Kinematic data were low pass filtered

Kinematic data were low-pass filtered (15Hz cutoff; Matlab R2012a), ground reaction force data were low-pass filtered (50Hz cutoff), and internal joint moments were computed using matched filter cutoffs (15Hz). EMG data were band pass filtered (15–300Hz cutoff), full-wave rectified, and low-pass filtered (15Hz cutoff). We analyzed data across TASIN-1 landing phase (from GRFz>20N to the time vertical center of mass velocity reached zero). We expressed joint angles in degrees, joint moments were normalized to bodyweight and subject height (BW•H), and EMG data were normalized to mean dynamic baseline activity recorded in the experimental condition with the lowest task demands (BW•H12.5) [2]. Prior to principal component analyses (PCA), we temporally normalized the landing phase to 101 points using cubic spline interpolation (Matlab R2012a). We then converted each time series waveform to z-scores by subtracting the subject׳s baseline mean and dividing by the baseline standard deviation (BW•H12.5). For each of the 12 lower extremity variables, we assembled 1026×101 dimension matrices (19 subjects×3 loads×2 heights×9 trials=1026; times series length=101). Principal components (PCs) explaining greater than 90% cumulative explained variance (EV) were extracted from the covariance matrix of each variable, and PC scores were computed for each trial and PC (Matlab R2012a). PC scorea means were aggregated for each subject from the 9 analyzed trials and used for inferential testing. Separate 3×2 (load×height) repeated measures analyses of variance (ANOVAs) were conducted for each extracted PC and used in evaluating movement pattern changes across conditions (IBM SPSS Statistics Version 20) [3,4]. We performed follow-up one-way repeated measures ANOVAs, simple main effects analysis, degree of freedom Huynh-Feldt corrections, and Bonferroni corrections during pairwise comparisons, as necessary [5]. Ensemble curves (aggregated among subjects) were plotted by condition for each lower extremity variable (Figs. 1–6, top). Extracted PCs were plotted below ensemble plots in order of descending EV. PC score means were presented beside each PC loading vector, indicating statistically significant condition differences where appropriate (α=0.05). Load and landing height PC score means were presented in each condition following significant interaction. Height PC score means were presented following a significant height main effect and load PC score means were presented following a significant load main effect. In each case, significant ANOVA results were included above PC score means, with significant pairwise comparisons highlighted (α=0.05). Hip angles and moments were show in Fig. 1, knee angles and moments were shown in Fig. 2, ankle angles and moments were shown in Fig. 3, GRFz and gluteus maximus were shown in Fig. 4, biceps femoris and vastus medialis were shown in Fig. 5, and medial gastrocnemius and tibialis anterior muscles were shown in Fig. 6. Significant pairwise differences and variable trends across conditions were summarized in Table 1 of Nordin and Dufek [1].
Data Aerodynamic raw data measured during the wind tunnel campaign were rotor torque (Qaero), rotor thrust (both in the longitudinal direction TX and in the transversal one TY), rotor rotational speed () and wind tunnel speed (). These data are here presented in a convenient form typically adopted in wind turbine engineering. As a matter of fact, to provide more insights on rotor behavior, aerodynamic torque (CQ,aero) and thrust (CTX and CTY) coefficients are provided in Tables 1–4. as a function of the equatorial Tip Speed Ratio (TSReq) computed at rotor equatorial diameter. In order to disclose the influence of the blade Reynolds number (Re) on aerodynamic torque, power and thrust coefficients, Figs. 2–6 show a comparison between the data obtained at two rotor angular velocities, respectively 200rpm (Re=1.38×105) and 300rpm (Re=2.05×105).