A comparison of methods for analyzing drop

jump performance
BACA ARNOLD

Medicine & Science in Sports & Exercise. 31:p 437-442, March 1999.

Author Information

Department of Biomechanics, IfS, University of Vienna, Vienna, AUSTRIA

Submitted for publication April 1997.

Accepted for publication June 1998.

Address for correspondence: Dr. Arnold Baca, Department of

Biomechanics, IfS, University of Vienna, Auf der Schmelz 6, A-1150 Wien,
Austria. E-mail: arnold.baca@univie.ac.at.

Abstract

A comparison of methods for analyzing drop jump performance. Med. Sci.

Sports Exerc., Vol. 31, No. 3, pp. 437-442, 1999.

Back to Top

Purpose:

Drop jumping is a popular form of plyometric training. Different techniques

are applied to determine parameter values quantifying drop jumps, such as
the jump height or the durations of the phases of downward and upward
movements of the center of mass (CM) during foot contact with the ground
after dropping. The flight-time method estimates the jump height from the
time between the instant of leaving the ground and the instant of landing. In
video-based methods, markers are placed on the skin of the subject to
define the positions of the body segments. The time-dependent positions of
the CM and parameter values are then calculated utilizing models of the
human body. If the vertical velocity of the CM can be estimated at one
instant, the parameter values can be calculated from the vertical ground
reaction forces.

Back to Top

Methods:

The purpose of this study was to find out which technique yields the lowest
errors compared with the results obtained by the double force plate
technique. In this investigation, two force plates were used, one located
under the drop platform. Twenty-five drop jumps were analyzed with eight
different methods. There were large differences between the reference
method and other methods. Using the height of the drop platform (0.39 m)
to estimate the velocity at the end of the free fall, in conjunction with data
from one force plate, resulted in a mean difference of 4.2% (SD: 9.6%) in
the calculated jump height. Using video information to estimate the time
that the velocity of the CM fell to zero after the drop phase, in conjunction
with data from one force plate, resulted in differences in the jump height of
up to 17%.

Back to Top

Results:

Differences between the reference method and video based methods were
comparatively small (mean value of differences in jump height: −0.007 m,
SD: 0.013 m for the best of these methods) but not negligible.

Back to Top

Conclusions:

Nevertheless, video based methods turned out to be the most promising

alternative to the reference method to determine accurate variables
concerning drop jump performance.

Drop jumps are exercises in which the athlete, by forced eccentric

contraction, attempts to enhance subsequent exercise performance (1,3-
7,10-12,17,21). Athletes performing drop jumps jump down from a raised
platform and, upon landing, execute a maximal vertical jump. Different
methods are applied to analyze drop jumps biomechanically. Parameters
such as the jump height, the durations of the eccentric (impact) and
concentric (push-off) phases, and the vertical displacements during these
phases are frequently used in the literature dealing with the biomechanics of
plyometric jumping (3,5,6,8,17,18).

The parameter values are determined by applying different methods.

Several authors (1,8,9,17) estimate the jump height from the flight time,
i.e., from the time between the instant of leaving the ground and the instant
of landing after the jump. This method works if and only if the height of the
center of mass (CM) at the instant of leaving the ground is exactly the same
as that at the instant of landing.

In other investigations (4-6,19,20), markers are placed on the skin of the

subjects to define the positions of the body segments. Motion analysis
systems are used to record the drop jumps. The coordinates of the marker
points are extracted, scaled, and filtered. By utilizing models of the
segmented human body, positions of CM of body segments as well as of the
whole body are computed. The jump height and other parameter values are
then derived from these time-dependent positions.

Parameter values quantifying drop jump performance can also be calculated

from the vertical ground reaction forces. If the velocity at the instant that
the toes touch the ground after dropping is known, the parameter values can
be derived from the integral of the net vertical force (vertical ground
reaction force minus body weight). Some authors (18) estimate this velocity
from the height of the drop platform, from which the subject jumps down.
Another way is to record the vertical ground reaction forces from an instant
before the subject begins to move on the drop platform. The results
obtained by this method have been taken as reference in this study. In other
methods, force and video data are recorded simultaneously, and video
information is used to get more precise parameter values from the force
data or force data information is used to get more precise values from the
video information. The present investigation was undertaken to compare
different measuring techniques (see Parameter Estimation Methods) to find
out which method yields the lowest errors compared with the results
obtained by the reference method.

