An Analysis of the Technical Aspects of Competitive Canoe Slalom

AN ANALYSIS OF THE TECHNICAL ASPECTS OF COMPETITIVE CANOE SLALOM

Adam Hunter

Thesis presented for the degree of Doctor of Philosophy

University of Canberra Faculty of Health June 2010

This thesis investigated the technical and tactical aspects of competitive canoe slalom with a view to provide information that could assist coaches and athletes in enhancing performance and designing training programs that are grounded in scientific research. To achieve these goals two methodology projects and five studies were undertaken.

A customised competition analysis program (CAP) and standardised set of operational definitions were developed. Assessment of the intra-observer and inter-observer reliability (> 78%) of data gathered using this system was within the limits considered acceptable in previous research. In addition to CAP, a paddle force analysis program was developed (absolute variance = 5.1 N) which allowed analysis and comparison of paddle force data with CAP data.

This thesis investigated the distribution and variability of paddlers in each category from historical information. Results demonstrated that athletes within the top ten in each category were grouped based on the category in which they paddled. Men‟s single canoe (C1M) paddlers were found to regularly complete courses within 105% of the for men‟s single kayak (K1M) time.

Competition performances K1M, C1M and women‟s single kayak (K1W) were analysed using the CAP. For a complex gate sequence paddlers using a spin rather than pivot manoeuvre had significantly (p < 0.05) faster split times for the gates before and after the execution of the manoeuvre. Therefore, specific strategies that were beneficial to performance were identified.

To understand further the relationship between strategy and performance, six trials for international level K1M (n = 11) and C1M (n = 6) were digitised for a left hand upstream gate. Results showed that there was a strong correlation (r >= 0.89) between boat trajectory and the total time. The boat trajectory of the fastest paddlers were found to be significantly closer (p < 0.05) from the slowest paddlers.

The effectiveness of combining paddle force data with CAP data was demonstrated through a case study of a C1M paddler. This amalgamation also showed how the paddler lost time

ii around an upstream gate due to an error, demonstrating the link between the force applied to the paddle, the strategy the paddler used and the path they chose.

Paddle shaft length was manipulated in a case study involving 12 repeat sprints on flatwater, to improve factors related to canoe slalom performance. Results showed that the shorter shaft allowed the paddler to accelerate the boat more rapidly, a key factor in canoe slalom performance.

Dissemination of information to coaches and athletes is the critical link in any analysis process, which is required for beneficial gains to be made from the information. Presentation of an example report demonstrates how this process was facilitated in this thesis.

The findings of these studies were related back to a model of critical variables for canoe slalom showing that this thesis was able to provide scientific information into key intrinsic variables which the paddler can influence including: run time, path taken, mean velocity, strokes used, technique, experience, variability and errors. The interactions of these variables and research findings are discussed, then improvements and future research directions presented.

iii CERTIFICATE OF AUTHORSHIP OF THESIS

Except where acknowledged clearly in footnotes, quotations and the bibliography, I certify that I am the sole author of the thesis submitted today entitled –

An Analysis of the Technical Aspects of Competitive Canoe Slalom

I certify further that to the best of my knowledge my thesis contains no material previously published or written by another person except where due reference is made in the text of this thesis. The material in this thesis has not been the basis of an award of any other degree or diploma except where due reference is made in the text of the thesis. The thesis complies with University requirements for a thesis as set out in www.canberra.edu.au/secretariat/goldbook/forms/thesisrqmt.pdf

Signature of Candidate

Signature of chair of the supervisory panel

Date:

iv ACKNOWLEDGEMENTS

I would like to thank the following people for their input, support and assistance during the course of this study:

Firstly, I would like to thank my academic supervisors Dr Margi Böhm, Professor Keith Lyons and Dr Jodie Cochrane for their guidance and encouragement in all aspects of this candidature. Each through their own knowledge have challenged and enlightened me in many areas and for that, I will always be grateful. Thanks to the University of Canberra, Australian Institute of Sport (AIS) and Canoeing Australia for financial support during the course of my work.

A huge thanks (and sorry) to Alexi Sachlikidis, Daniel McFarlane, Angela McCombe and Hamish Jeacocke for the early mornings and late nights during testing, for enduring long weekends in Penrith and for climbing in and out of the slippery, slimy, icy cold water to set up the calibration.

Thanks to the many colleagues at the AIS for their advice, support, encouragement and willingness to share information. It has been a pleasure to work alongside each of you and has improved my knowledge of biomechanics. Special thanks to Nick Brown, John Baker, and Alison Alcock for critiquing my work along the way.

To the coaches of the AIS slalom program, Mike Druce and Miriam Fox, thanks for educating me about the finer details of canoe slalom and for your patience when things didn‟t quite go to plan. Particular thanks goes to the athletes involved in the studies for their time and the effort, which they put into each of the trials. Without their effort, the investigation would not have been a success.

