccadesroot plates. in the compared stands were also found between felling and cutting off the root plates.

control areas - average control areas - moving control areas - processing control areas - cutting damaged areas - average damaged areas - moving damaged areas - processing damaged areas, cutting off

0 0.2 0.4 0.6 Saccade [ms.]

Figure 4. The duration of saccades at the work station of the harvester operator. Figure 4. The duration of saccades at the work station of the harvester operator.

Table 4. Results of the U Mann-Whitney test: saccade duration in the damaged and undam- Table 4. Results of the U Mann-Whitney test: saccade duration in the damaged and undamaged aged stands. stands. Operation Z p Operation Z p Total of all operations −1.839 0.049 TotalCutting/Cutting of all operations off −−2.1431.839 0.0300.049 Cutting/CuttingProcessing off −−0.5102.143 0.6100.030 ProcessingMoving − 0.7140.510 0.4700.610 Moving 0.714 0.470 In both types of stands, the shortest saccades were observed during processing (39 ms andIn both 43 ms types respectively). of stands, The the main shortest element saccades of the operator’swere observed vision during was the processing log being (39 processed in the vicinity of the harvester head and the cutting device (Figure5). When the ms and 43 ms respectively). The main element of the operator’s vision was the log being processed in the vicinity of the harvester head and the cutting device (Figure 5). When the log being processed was moved from the right to the left side of the operator’s field of view, over 50% of the saccades were located on the processed stem at the point of its cross-

Forests 2021, 12, 168 8 of 16 Forests 2021, 12, x FOR PEER REVIEW 8 of 16

Forests 2021Forests, 12 2021, x FOR, 12, PEERx FOR REVIEW PEER REVIEW 8 of 16 8 of 16 log being processed was moved from the right to the left side of the operator’s field of view, cuttingover 50%(i.e., of AOI the saccades2). When were the locatedlog was on moved the processed to the right stem atside—which the point of itswas cross-cutting less frequent— (i.e., AOI 2). When the log was moved to the right side—which was less frequent—the the same AOI cutting2 attracted (i.e., AOI only 2). about When the37% log of was the moved saccades. to the The right remaining side—which observations was less frequent— same AOIcutting 2 attracted (i.e., AOI only 2). When about the 37% log of was the saccades.moved to Thethe right remaining side—which observations was less were frequent— were evenly distributedthe same AOI between 2 attracted the harvester only about head 37% andof the the saccades. place where The remaining the wood observations was evenlythe distributed same AOI between 2 attracted the harvester only about head 37% and of the the place saccades. where The the woodremaining wasstored observations were evenly distributed between the harvester head and the place where the wood was stored(AOI (AOI 1,were AOI 1, 3).evenlyAOI 3). distributed between the harvester head and the place where the wood was storedstored (AOI 1,(AOI AOI 1, 3). AOI 3).

Figure 5. Heatmap of the harvester operator’s field of view during processing in the control and the Figure 5. HeatmapFigureFigure 5. Heatmap of the 5. Heatmap harvester of the harvesterof operator’sthe harvester operator’s field operator’s of field view of field viewduring of during view processing during processing processing in thein the control controlin the and control and and damaged stands.the damaged stands. the damagedthe damagedstands. stands.

Large differencesLarge Largedifferences in differences the duration in the induration ofthe saccades duration of sacc between ofades sacc betweenades the damagedbetween the damaged the area damaged (42 area ms) (42 andarea ms) (42 and ms) and Large differences in the duration of saccades between the damaged area (42 ms) and the controlthe control areathe control (47 area ms) (47 area were ms) (47 were found ms) foundwere for travellingfound for trav forelling trav between ellingbetween successivebetween successive successive operation operation operation sites. sites. Thesites. The theThe control fieldfield ofarea viewoffield view(47 of of ms) the ofview the machinewere ofmachine the found machine operator operator for trav duringoperator duringelling this during betweenthis operation operation this successiveoperation was was definitely definitely was operation definitely narrower narrower sites. narrower in inThe the in the fieldthe of damaged viewdamaged ofdamaged standsthe stands machine (AOI stands (AOI 5) operator (Figure5)(AOI (Figure 5)6 ).(Figureduring 6). It coveredIt covered 6). this It covered anoperation an area area withan with areawas numerous numerous withdefinitely numerous root root narrower plates plates root of plates of in wind- the of wind- damagedwindthrownthrown stands trees,thrown trees,(AOI directly trees, directly5) (Figure indirectly frontin front6). of inIt the coveredoffront harvesterthe of harves thean onareaharvester both on withter sidesboth onnumerous ofsidesboth the ofsides crane. the root of Undercrane. platesthe crane. thoseUnder of wind- Under those those difficult working conditions, more than 70% of saccades were concentrated on the access thrown trees,difficult directlydifficult working workingin conditions,front conditions, of the more harves thanmoreter 70% than on of 70%both saccades of sides saccades were of theconcentratedwere crane. concentrated Under on the onthose access the access corridor. In the undamaged control stands, the operator searched for trees designated for difficult workingcorridor.corridor. Inconditions, the Inundamaged the undamaged more controlthan control70% stands, of stands,saccades the operator the wereoperator searched concentrated searched for trees for on designatedtrees the designated access for for felling, growing both in front of the cab and on both of its sides. Therefore, the operator’s corridor. felling,In thefelling, undamagedgrowing growing both control in both front in stands, offront the ofcab thethe and caboperator on and both on searched of both its sides.of its for sides.Therefore, trees Therefore, designated the operator’s the operator’sfor field of vision was wider and focused on the stand (approx. 50% of the saccades), on the field offield vision of vision was wider was wider and focused and focused on the on stand the stand(approx. (approx. 50% of 50% the ofsaccades), the saccades), on the on the felling,processed growing wood both (almost in front 30% of of the the saccades) cab and and on onboth the of harvester its sides. head Therefore, and crane the (approx. operator’s processedprocessed wood wood(almost (almost 30% of 30% the ofsaccades the saccades) and )on and the on harvester the harvester head andhead crane and crane(ap- (ap- field17%) of (Figurevision7 was). The wider wide fieldand offocused vision wason the the stand result of(approx. the operator’s 50% of eyeball the saccades), activity as on the prox. prox.17%) (Figure17%) (Figure 7). The 7). wide The fieldwide offield visi ofon visi wason the was result the resultof the of operator’s the operator’s eyeball eyeball processedwell as his wood head (almost movements, 30% which, of the however, saccades was) and not on analysed the harvester separately. head and crane (ap- activityactivity as well as as well his ashead his movements,head movements, which, wh however,ich, however, was not was analysed not analysed separately. separately. prox. 17%) (Figure 7). The wide field of vision was the result of the operator’s eyeball activity as well as his head movements, which, however, was not analysed separately.