Back to Top

METHODS

Subjects and recording techniques. Five volunteer male subjects aged 18-
25 yr (height = 1.83 ± 0.034 m, body mass = 75.0 ± 5.5 kg) and in good
physical condition participated in this study. Informed consent was obtained
from the subjects in accordance with the policy statement of the American
College of Sports Medicine. The subjects performed five drop jumps. They
were instructed to keep their hands on their hips and to jump as high as they
could. Drop height was 0.39 m.

The motion was characterized using two-dimensional kinematic and force

data. As illustrated in Figure 1, markers were attached to the skin over bony
landmarks or over average joint centers defined by the 17-segment model of
the human body of Hatze (13). These markers defined the position of the
abdomino-thoracic segment, head, shoulders, arms, forearms, abdomino-
pelvic-segment, upper legs, lower legs, and feet. While jumping, the
subjects were recorded with a Kodak Ektapro 1000 high-speed video
system operating at a frame rate of 250 Hz. The optical axis of the camera
was approximately perpendicular to the plane of motion. Calibration
parameters were obtained from the image of a calibration frame recorded
previously to the motions. A modified direct linear transformation approach
was applied (15). The method described by Baca (2) was used to calculate
image coordinates of the centers of the circular markers. Absolute
coordinates were derived using the calibration parameters. The filtering
method proposed by Hatze (14) was used for data smoothing and
calculation of derivatives. Segment masses and centers of segment masses
relative to the segment fixed coordinate systems were estimated using the
anthropometric-computative method proposed by Hatze (13). In
combination with the coordinate data, these inertial parameters were used to
compute the absolute positions of CM of body segments, as well as the
position of the CM of the whole body.

perimental setup and position of markers applied to the skin of the subject.

Vertical and horizontal components of the ground reaction force were

registered simultaneously during the whole motion. Force data were
sampled at 1000 Hz using two synchronized force platforms (Kistler type
9281 B and Kistler type 9261 A, Kistler Instrumente AG, Winterthur,
Switzerland). A 16-bit analog-to-digital converter board (Analog Devices
type RTI850F, Norwood, MA) was used for that purpose. The high
sampling frequency used in collecting force data ensured that the impact
periods were accurately represented. Figure 2 gives an example of a record
obtained from both (synchronized) force plates. Figure 2 also identifies the
key time points used in the various analyses. As depicted in Figure 1, the
subjects jumped down from a platform made of pressboard. This drop
platform was mounted on a box of hardwood. The drop platform was bolted
to four aluminum pipes, which themselves were bolted to the four corners
of the first Kistler platform. In this way the box was pressed between the
drop and force platform. After dropping down, the subjects landed on the
second platform. Trials were evaluated only if the subjects landed on the
second force platform after the vertical jump again, which caused a lot of
trials to be repeated. Four-parameter estimation methods required the
integration of the net vertical forces over different time phases. A trapezoid
method was used for all the integrations.

mple of a force-time curve of the vertical ground reaction force during a drop jump. At t0, the subjects stands still on the first force platform. t1 is the instant of take-off
from this platform, t2 that of landing on the second force platform. At t3, the vertical velocity of the total CM is zero. t4 is the instant of leaving the ground, t5 that of touch-
down. At t6, which has been estimated by visual inspection of the force data, the subject stands still again. tecc is the duration of the eccentric phase, tconc that of the
concentric phase. Portions of data which came from each platform are indicated.

Back to Top

Parameter Estimation Methods

Double force plate technique (reference method, method A). From the
ground reaction forces recorded by the first force plate vL (the vertical
landing velocity) was calculated as follows: Equation (1) where Fz(t) is the
vertical ground reaction force, g is the acceleration due to gravity and m is
the body mass. This velocity can be used to determine vOFF, hJUMP, and
hDROP (see Table 1): Equations (2)-(4) tECC and tCONC are calculated from t2
and t4(Fig. 2), which are estimated from the force data of the second force
platform and by calculating the instant t3 fulfilling the condition Equation
(5) ΔzECC and ΔzCONC(Table 1) can be calculated as follows: Equation (6)
and Equation (7) where τ is an integration variable.

List and definition of all parameters calculated.

One force platform technique (method B). In this method, vL was
calculated by Equation (8) The remaining parameter values were calculated
in the same way as in reference method A (equations 2-7). No force data
from the first force plate are required for this method.

Potential energy method (method C). In this method, vL was calculated

using the drop platform height as an estimation of hDROP. From the
principle of transformation of potential into kinetic energy, it follows that:
Equation (9) The remaining parameter values were calculated in the same
way as in the reference method A (equations 2-7).
Flight time method (method D). In this method, vOFF was calculated from
the flight time. This flight time tF was estimated from the force data of the
second force platform (tF = t5 − t4). Equation (10) From vOFF the jump
height can be computed equation 3. Other parameter values have not been
calculated from the flight time.