Finally thank you to my family and friends, you have supported me in various ways through my journey, it definitely has not been easy, and I deeply appreciate your encouragement, patience and understanding.

v DEDICATION

To my parents Judy and Roy Hunter and my grandma Doreen Hunter.

vii TABLE OF CONTENTS

AN ANALYSIS OF THE TECHNICAL ASPECTS OF COMPETITIVE CANOE SLALOM ...... I

ABSTRACT ...... II

CERTIFICATE OF AUTHORSHIP OF THESIS ...... IV

ACKNOWLEDGEMENTS ...... V

PUBLICATIONS AND PRESENTATIONS BY THE CANDIDATE RELEVANT TO THE THESIS ..... VI

DEDICATION ...... VII

TABLE OF CONTENTS ...... VIII

LIST OF FIGURES ...... XII

LIST OF TABLES ...... XIX

LIST OF APPENDICES ...... XXI

DEFINITIONS AND NOMENCLATURE ...... XXII

1 INTRODUCTION ...... 1

1.1 HISTORICAL PERSPECTIVE ...... 1 1.2 CURRENT SLALOM COMPETITIVE ENVIRONMENT ...... 6 1.3 MOTIVATION FOR A TECHNICAL ANALYSIS OF CANOE SLALOM ...... 9 1.3.1 Visualisation Skills ...... 10 1.3.2 Mental Arousal ...... 10 1.3.3 Physiology of slalom paddlers ...... 11 1.3.4 Body composition of elite slalom paddlers ...... 11 1.3.5 Technique ...... 12 1.3.6 Force Application ...... 12 1.3.7 Repeatability of Athletic Performance ...... 13 1.3.8 Equipment Setup ...... 13 1.3.9 Augmented information ...... 14 1.3.10 Summary of Key Intrinsic Attributes ...... 15 1.4 STATEMENT OF THE RESEARCH PROBLEM FOR THIS THESIS ...... 15 1.5 SCOPE OF THE THESIS ...... 17 1.6 THESIS LAYOUT AND STRUCTURE ...... 20

2 METHODOLOGY ...... 21

2.1 PREVIOUS RESEARCH METHODOLOGY ...... 21 2.1.1 Time-motion analysis ...... 21 2.1.2 Need for Customised Analysis Software for Canoe Slalom Time Motion Analysis ...... 23 2.1.3 Motion Analysis ...... 24

viii 2.1.4 Global Positioning Systems ...... 25 2.1.5 Force Curves ...... 27 2.2 DEVELOPMENT OF ANALYSIS SYSTEMS FOR CANOE SLALOM ...... 28 2.2.1 Competition Analysis Program (CAP) for Canoe Slalom ...... 28 2.2.1.1 Development of CAP ...... 28 2.2.1.1.1 Platform for CAP ...... 29 2.2.1.1.2 Operational Definition Set ...... 29 2.2.1.1.3 Navigating CAP ...... 35 2.2.1.1.4 New Analysis ...... 36 2.2.1.1.5 Load Analysis ...... 40 2.2.1.1.6 Crash Recovery ...... 40 2.2.1.1.7 Set Hot Keys ...... 40 2.2.1.1.8 Individual Report One and All ...... 41 2.2.1.1.9 Summary Report ...... 47 2.2.1.2 Reliability Assessment for CAP ...... 53 2.2.1.2.1 Methodology ...... 53 2.2.1.2.2 Results ...... 55 2.2.1.2.3 Discussion ...... 58 2.2.1.2.4 Conclusions ...... 61 2.2.1.3 Summary of CAP...... 61 2.2.2 Paddle Force Curve Analysis Program for Slalom ...... 62 2.2.2.1 Development ...... 62 2.2.2.1.1 Capturing...... 63 2.2.2.1.2 Displaying the Data ...... 64 2.2.2.1.3 Calibration ...... 66 2.2.2.1.4 Drift or mechanical shift correction ...... 71 2.2.2.1.5 Trimming the Trial ...... 74 2.2.2.1.6 Filtering ...... 75 2.2.2.1.7 Switching Channels of Data ...... 76 2.2.2.1.8 Exporting and Saving ...... 77 2.2.2.1.9 Analysis ...... 77 2.2.2.1.10 Reporting...... 81 2.2.2.1.11 Conclusions ...... 83 2.2.2.2 Paddle Force calibration and error assessment ...... 83 2.2.2.2.1 Methods ...... 84 2.2.2.2.2 Results ...... 87 2.2.2.2.3 Conclusions ...... 88 2.3 SLALOM ANALYSIS IN COACHING AND RACING ENVIRONMENT ...... 89 2.3.1 Analysis of Canoe Slalom Category Clustering ...... 89 2.3.2 Canoe Slalom Competition Analysis Program (CAP) ...... 90 2.3.3 Canoe Slalom Boat Trajectory while Negotiating an Upstream Gate out of Competition ...... 93 2.3.4 Interaction between Strategy and Paddle Force ...... 97 2.3.5 Equipment Setup – Canoe Slalom Paddle Length Comparison ...... 99