Figure 6. Heatmap of the harvester operator’s field of vision during a run in the post-disaster Figure 6.FigureHeatmap 6. Heatmap of the harvester of the harvester operator’s operator’s field of vision field of during vision a runduring in the a run post-disaster in the post-disaster stands. stands.stands. Figure 6. Heatmap of the harvester operator’s field of vision during a run in the post-disaster stands.

Figure 7.FigureHeatmapFigure 7. Heatmap of 7. the Heatmap harvester of the harvesterof operator’s the harvester operator’s field operator’s of visionfield of duringfield vision of avisionduring run in during a the run standard in a runthe standardin stands. the standard stands. stands.

Figure 7. Heatmap of the harvester operator’s field of vision during a run in the standard stands.

Forests 2021, 12, x FOR PEER REVIEW 9 of 16

The largest differences between the duration of saccades in the damaged and the un- Forests 2021, 12Forests, x FOR2021 PEER, 12, 168REVIEW 9 of 16 9 of 16 damaged control stands were observed for tree felling (Activity 1 - performed on undam- aged plots) and for cutting stems off root plates (Activity 2 - performed on damaged plots) (Figure 8,9). During work in the control stands, saccades were longer by approx. 20% (49 The largestms). TheThe differences operator’s largest differencesbetween field of the view between duration basically the of saccades durationcovered inthe of the saccadesharvester damaged in head theand (AOIdamaged the un- 1), on and which the damaged controlalmostundamaged 70%stands of control wereobservations observed stands focussed, werefor tree observed felling and the (Activity for tree tree trunk 1 felling - performed at the (Activity place on of undam- 1 cutting - performed and just on aged plots)belowundamaged and for it (AOI cutting plots) 2) (approx.stems and off for 20% root cutting ofplat the stemses saccades (Activity off root). 2 Only plates- performed about (Activity 10% on 2damagedof - observations performed plots) on were damaged con- (Figure 8,9).nectedplots) During (Figureswith work felling8 andin the site9). control During(AOI 4). stands, work Cutting in saccades the stem controls offwere stands,root longer plates saccades by in approx. the were damaged 20% longer (49 stands by approx. was ms). The operator’susually20% (49 done ms). field in The of such view operator’s a basicallyway that field coveredthe of place view the of basically harvester cuttingcovered remained head (AOI the invisible harvester 1), on whichto headthe operator, (AOI 1), almost 70%thereforeon of which observations he almost mainly 70%focussed, observed of observations and the the harvester tree focussed, trunk head at and (approx.the the place tree 40%of trunk cutting of the at the andsaccades) place just of and cutting the below it (AOIprocessedand 2) just (approx. below wood 20% it (approx. (AOI of the 2) 37%(approx.saccades of the). 20% Onlysaccades) of about the saccades).(AOI 10% 1, of AOI observations Only 2). about 10%were of con- observations nected withwere felling connected site (AOI with 4). Cutting felling site stem (AOIs off 4).root Cutting plates in stems the damaged off root plates stands in was the damaged usually donestands in such was a usually way that done the in place such of a waycutting that remained the place invisible of cutting to remained the operator, invisible to the therefore heoperator, mainly thereforeobserved hethe mainly harvester observed head the(approx. harvester 40% headof the (approx. saccades) 40% and of the the saccades) processed andwood the (approx. processed 37% wood of the (approx. saccades) 37% (AOI of the 1, AOI saccades) 2). (AOI 1, AOI 2).