Back to Top

Methods Based on Video Data Only

Video positions technique (method E). In this method, all parameter

values quantifying the drop jump performance were determined from video
information only. Hence, it was necessary to use video data to estimate t2,
t3, and t4(Fig. 2). The time sequences of the absolute coordinate values of
the marker point fixed to the fifth metatarsophalangeal joint (see Fig. 1)
were used to estimate t2 and t4. It was easier to estimate t2, because it
coincided with that of the maximum downward velocity of the marker. t3
was determined using the lowest vertical position of the total CM. hJUMP
was identified using the heights of the total CM at t4 and at the vertex of the
flight phase. hDROP was calculated as the change in height of the total CM
from the vertex of the drop phase to the point of landing assuming a vertical
velocity greater than or equal to zero at t1. vL and vOFF were calculated
from the jump and effective drop heights by equation 9 and Equation (11)
tECC and tCONC were computed using t2, t3 and t4, ΔzECC and ΔzCONC using
the vertical positions of the total CM at these instants.

Video positions and velocities technique (method F). t2 and t4(Fig. 2)

were estimated from video position data as in the previous method. vL and
vOFF were given from the derivatives of the position data of the total CM.
Other parameter values have not been calculated by this method.

Back to Top

Methods Based on Video and Force Platform Data

Video positions plus force data technique (method G). In this method,
the parameter values were obtained by combining force and video data. t2
and t4(Fig. 2) were estimated from the force data. t3 was determined from
the video data using the lowest vertical position of the total CM. hJUMP was
identified using the heights of the CM at t4 and at the vertex of the flight
phase. hDROP was calculated as the change in height of the total CM from
the vertex of the drop phase to the point of landing assuming a vertical
velocity greater than or equal to zero at t1. vL and vOFF were calculated
from hJUMP and hDROP by Equations 9 and 11. tECC and tCONC were
computed using t2, t3 and t4, ΔzECC and ΔzCONC using the vertical positions
of the total CM at these instants.

Zero velocity technique (method H). This technique is another method of

combining force and video data. The instants t2, t3, and t4 were estimated as
in the previous method. vL and vOFF were determined by integration of the
net vertical force data ending or starting at the instant t3: Equations (12) and
(13) tECC and tCONC were calculated using t2, t3, and t4. The vertical
displacements were found using equations 6 and 7. hJUMP and hDROP height
were calculated using equations 3 and 4.
Back to Top

RESULTS AND DISCUSSION

Because the accuracy of force plates is very high, very precise parameter
values can be expected from the double force plate method, if an adequate
sampling frequency is chosen. Therefore, this technique has been selected
as the reference method. The variables quantifying the performance of the
drop jumps were calculated for each subject from this method (Table 2).
Percentage change values of these variables applying the other measuring
techniques are presented in Figures 3-7.

Variables quantifying drop jump performance. Values were found using the reference method.

Means ± SD of the percentage change values of the jump heights.

Means ± SD of the percentage change values of the effective drop heights.

Means ± SD of the percentage change values of the durations of the eccentric and concentric phases.
Means ± SD of the percentage change values of the downward and upward vertical displacements during the eccentric and concentric phases.

Means ± SD of the percentage change values of the velocities vL and vOFF.

There are some errors, which may still be present in the double force plate
method: nonlinearity of the force measuring system; incorrect calibration;
errors due to analog to digital conversion, rounding, numerical integration,
and due to incorrect estimation of the instants of landing after the drop
phase and take-off from the ground. Most of these factors were discussed
by Kibele and Schwirtz (16). The data collection system was checked for
linearity by statically loading the drop platform and force platform 2 (see
Fig. 1) with weights up to 1000 N, and the linearity was better than 0.2%.
Beyond that, these weights have been put in different locations on force
platform 1 both with and without the box/pipe/drop platform construction.
Differences in the measured values were negligible. There are no impact
phases during take-off from the drop platform, which could cause problems
due to reduced resonant frequency of the measuring system. Before each
experimental session, the calibration of the data collection system was
controlled by measuring the body weight of the subject performing the drop
jumps by both force plates. Errors due to analog-digital conversion were
neglected, because the resolution of the converter used was high (16 bit)
and rounding errors were expected to cancel out during the integration. If
the sampling rate of the force data is too low, errors can be introduced by
numerical integration. Because, however, numerical integration based on
every second sampled force value resulted in almost the same results, these
errors were considered negligible for the present investigation. Small errors
were unavoidable in the estimation of t2 and t4. These errors should be,
however, not greater than 0.002 s because of the sampling frequency of
1000 Hz. A timing error of 0.002 s in the estimation of t2 would result in a
mean error of about 2% in the downward vertical displacement and a mean
error of less than 2% in the estimation of the drop height. The same error in
the estimation of t4 would result in mean errors of less than 2% in the
upward vertical displacement and the jump height.

