This is an exploratory and descriptive study using anthropometrical measurements, followed by a three-dimensional motion capture, synchronized with ground-reaction forces taken from multiple force platforms in standardized laboratory conditions. The combination of the techniques allowed for a complete, whole-body, biomechanical analysis of the pop-up motion. This study was conducted according to the Helsinki Declaration and approved by the local ethics committee (project number: CEFAD 27.2014). Written informed consent was obtained from all participants involved in the study.
2.2. Experimental Procedures
Prior to the pop-up analysis, the surfers’ anthropometric measurements and body-composition analysis were obtained, both while in the standing position. Body composition was assessed through multifrequency bioimpedance analysis using the InBody 230 (Biospace Co., Ltd., Seoul, Korea). Together, these tests yielded body mass index (BMI), percentage of skeletal muscle mass (SMM%), percentage of body fat mass (BFM%), and waist-to-hip ratio (WHR).
A 12-camera digital motion-capture system (mocap) (Qualisys, Gothenburg, Sweden) was used to record three-dimensional movement of the participants at a 200 Hz sampling frequency. A volume of approximately 45 m
3 (5 m long, 3 m wide, and 3 m deep) was calibrated using an L-shaped reference structure and wand according to the manufacturer’s recommendations (0.7 mm standard deviation error calibration mean). Spherical retro-reflective markers were attached to the skin by double-faced adhesive tape, and clusters were fastened with an elastic strap. The full-body model consisted of 46 markers at the following anatomical locations (
Figure 1): right anterior head, left anterior head, right posterior head, left posterior head, right acromion, left acromion, right lateral epicondyle, left lateral epicondyle, right medial epicondyle, left medial epicondyle, right styloid process of ulna, left styloid process of ulna, right styloid process of radius, left styloid process of radius, right second metacarpal head, left second metacarpal head, right third metacarpal head, left third metacarpal head, right fifth metacarpal head, left fifth metacarpal head, right scapula inferior angle, left scapula inferior angle, right sternum, left sternum, seventh cervical spinal process, tenth thoracic spinal process, right anterior superior iliac spine, left anterior superior iliac spine, right posterior superior iliac spine; left posterior superior iliac spine, right medium iliac spine, left medium iliac spine, right medial knee (medial epicondyle of the knee), left medial knee (medial epicondyle of the knee), right lateral knee (lateral epicondyle of the knee), left lateral knee (lateral epicondyle of the knee), right lateral malleolus, left lateral malleolus, right medial malleolus, left medial malleolus, right first metatarsal head, left first metatarsal head, right fifth metatarsal head, left fifth metatarsal head, right calcaneus, and left calcaneus.
Ground-reaction-force data were obtained using four force plates (
Figure 1c). Two of the force plates, FP1 and FP2, were 60 × 40 cm
2; while the other two, FP3 and FP4, were 60 × 90 cm
2 (Bertec, Columbus, OH, USA). The force plates were mounted flush with the laboratory floor, and ground-reaction-force data were acquired at 1000 Hz. The two-dimensional outline of a typical short surfboard (74 inches in length) was created on the laboratory floor using tape. The image was constructed such that each of the four force plates were positioned in a quadrant of the board and could register the force generated by each upper and lower limb in action during the pop-up motion.
The surfers were instructed to familiarize themselves with the representative surfboard and force plates and to adjust their prone position according to their personal preferences. The researcher then explained the procedures for simulating the pop-up movement freely, while respecting the surfboard dimensions on the floor. First, the surfers were asked to simulate 3–4 paddling movements and then performed the pop-up. If the surfer’s hands were not entirely on the force plates we asked the surfer to move fore or aft along the surfboard. This adjustment did not impose any influence on the movement. Surfers were encouraged to direct their gaze forward as if they were surfing. Following this familiarization period, surfers were allowed to perform the entire pop-up motion freely, without any restriction from the researcher (
Figure 2).
2.3. Data Processing and Analysis
Qualisys Track Manager–QTM version 2.2 (Qualisys, Gothenburg, Sweden) software was used to acquire the 3D kinematic and ground-reaction-force data simultaneously and synchronized. Each reflective track marker was identified using the respective anatomical reference label. The marker reconstruction accuracy reached 100%. After this treatment, the data files were exported using a public-domain binary file format, C3D (Coordinate 3D–C3D.org), which stores 3D data and their associated parameters (i.e., 3D ground reaction forces) in a single file.
Visual 3D Professional version 6.0 (C-motion Inc., Rockville, MD, USA) software was used to process the kinematic and ground-reaction-force data from the C3D files. A full-body biomechanical model was created. All body segments were created based on the markers fixed at the anatomical points. A bilateral marker set and two tracking markers were actually provided for each segment: head, thorax/ab, pelvis, upper-arms, fore-arms, hands, thighs, shank, and feet. The inverse kinematics algorithm was selected to compute the model according to recommendations from the C-motion Visual 3D documentation and applied in each file, taking into account each surfer’s body weight.