Figure 8. Heatmap of the harvester operator’s field of view during tree cutting in the undamaged stands.

Figure 8. HeatmapFigure 8.of theHeatmap harvester of theoperator’s harvester field operator’s of view during field oftree view cutti duringng in the tree undamaged cutting in the undam- stands. aged stands.

Figure 9. Heatmap of the harvester operator’s field of view while cutting stems off root plates in the damaged stands. Figure 9. Heatmap of the harvester operator’s field of view while cutting stems off root plates in the damagedThe average stands. eye fixation duration in the damaged stands was shorter than in the control stands by approx. 20% (444 ms and 534 ms, respectively). This difference was statistically The average significant eye fixation (Figure 10duration). The coefficient in th e damaged of variation stands of thewas fixation shorter duration than in wasthe Figure 9. Heatmapcontrol68% for ofstands the the control harvester by approx. stands operator’s 20% and 75%(444 field forms of theviewand damaged 534while ms, cutting respectively). stands. stems We off observed root This plates difference a lowerin variationwas sta- the damagedtisticallyin stands. the eye significant fixation length (Figure in the10). damaged The coefficient stands, of where variation halfof of the the observations fixation duration fell within was the 200–600 ms range. However, the share of outliers was higher in this group. The average eye fixation duration in the damaged stands was shorter than in the control stands by approx. 20% (444 ms and 534 ms, respectively). This difference was sta- tistically significant (Figure 10). The coefficient of variation of the fixation duration was

Forests 2021, 12, x FORForests PEER 2021 REVIEW, 12, x FOR PEER REVIEW 10 of 16 10 of 16

68% for the control68% stands for the and control 75% forstands the damagedand 75% forstands. the damaged We observed stands. a lower We observed varia- a lower varia- Forests 2021, 12, 168 10 of 16 tion in the eye fixationtion in lengththe eye in fixation the damaged length stands,in the damaged where half stands, of the where observations half of thefell observations fell within the 200–600within ms range. the 200–600 However, ms range.the sh areHowever, of outliers the wasshare higher of outliers in this was group. higher in this group.

Figure Figure10. Positional 10. PositionalFigure statistics 10. statistics Positional of the fixation of thestatistics fixation time of in the time the fixation damaged in the damagedtime and in undamagedthe and damaged undamaged stands. and undamaged stands. stands.

The summaryThe summary durationThe durationsummary of saccades, of duration saccades, within of within saccades,activities activities observedwithin observed ac tivitiesduring during observedthe processing the during processing the processing of successiveof successive trees,of successive trees,occurred occurred in trees, sequences in occurred sequences showing in sequences showing repeated repeated sh periodsowing periods repeated of variable of periods variable eyeball of eyeball variable eyeball activityactivity (Figure (Figure 11).activity As 11a (Figurerule,). As longer a11). rule, Assummary longera rule, longer summarytimes of summary saccades times times of were saccades offollowed saccades were by were followedmark- followed by by mark- edly shortermarkedly ones. shorteredly shorter ones. ones.

45 (A) 45 40 felling,40 post disaster areas felling, post disaster areas 35 felling,35 standard areas felling, standard areas 30 30 25 25 20 20 15 15 Saccade [s] Saccade [s] 10 10 5 5 0 0 1 4 7 10 13 161 19 4 22 7 25 10 28 13 31 16 34 19 37 22 40 25 43 28 46 31 49 34 52 37 55 40 58 43 61 46 64 49 67 52 70 55 58 61 64 67 70 Observation number Observation number

90 (B) 90 80 80 processing, post disaster areas processing, post disaster areas 70 processing, standard70 areas processing, standard areas 60 60 50 50 40 40 30 30 Saccade [s] Saccade [s] 20 20 10 10 0 0 1 4 7 1013161922252831343740434649525558616467701 4 7 101316192225283134374043464952555861646770 Observation number Observation number

40 (C) 40 35 moving, post 35disaster areas moving, post disaster areas moving, standard areas 30 moving, standard30 areas 25 25 20 20 15 15 10 10 Saccade [s] Saccade Saccade [s] Saccade 5 5 0 0 -5 -5 1 4 7 10 13 161 19 4 22 7 25 10 28 13 31 16 34 19 37 22 40 25 43 28 46 31 49 34 52 37 55 40 58 43 61 46 64 49 67 52 70 55 58 61 64 67 70 Observation number Observation number Figure 11. Sequence of saccades in the damaged and undamaged stands during felling (A), processing (B), moving (C).