Although the results of the one force platform method (B) were closest to
those from the reference method, this method is not a practical one for
analyzing drop jump performance. It was a difficult coordination task for
the subjects to land on the second force platform with dimensions 0.6 × 0.4
m after the jump again and to come to a standstill as soon as possible. This
means that the subject not only has to concentrate on the primary task of
jumping as high as possible but also on subsequent motions. Nevertheless,
the longer the integration period, the more linearity, calibration, numerical
integration, and rounding errors will affect the results.
General problems inherent in all video-based methods are the inadequacy of
mathematical models of the human body, the reduced sampling rate, and
errors introduced by the optical system. One major error source is the
marker shifting relative to the skeleton. Deviations of the point O1 fixed to
the subject's torso (see Fig. 1) strongly affected the results of the present
investigation.

Additional errors occur when incorrect time-points are pulled from video
data. In particular, estimating t3, the instant that the velocity of the CM was
zero, from the video data was a difficult task. The largest errors (−16.8% in
the jump height and −12.7% in the upward vertical displacement) in method
H, which estimates t3 from video information, were observed in a trial in
which t3 was estimated as being not more than 0.008 s (period of two
frames) later than that found in the reference method, using force data.

Differences of the video position technique (E) and the video position plus
force data technique (G) with regard to the reference method were
comparatively small (E: 0.000 ± 0.015 m, G: −0.007 ± 0.013 m for jump
height). Although the mean of the percentage change values of the jump
heights differed less for method E (see Fig. 3), it can be inferred from the
other results (in particular from the standard deviations, see Figs. 4-7) that
method G is superior to method E. A plausible explanation for the
comparatively high mean difference in the calculated effective drop heights
is apparent when the subjects' jumping technique is considered. The
subjects raised the leg more distant from the camera before dropping down.
At the instant that free fall started, the CM of this leg was still elevated.
Therefore, the assumption of a symmetric motion from the very beginning
is not justified, and the vertical position of the CM was underestimated. As
a consequence, the calculated effective drop height was too small. An
additional problem arises if the subjects are still in contact with the drop
platform at the instant that the CM reaches its maximum height. These
problems can be overcome by estimating vOFF from the jump height ((11))
and calculating the velocity of the CM at the instant of impact after the drop
phase by backward integration of the net vertical forces then. Equation (4)
can then be used to calculate the effective drop height. Large relative
differences were also observed for the vertical displacements during the
eccentric phase (Fig. 6). The corresponding absolute differences were,
however, not tremendously large. Marker shifting might be the main reason
for these differences, which were less than 11 mm on average.

Because of this marker shifting and imperfection of the measuring methods

and devices, errors are also introduced in the calculation of marker
velocities and consequently the velocity of the CM. It was therefore not
surprising (Fig. 7) that methods E and G (velocities estimated from the
heights) were more accurate than method F (velocities estimated from the
derivatives of the marker position data).

Method D overestimated the jump height as well as the velocity at the end
of the concentric phase, because in most trials the subject was in a more
extended position at the end of the concentric phase than at the end of the
flight phase. From the mean values and standard deviations of the relative
differences of the individual subjects' five jump heights with regard to the
reference method (M.J. :8.4 ± 3.3%; C.S.: 5.2 ± 4.4%; M.M.: 5.6 ± 2.8%;
W.J: −0.9 ± 2.6%; H.K.: 1.8 ± 4.4%), it can be inferred that method D is
better suited to compare trials of the same subject than to compare trials of
different subjects or to calculate exact parameter values.

In summary, it can be concluded from this study, which was based on

sampling video data at 250 Hz and force data at 1000 Hz, that the video
position plus force data technique (G) and the video position technique (E)
based on adequate models of the human body are the most promising
alternative to the reference method to determine accurate variables
concerning drop jump performance. If force data are recorded using one
force plate only, method G should be preferred to method E and the instants
of landing after the drop phase and take-off from the ground should be
estimated from force data. The results of this investigation indicate that the
flight time method (D) should only be applied to compare trials of the same
subject, if kinematic data cannot be obtained and only one force plate is
available. The potential energy method (C), the zero velocity method (H)
and the video positions and velocities technique (F) are not recommended
to analyze drop jump performance. Although the one force plate technique
turned out to be very accurate, it is not recommended, because the
requirement to stand still as soon as possible on the force plate after the
jump might influence the jumping technique.