The beginning of the pop-up was defined as the point when at least one hand touched the surfboard (FP1 or FP2), right after the simulated sprint paddling. Thereafter, the data-mining process and visual inspection of the motion technique analysis reveal three distinct serial and complementary phases: (i) the push-up phase is defined as the time between first hand contact with the surfboard, followed by the push-up movement, up to the point when both hands leave the surfboard; (ii) the transition phase, which can be performed in two distinct ways: (1) Wipe-transition—the surfer experiences a flight phase (no contact) or immediately after their hands leave the surfboard, one or both of their feet contacts the surfboard—the duration counts the time between hands leave the surfboard until any foot touch the surfboard; (2) Overlap-transition—the surfers’ hands remain in contact with the surfboard while one or both feet contact the surfboard—the duration starts to count when any foot touches the surfboard until both hands leave the surfboard; and (iii) the reaching phase refers to the period between either foot touching the surfboard and the surfer achieving stabilization of their own weight on the surfboard. The stabilization of the weight was defined as the instant when the surfers reached their exact body weight, measured by FP3 and FP4. The reaching phase should result in the surfers’ feet individually positioned front foot on FP3 and back foot on FP4 (
Figure 1c and
Figure 2f). If this did not happen, the surfer repeated the action, making adjustments until his technique produced the desired motion. The stabilization time defined the end of the pop-up movement.
A specific pipeline script command for Visual 3D was created to identify events that define the pop-up phases, and to extract the related parameters to be analyzed. The principal time-related events and parameters were: (a) hands touch FP1 and FP2; (b) distance between hands using the lateral markers positioned in the metacarpus of the little fingers; (c) pushing-up peak-force result from FP1–2, and the respective elbows angles in this instant; (d) instant when the hands leave FP1–2; (e) feet touched FP3–4; (f) flight phase, the period within the wipe-transition where the surfer has no contact with the surfboard; (g) reaching peak-force resulting from FP3–4 and the respective knee angles; and (h) weight stabilization and the feet base distance (using the lateral markers positioned on the metatarsus of the 5th toes).
The ground-reaction-force values in individual time curves were normalized by individual body mass (N/kg). To calculate the impulse and impact absorption during each phase, the time integral of the force/time curves were calculated.
Data processing generated outputs in text files for statistical analysis. The principal events generated were: hand-touch time, push-peak force (independent and both hands), hand-push-out time, reach time, reach-peak force (independent and both feet), and stabilization. A script was written to automatically extract the events. Events that were not automatically detected were identified manually using force-plate data. We expected the operator variability to be minimal for these cases because the force data were high resolution and events were clear and easy to identify. All the events registered (both manual and automated) were visually inspected to confirm accuracy.
The kinematic data were interpolated using the Piecewise Cubic Hermite Interpolating Polynomial method with 100 samples, to ensure that each trial included the same number of samples. A symmetry function (SF) with temporal dependence was then applied to identify the percent difference between the right X
R(t) and left X
L(t) sides, for elbow and knee angle, relative to the respective average range of motion (ROM) [
18]. The positive or negative signal indicated the leading side. A score close to zero indicated symmetry (equality) between limbs. Magnitudes of 10% were typically reported in symmetrical populations when other movements were analyzed [
18]. Therefore, an asymmetry of 15% or more is thought to indicate substantial asymmetry [
18].
2.4. Statistical Analysis
All data were analyzed using the statistical software Statistica 12 (StatSoft©, Tulsa, OK, USA) and Excel 2016 (Microsoft Corp., Redmond, WA, USA) with p ≤ 0.05 significance level. All variables were reported with descriptive statistics (mean and standard deviation). For the temporal variables, descriptive statistics were used for analysis, and the pop-up subphases were presented relative to the total duration of the gesture. A significance level (α) of 0.05 was used for the inferential tests. Factorial ANOVA was used to investigate differences in kinematic parameters (elbow and knee angles during the peak-forces) between hand-dominance and stance-feet-base. A one-way ANOVA was conducted to verify the existence of differences between the type of transition and the performance of the pop-up. A dependent-samples t-test was used to identify differences between individual hand peak-forces during the push-up phase, and also for the reaching phase comparing front and rear feet peak-force distribution. A multiple linear regression model was applied to analyze the influence of body characteristics (skeletal muscle mass—SMM%; body fat mass—BFM%; body height) on the performance of the pop-up technique (duration). Finally, another multiple linear regression model was used to analyze the influence of the performance of the pop-up technique (duration) on the kinetic variables (ground-reaction-force peak during the pushing and reaching phases).