Forests 2021, 12, x FOR PEER REVIEW 11 of 16

Figure11. SequenceForests of2021 saccades, 12, 168 in the damaged and undamaged stands during felling (A), processing (B), moving (C). 11 of 16

The mean difference between the duration of subsequent saccades for all of the ana- lysed activities amountedThe mean to 31% difference in the between undamaged the duration control of subsequentstands, while saccades in the for damaged all of the anal- stands it was 94%.ysed The activities variability amounted of the to 31% sums in the of undamagedsaccade duration control stands, observed while during in the damaged the processing of successivestands it was trees 94%. was The determined variability of in the accordance sums of saccade with duration the formula observed no. during 1. Their the pro- high differentiationcessing between of successive the analysed trees was determinedstands could in accordance be observed with especially the formula in no. the 1. Their case high of processing indifferentiation post-disaster between stands the(Spp analysed = 207%). stands For couldprocessing, be observed the presence especially of in signif- the case of processing in post-disaster stands (Spp = 207%). For processing, the presence of significant icant second-ordersecond-order and fourth-order and fourth-order autocorre autocorrelationslations was was also also established, established, which which provesproves that a that a longer saccadelonger was saccade clearly was clearlyfollowed followed by a byshorter a shorter one one (Figure (Figure 12). 12).

FigureFigure 12. Autocorrelation 12. Autocorrelation function offunction the time of series the oftime saccades series observed of saccades during observed processing during in the damagedprocessing stands. in the damaged stands. Harvester runs were characterised by similar variability of saccade cycles on both plots: Sms = 55% in the undamaged control stands and Smp = 50% in the damaged stands. Harvester runsThe variability were characterised of saccade cycles by calculatedsimilar variability for tree cutting of saccade in the undamaged cycles on stands both was plots: Sms = 55%higher in the (Scp undamaged = 25%) than control for cutting stands off root and plates Smp in= the50% post-disaster in the damaged stands stands. (Scs = 15%). The variability of saccade cycles calculated for tree cutting in the undamaged stands was higher (Scp = 25%)4. Discussion than for cutting off root plates in the post-disaster stands (Scs = 15%). The evaluation of any harvesting technology should take into account economic, 4. Discussion ecological and ergonomic factors [34–37]. Productivity is no longer regarded as the main evaluation criterion, and the ergonomic assessment of work tasks is gaining special impor- The evaluationtance. of Years any ago,harvesting such assessment technolo concernedgy should mainly take the into physiological account economic, workload associated eco- logical and ergonomicwith manual factors labour. [34–37]. Today, Productivi the affirmationty is no of automatedlonger regarded technologies as the has main removed eval- hard uation criterion,manual and the work ergonomic and has transformed assessment much of of anwork operator’s tasks taskis gaining into supervising special theim- opera- portance. Years tionago, of such a system. assessment Thus, ergonomic concerned analyses mainly currently the physiological include also the workload tracking ofasso- changes in the human psyche because modern forms of work require complex thought processes ciated with manualand entail labour. a significant Today, cognitive the affir workload.mation of automated technologies has re- moved hard manualThe work introduction and has oftransformed harvesters and mu forwardersch of an operator’s in forestry meanttask into that supervis- their operators ing the operationdeveloped of a system. forms Thus, of fatigue ergonomi typical ofc mentalanalyses work currently and characterised include byalso a reduced the track- capacity ing of changes forin concentration,the human psyche impaired because thinking, modern slowed and forms weakened of work perception, require decreased complex motiva- thought processestion and to complete entail a thesignificant job, emotional cognitive disorders, workload. focus on rest. Grzywi´nskiand Hołota [27], The introduction of harvesters and forwarders in forestry meant that their operators developed forms of fatigue typical of mental work and characterised by a reduced capac- ity for concentration, impaired thinking, slowed and weakened perception, decreased mo- tivation to complete the job, emotional disorders, focus on rest. Grzywiński and Hołota