Back to Top

REFERENCES

1. Asmussen, E., and F. Bonde-Petersen. Storage of elastic energy in

skeletal muscles in man. Acta Physiol. Scand. 92:385-392, 1974.
Cited Here...
2. Baca, A. Spatial reconstruction of marker trajectories from high speed
video image sequences. Med. Eng. Phys. 19:367-374, 1997.
Cited Here...
3. Bedi, J. F., A. Cresswell, G. Andrew, T. Engel, and S. M. Nicol. Increase
in jump height associated with maximal effort vertical depth jumps. Res. Q.
Exerc. Sport 58:11-15, 1987.
Cited Here...
4. Bobbert, M. F., M. Mackay, D. Schinkelshoek, P. A. Hujing, and G. J.
Van Ingen Schenau. Biomechanical analysis of drop and countermovement
jumps. Eur. J. Appl. Physiol. 54:566-573, 1986.
Cited Here...
5. Bobbert, M. F., P. A. Hujing, and G. J. Van Ingen Schenau. Drop
jumping: I. The influence of jumping technique on the biomechanics of
jumping. Med. Sci. Sports Exerc. 19:332-338, 1987.
Cited Here...
6. Bobbert, M. F., P. A. Hujing, and G. J. Van Ingen Schenau. Drop
jumping. II. The influence of drop height on the biomechanics of jumping.
Med. Sci. Sports Exerc. 19:339-346, 1987.
Cited Here...
7. Bobbert, M. F. Drop jumping as a training method for jumping ability.
Sports Med. 9:7-22, 1990.
Cited Here...
8. Bosco, C., P. V. Komi, and E. Locatelli. Physiologische Betrachtungen
zum Tiefsprungtraining. Leistungssport 9:434-439, 1979.
Cited Here...
9. Bosco, C., and C. Pittera. Zur Trainingswirkung neuentwickelter
Sprungübungen auf die Explosivkraft. Leistungssport 12:36-39, 1982.
Cited Here...
10. Clutch, D., M. Wilton, C. McGown, and G. R. Bryce. The effect of
depth jumps and weight training on leg strength and vertical jump. Res. Q.
Exerc. Sport 54:5-10, 1983.
Cited Here...
11. Gollhofer, A., and H. Kyröläinen. Neuromuscular control of the human
leg extensor muscles in jump exercises under various stretch-load
conditions. Int. J. Sports Med. 12:34-40, 1991.
Cited Here...
12. Haekkinen, K., and Komi, P. V. Alterations of mechanical
characteristics of human skeletal muscle during strength training. Eur. J.
Appl. Physiol. 50:161-172, 1983.
Cited Here...
13. Hatze, H. A mathematical model for the computational determination of
parameter values of anthropomorphic segments. J. Biomech. 13:833-843,
1980.
Cited Here...
14. Hatze, H. The use of optimally regularized Fourier series for estimating
higher-order derivatives of noisy biomechanical data. J. Biomech. 14:13-18,
1981.
Cited Here...
15. Hatze, H. High-precision three-dimensional photogrammetric
calibration and object space reconstruction using a modified DLT-approach.
J. Biomech. 21:533-538, 1988.
Cited Here...
16. Kibele, A., and A. Schwirtz. Zur Methodik der Sprungkraftdiagnostik:
von der reinen Sprunghöhenbestimmung zur biomechanischen
Kennwertbildung. In: Integrative Ansätze in der Bio&Mechanik, A.
Gollhofer (Ed.). Sankt Augustin, Germany: Academia Verlag, 1996, pp.
267-276.
Cited Here...
17. Komi, P., and C. Bosco. Utilization of stored elastic energy in leg
extensor muscles by men and women. Med. Sci. Sports Exerc. 10:261-265,
1978.
Cited Here...
18. Lees, A., and E. Fahmi. Optimal drop heights for plyometric training.
Ergonomics 37:141-148, 1994.
Cited Here...
19. McNitt-Gray, J. L. Kinetics of the lower extremities during drop
landings from three heights. J. Biomech. 26:1037-1046, 1993.
Cited Here...
20. McNitt-Gray J. L. Kinematics and impulse characteristics of drop
landings from three heights. Int. J. Sport Biomech. 7:201-224, 1991.
Cited Here...
21. Schmidtbleicher, D. Welche Absprunghöhen bei Tiefsprüngen?
Leichtathletik 34:47-49, 1983.
Cited Here...
Keywords:

BIOMECHANICS; JUMP HEIGHT; MOTION ANALYSIS;

PLYOMETRICS

© 1999 Lippincott Williams & Wilkins, Inc.