Forests 2021, 12, 168 12 of 16

Sullman and Gellersted [28], Berger [29], who studied the mental workload of forest work- ers, pointed out that the job of the harvester operator involves mostly mental effort with very high levels of mental stress. Häggström et al. [38] noted also a high level of automatic- ity in the work of harvester operators, including the recognition of familiar, small objects without significant visual involvement [39]. The visual standard for the cognition of reality, namely eye tracking, is often used in neurobiology, psychology, marketing and computer science [40,41]. In forestry, the eye tracking study technique has been introduced in the last few years only [38,42]. Among other things, such studies probe the significance of recognising objects, planning tasks and establishing priorities, as well as top-down and bottom-up cognitive processes [43]. For simple and clearly defined tasks there is a high correlation between eyeball movement and mental stress [44,45]. The making and changing of decisions are closely linked to changes in eyeball movement patterns [46–48]. Longer fixations can be associated with a more difficult cognitive process occurring during observation of a large number of objects or a phenomenon that is not fully rec- ognizable [49]. Rayner [40] drew attention to the variability of information obtained by means of eyesight, and thus to the relativity of such assessments. In the current experiment, clear differences in the duration of eyesight fixations indicate probably a different nature of the operator’s work in the standard and the damaged stands. Although the average duration of fixations was longer under the supposedly easier work conditions offered by the undamaged stand, the data collected in the damaged area contained many more outliers—i.e., much longer fixations. These probably occurred while maneuvering the machine and the harvester head for the purpose of allowing trouble-free and safe cutting of difficult-to-reach windthrown trees. Very similar results were obtained by Szewczyk et al. [42], who analysed the work of a harvester operator on steep slopes: performing tasks in difficult terrain also involved fixations that were longer and had greater variability of duration than recorded for the same work and operator on gentle terrain. In another simi- lar experiment, Häggström et al. [38] compared the fixation times of a harvester operator working in a thinning and a mature stand. Here, the occurrence of longer fixations was observed in thinning stands as compared to mature stands, where work is more dangerous and more valuable assortments are obtained. All these studies concur in associating longer fixation times with the easier work entailed in routine tasks, when a visual model of task execution is not created at the beginning of the task, but is constructed on an ongoing basis [50]. Conversely, complicated vision scenes determine shorter fixations, as already indicated by Duchowski [41] and Molnar [51]. According to Rayner and Castelhano [52], the mean duration of a fixation is 260–330 ms., which is slightly shorter than in our research (between 400 and 500 ms). As found in our study, the occurrence of periodic extensions of fixation duration up to 1.5 s in the case of tasks characterised by high intensity is consistent with the results provided by Steinman [53]. Saccadic eyeball movements reflect changes in mental load, and thus in the level of stress [45,54]. Increased load conditions shorten the reaction time and cause a rapid shifting of the eyesight to another part of the vision scene. Again, that is the case of the studies by Szewczyk et al. [42] for a harvester operator working on steep terrain, and by Häggström et al. [38] for a harvester operator engaged with final cutting. The present study aligns with those findings, reporting a shorter (about 14%) duration of saccades in the damaged stand, compared with the undamaged stand. The greatest differences between the duration of saccades in the analysed stands were observed between tree cutting (undamaged stands) and cutting off root plates (damaged stands). In fact, while these two operations might be considered functionally equivalent (as they both aim at separating the stem from the roots), they are inherently different in their technical execution and the latter is undoubtedly more difficult than the former. On both research plots, the shortest saccades were observed in the case of processing. As observed in our experiment, the pattern of visual scanning of the terrain with areas of interest located on the harvester head (up to 50% of saccades), is consistent with other analyses. In the study by Häggström et al. [38] most saccades (58%) occurred between Forests 2021, 12, 168 13 of 16

fixations located on the of the harvester head. This amounted to 23% more than the mean from the entire processing of the wood. That could indicate that the maximum mental stress of the harvester operator occurs during processing, when a considerable cognitive effort is required in an attempt to maximize value recovery. Decisions made during processing strongly impact operation profitability and are often crucial to financial success. The greatest dynamics of changes in the length of saccades, visible in our experiment, and their clear cyclic pattern characterise well the way in which that operation is performed. Most likely, shorter saccades are associated with the actual cross-cutting, while longer ones relate to the search for new tasks. Processing is the only fully automated operation performed by the operator. Interestingly, it is this operation that requires the greatest mental strain and generates the highest level of stress, likely due to its crucial impact on value recovery and financial gain. The high psychological tension observed especially in the post-disaster areas was probably what generated the variable length of the total duration of saccades observed during the processing of successive trees. Longer saccade durations following shorter ones were probably related to the operator’s rest during activities that are not formal breaks, just after work burdened with considerable psychological tension. What is visible here is a system of specific breaks concealed within work, as described by Rosner [55] and Grandjean [56]. The long duration of saccades when the machine was travelling, especially in the undamaged stands, are consistent with the findings of Häggström et al. [38]. New findings that emerged as a result of our research concerned the scanning of the working space (the corridor and the stand). In the damaged stands, scanning was mostly restricted to the access corridor, probably due to more difficult working conditions. When the machine was travelling, the operator focused on the route and did not collect information about the stand and the next work station to the same extent as when working in the undamaged stand. The work pattern in which the harvester operator plans subsequent tasks before completing the current ones speeds up the work and makes it smoother [50]. According to Traschütz et al. [57] peripheral vision is used by the machine operator continuously in order to assess the machine’s position and speed when travelling. This phenomenon was also evident in our experiment when the harvester was moving in the undamaged stands, in which half of the eyesight fixations during travelling included scanning of the surrounding stand.

5. Conclusions Conducted with the eye tracking method, an analysis of the visual information reach- ing harvester operators can help probing the cognitive strain experienced under variable conditions. Comparing the duration and the variability of eyesight fixations and saccades may enable a fair assessment of the cognitive workload imposed on a harvester operator by work in damaged and undamaged stands. Such indicators point at the significantly higher cognitive demand required by work in damaged stand. High mental loads borne by opera- tors working on post-disaster plots affect the way they work while processing subsequent trees. They compensate for the high stress by cyclically unintentionally slowing down the most strenuous activities. Furthermore, within a complex task, eye tracking helps emerging those specific task elements where cognitive demand is higher, thus indicating where remedial action should be prioritized. That way, new and adapted work standards could be developed that take into account the challenging work conditions offered by windthrown stands. Development of such new standards may have a strategic value, given the expected rapid increase of storm damage events across Europe. An objective evaluation of cognitive stress may also help assessing harvester operators’ mental resilience, and predicting long- term productivity under especially stressful work conditions. Thanks to the optimisation of work techniques, this knowledge may also be used in harvester operator training.

Author Contributions: Conceptualization, G.S. and P.T.; methodology, G.S. and P.T.; software, B.M.; data curation, K.A.; validation, G.S., P.T. and D.K.; writing of first manuscript draft, G.S., R.S. and N.M.; review and editing of manuscript versions, G.S., P.T., D.K., R.S. and N.M.; visualization, B.M. All authors have read and agreed to the published version of the manuscript. Forests 2021, 12, 168 14 of 16

Funding: Funded by a subsidy of the Ministry of Education and Science for the University of Agriculture in Krakow for the year 2020. Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to analyzes related to the way work is performed by third parties. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Pickett, S.T.A.; White, P.S. The Ecology of Natural Disturbance and Patch Dynamics; Academic Press: Cambridge, MA, USA, 1985. 2. Ziembli´nska,K.; Urbaniak, M.; Merbold, L.; Black, T.A.; Jagodzi´nski,A.M.; Herbst, M.; Qiu, C.; Olejnik, J. The carbon balance of a Scots pine forest following severe wind throw: Comparison of techniques. Agr. Forest Meteorol. 2018, 260–261, 216–228. [CrossRef] 3. Schelhaas, M.; Nabuurs, G.; Schuck, A. Natural disturbances in the European forests in the 19th and 20th centuries. Glob. Change Biol. 2003, 11, 1620–1633. [CrossRef] 4. Don, A.; Bärwolff, M.; Kalbitz, K.; Andruschkewitsch, R.; Jungkunst, H.F.; Schulze, E.D. No rapid soil carbon loss after a windthrow event in the High Tatra. Forest Ecol. Manag. 2012, 276, 239–246. [CrossRef] 5. Zaj ˛aczkowski,J. Odporno´s´cLasu na Szkodliwe Działanie Wiatru i Sniegu´ (Forest Resistance to the Harmful Effects of Wind and Snow); Wydawnictwo Swiat:´ Warszawa, Poland, 1991. 6. Gardiner, B.A.; Quine, C.P. Management of forests to reduce the risk of abiotic damage—A review with particular reference to the effects of strong winds. For. Ecol. Manag. 2000, 135, 261–277. [CrossRef] 7. Panferov, O.; Doering, C.; Rauch, E.; Sogachev, A.; Ahrends, B. Feedbacks of windthrow for Norway spruce and scots pine stands under changing climate. Environ. Res. Lett. 2009, 4, 1–10. [CrossRef] 8. Frank, D.; Reichstein, M.; Bahn, M.; Thonicke, K.; Frank, D.; Mahecha, M.D.; Smith, P.; Velde, M.; Babst, F.; Beer, C.; et al. Effects of climate extremes on the terrestrial carbon cycle: Concepts, processes and potential future impacts. Glob. Change Biol. 2015, 21, 2861–2880. [CrossRef] 9. Leckebusch, G.C.; Koffi, B.; Ulbrich, U.; Pinto, J.G.; Spangehl, T.; Zachsarias, S. Analysis of frequency and intensity of European winter storm events from a multi-model perspective, at synoptic and regional scales. Clim. Res. 2006, 31, 59–74. [CrossRef] 10. Beniston, M.; Stephenson, D.B.; Christensen, O.B.; Ferro, C.A.T.; Frei, C.; Goyette, S.; Halsnaes, K.; Holt, T.; Jylha, K.; Koffi, B.; et al. Future extreme events in European climate: An exploration of regional climate model projections. Clim. Chang. 2007, 81, 71–95. [CrossRef] 11. Lorz, C.; Fürst, C.; Galic, Z.; Matijasic, D.; Podrazky, V.; Potocic, N.; Simoncic, P.; Strauch, M.; Vacik, H.; Makeschin, F. GIS-based probability assessment of natural hazards in forested landscapes of Central and South-Eastern Europe. Environ. Manag. 2010, 46, 920–930. [CrossRef] 12. Ikonen, V.P.; Kilpeläinen, A.; Zubizarreta-Gerendiain, A.; Strandman, H.; Asikainen, A.; Venäläinen, A.; Kaurola, J.; Kangas, J.; Peltola, H. Regional risks of wind damage in boreal forests under changing management and climate projections. Can. J. For. Res. 2017, 47, 1632–1645. [CrossRef] 13. Bilici, E.; Andiç, G.V.; Akay, A.E.; Sessions, J. Productivity of a portable winch system used in salvage of storm-damaged timber. Croat. J. For. Eng. 2019, 40, 311–318. [CrossRef] 14. Szewczyk, G.; Sta´nczykiewicz, A. Model szacowania pracochłonno´scipozyskiwania drewna w drzewostanach ze ´sniegołomami (A model for estimating labour requirements during timber harvesting in snowbreak stands). Le´snePr. Badaw. 2012, 2, 167–173.

15. Moskalik, T.; Borz, S.A.; Dvoˇrák, J.; Ferencik, M.; Glushkov, S.; Muiste, P.; Lazdin, š, A.; Styranivsky, O. Timber Harvesting Methods in Eastern European Countries: A Review. Croat. J. For. Eng. 2017, 2, 231–241. 16. Frutig, F.; Fahrni, F.; Stettler, A.; Egger, A. Mechanisierte Holzernte in Hanglagen. Wald Und Holz 2007, 4, 47–52. 17. Dvoˇrák, J.; Bystrický, R.; Hošková, P.; Hrib, M.; Jarkovská, M.; Kováˇc,J.; Krilek, J.; Natov, P.; Natovová, L. The Use of Harvester Technology in Production Forests; Folia Forestalia Bohemica: Kostelec nad Cernˇ ýmilesy, Czech Republic, 2011. 18. Szewczyk, G.; Sowa, J.M.; Grzebieniowski, W.; Kormanek, M.; Kulak, D.; Sta´nczykiewicz,A. Sequencing of harvester work during standard cuttings and in areas with windbreaks. Silva. Fenn. 2014, 48, 1–16. [CrossRef] 19. Szewczyk, G.; Sowa, J.M.; Michalec, K.; Gaj-Gielarowec, D.; Gielarowiec, K. Salvage condition assesment of timber volume in disturbed areas. Balt. For. 2017, 23, 619–625. 20. Bort, U.; Mahler, G. Kranvollernter einsatz bei der sturmholz aufarbeitung. AFZ Wald 1990, 14–15, 366–368. 21. Bort, U.; Mahler, G.; Pfeil, C. Sturmholz auf arbeitung mit Kranvollernten. AFZ Wald 1990, 25–26, 640–641. 22. Brzózko, J. Pozyskiwanie drewna z obszarów pokl˛eskowych—Czynnikiryzyka i sposoby zwi˛ekszaniabezpiecze´nstwapracy (Wood harvesting from calamity areas risk factors and methods of work safety increasing). Tech. Rol. Ogrod. Le´sna 2009, 1, 10–12. 23. Brzózko, J. Metoda Prognozowania Wydajno´sciMaszynowego Pozyskiwania Drewna Pokl˛eskowegona Podstawie cech Uszkodzonej Powierzchni Le´snej(The Method of Forecasting of Windthrown Wood Machine Harvesting Performance Based on the Characteristics of the Damaged Forest Area); Wydawnictwo SGGW: Warszawa, Poland, 2014. 24. Brzózko, J.; Kaluga, T. Investigations on technological process of after-calamity site preparation to logging with the harvester. Ann. Wars. Univ. Life Sci. SGGW Agric. 2010, 56, 79–87. Forests 2021, 12, 168 15 of 16

25. Heinimann, H.R. Forest operations engineering and management—The ways behind and ahead of a scientific discipline. Croat. J. For. Eng. 2007, 28, 107–121. 26. Spinelli, R.; Magagnotti, N.; Labelle, E.R. The effect of new silvicultural trends on mental workload of harvester operators. Croat. J. For. Eng. 2020, 41, 177–190. [CrossRef] 27. Grzywi´nski,W.; Hołota, R. Subjective assessment of the fatigue of forest workers based on japanese questionnaire. Acta Sci. Pol. 2006, 5, 27–37. 28. Sullman, M.; Gellerstedt, S. The mental workloads of mechanized harvesting. N. Z. For. 1997, 11, 48. 29. Berger, C. Mental stress on harvester operators. In Proceedings of the Austro2003: High Tech Forest Operations for Mountainous Terrain, Schlaegl, Austria, 5–9 October 2003. 10p. 30. Stampfer, K. Stress and Strain Effects of Forest Work in Steep Terrain. In Proceedings of the IUFRO/FAO Seminar on Forest Operations in Himalayan Forests with Special Consideration of Ergonomic and Socio-Economic Problems, Thimphu, Bhutan, 20–23 October 1997; Heinimann, H.R., Sessions, J., Eds.; Kassel University Press: Kassel, Germany, 1998; pp. 113–119. Available online: http://www.uni-kassel.de/upress/online/inhalt/978-3-933146-12-0.inhalt.pdf (accessed on 30 January 2021). 31. Gomolka, Z.; Kordos, D.; Zeslawska, E. The application of flexible areas of interest to pilot mobile eye tracking. Sensors 2020, 20, 986. [CrossRef] 32. Olsen, A. The Tobii I-VT fixation filter algorithm description. Tobii Technol. 2012, 21, 1–21. Available online: https://www.tobiipro. com/siteassets/tobii-pro/learn-and-support/analyze/how-do-we-classify-eye-movements/tobii-pro-i-vt-fixation-filter.pdf (ac- cessed on 17 April 2020). 33. Box, G.E.P.; Jenkins, G.M.; Reinsel, G.C. Time Series Analysis: Forecasting and Control; Wiley: Hoboken, NJ, USA, 2008. 34. Giefing, D.F. Badania nad opracowaniem proekologicznych procesów pozyskiwania drewna. In Model Optymalnych dla Srodowiska´ Procesów Pozyskiwania Drewna; IBL: Warszawa, Poland, 1995; pp. 52–59. 35. Paschalis, P. Kryteria zrównowazonej˙ gospodarki le´snejw uzytkowaniu˙ lasu (Sustainable forest management criteria in forest utilisation). Prz. Tech. Rol. Le´snej 1997, 62, 22–29. 36. Skoupý, A. Sophisticated model for nature-friendly timber haulage evaluation. In Technology and Ergonomics in the Service of Modern Forestry; Starzyk, J., Ed.; Publishing House of the University of Agriculture: Krakow, Poland, 2011; pp. 241–251. ISBN 978-83-60633-51-9. 37. Kühmaier, M.; Stampfer, K. Development of a multi-criteria decision support for energy wood supply management. Croat. J. For. Eng. 2012, 33, 181–198. 38. Häggström, C.; Englund, M.; Lindroos, O. Examining the gaze behaviors of harwester operators: Aneye-tracking study. Int. J. For. Eng. 2015, 2, 96–113. [CrossRef] 39. Land, M.; Mennie, N.; Rusted, J. The roles of vision and eye movements in the control of activities of daily living. Perception 1999, 28, 1311–1328. [CrossRef] 40. Rayner, K. Eyemovements in reading and information processing: 20 years of research. Psychol. Bull. 1998, 124, 372–422. [CrossRef][PubMed] 41. Duchowski, A. Eye tracking methodology. In Theory and Practice; Springer: London, UK, 2007. 42. Szewczyk, G.; Spinelli, R.; Maganotti, N.; Tylek, P.; Sowa, J.M.; Rudy, P.; Gaj-Gielarowiec, D. The mental workload of harvester operators working in steep terrain conditions. Silva. Fenn. 2020, 54, 1–18. [CrossRef] 43. Schütz, A.C.; Braun, D.I.; Gegenfurtner, K.R. Eye movements and perception: A selective review. J. Vis. 2011, 11, 1–30. [CrossRef] [PubMed] 44. Hayhoe, M.M.; Rothkopf, C.A. Vision in the natural world. Wiley Interdisciplinary Reviews: Cognitive Sci. 2011, 2, 158–166. [CrossRef][PubMed] 45. A¸stefănoaei, C.; Creangă, D.; Pretegiani, E.; Optican, L.M.; Rufa, A. Dynamical complexity analysis of saccadic eye movements in two different psychological conditions. Rom. Rep. Phys. 2014, 66, 1038–1055. [PubMed] 46. Calvo, M.G.; Lang, P.J. Gaze patterns when looking at emotional pictures: Motivationally biased attention. Motiv. Emotion. 2004, 28, 221–243. [CrossRef] 47. Patalano, A.L.; Juhasz, B.J.; Dicke, J. The relationship between decisiveness and eye movement patterns in a decision making informational search task. J. Behav. Decis. Making. 2010, 23, 353–368. 48. Przybyło, J.; Ka´ntoch,E.; Augustyniak, P. Eyetracking-based assessment of affect-related decay of human performance in visual tasks. Future Gener. Com. Sys. 2018.[CrossRef] 49. Holmqvist, K.; Nyström, M.; Andersson, R.; Dewhurst, R.; Jarodzka, H.; Van de Weijer, J. Eyetracking: A Comprehensive Guide to Methods and Measures; Oxford University Press: New York, NY, USA, 2011. 50. Ballard, D.H.; Hayhoe, M.M.; Li, F.; Whitehead, S.D.; Frisby, J.; Taylor, J.; Fisher, R. Hand-eye coordination durings equential tasks. Philos. T. Roy. Soc. B 1992, 337, 331–339. 51. Molnar, F. About the Role of Visual Exploration in Aesthetics. In Advances in Intrinsic Motivation and Aesthetics; Day, H., Ed.; Plenum Press: New York, NY, USA, 1981. 52. Rayner, K.; Castelhano, M.S. Eye Movements during Reading, Scene Perception, Visual Search, and While Looking at Print Advertisements; Pieters, R., Wedel, M., Eds.; Erlbaum: Mahwah, NJ, USA, 2007. 53. Steinman, R.M. Gaze control in natural conditions. In The Visual Neuroscience; Werner, J.S., Chalupa, L.M., Eds.; The MIT Press: Cambridge, MA, USA, 2003; Volume 2. Forests 2021, 12, 168 16 of 16

54. Tokuda, S. Using saccadic eye movements to measure mental workload. J. Vision. 2008, 8, 86. [CrossRef] 55. Rosner, J. Ergonomia (Ergonomics); Pa´nstwoweWydawnictwo Ekonomiczne: Warszawa, Poland, 1985. 56. Grandjean, E. Physiologische arbeitsgestaltung. Zeitschrift für Präventivmedizin 1958, 3, 253–260. [CrossRef] 57. Traschütz, A.; Zinke, W.; Wegener, D. Speed change detection in foveal and peripheral vision. Vis. Res. 2012, 72, 1–13. [CrossRef] [PubMed]