ix 3 ANALYSIS OF CANOE SLALOM CATEGORY CLUSTERING ...... 101

3.1 RESULTS ...... 101 3.2 DISCUSSION ...... 104 3.3 CONCLUSIONS ...... 106

4 CANOE SLALOM COMPETITION ANALYSIS ...... 107

4.1 INTRODUCTION ...... 107 4.2 RESULTS ...... 107 4.2.1 Gate Split Times ...... 107 4.2.2 Strategy ...... 108 4.2.3 Turns ...... 110 4.2.4 Strokes ...... 111 4.2.5 Penalties ...... 117 4.3 DISCUSSION ...... 117 4.3.1 Gate Split Times ...... 117 4.3.2 Strategy ...... 118 4.3.3 Turns ...... 119 4.3.4 Strokes ...... 120 4.3.5 Penalties ...... 121 4.4 CONCLUSIONS ...... 122

5 CANOE SLALOM BOAT TRAJECTORY WHILE NEGOTIATING AN UPSTREAM GATE ... 123

5.1 INTRODUCTION ...... 123 5.2 RESULTS ...... 123 5.3 DISCUSSION ...... 127 5.4 CONCLUSIONS ...... 129

6 INTERACTIONS BETWEEN STRATEGY AND PADDLE FORCE...... 130

6.1 INTRODUCTION ...... 130 6.2 RESULTS ...... 130 6.3 DISCUSSION ...... 140 6.4 CONCLUSIONS ...... 141

7 EQUIPMENT SETUP – CANOE SLALOM PADDLE LENGTH COMPARISON ...... 143

7.1 INTRODUCTION ...... 143 7.2 RESULTS ...... 143 7.3 DISCUSSION ...... 145 7.4 CONCLUSIONS ...... 146

8 DISSEMINATION OF INFORMATION TO COACHES AND ATHLETES ...... 147

8.1 INTRODUCTION ...... 147 8.2 EXAMPLE OF A RACE REPORT SENT TO THE COACH AND ATHLETE ...... 148

x 8.2.1 Comparison of gate interval times ...... 148 8.2.2 Stroke count ...... 150 8.2.3 Stroke duration ...... 151 8.2.4 Race Profile ...... 153 8.3 CONCLUSIONS ...... 158

9 CONCLUSIONS ...... 161

9.1 RUN TIME, PATH TAKEN AND MEAN VELOCITY...... 162 9.2 STROKES USED ...... 163 9.3 TECHNIQUE AND ERRORS ...... 164 9.4 EXPERIENCE AND VARIABILITY ...... 164 9.5 PADDLE SETUP ...... 164 9.6 IMPROVING THE ANALYSIS PROCESS ...... 165 9.7 FUTURE RESEARCH DIRECTIONS ...... 168

LIST OF REFERENCES ...... 169

xi LIST OF FIGURES

FIGURE 1.1: GATE 2 REPRESENTS A DOWNSTREAM GATE (GREEN (G) AND WHITE POLE) THROUGH WHICH A

PADDLER CAN CONTINUE AS PART OF THEIR DOWNSTREAM LINE. GATE 3 IS AN UPSTREAM GATE (RED (R)

AND WHITE POLES) INDICATING THAT THE PADDLER MUST PADDLE AROUND THE GATE TO THE RIGHT OR

LEFT (BASED ON POSITIONING RELATIVE TO THE BANK) AND THEN PASS THROUGH FROM THE DOWNSTREAM DIRECTION...... 8 FIGURE 1.2: A MODEL OF THE CRITICAL VARIABLES IN A CANOE SLALOM COMPETITION, THEIR INTERACTION AND

THEIR GENERAL GROUPING BASED ON TYPE. THE MODEL HIGHLIGHTS THE VARIABLES WHICH THE PADDLER CAN INFLUENCE (GREEN)...... 17 FIGURE 2.1: MAIN MENU GUI (GRAPHICAL USER INTERFACE) ALLOWED THE OPERATOR TO NAVIGATE THE

FUNCTIONALITY OF THE COMPETITION ANALYSIS PROGRAM (CAP). NEW ANALYSIS IS FOR STARTING

ANALYSIS OF A NEW COMPETITION RUN. LOAD ANALYSIS OPENS AN EXISTING ANALYSIS FOR REVIEW OR

EDITING. CRASH RECOVERY OPENS THE CURRENT ANALYSIS IF THE PROGRAM WAS INADVERTENTLY CLOSED.

SET HOT KEYS ALLOWS THE USER TO SETUP THEIR PERSONALISED KEYBOARD SHORTCUTS FOR THE ANALYSIS

GUI. INDIVIDUAL REPORT ONE AND ALL PROCESS THE ANALYSED DATA AND PRODUCE BASIC STATISTICS ON

EACH FILE. SUMMARY REPORT SENDS THE DATA TO A SINGLE SUMMARY FILE FOR STATISTICAL COMPARISON...... 35 FIGURE 2.2: DIALOG BOX FOR THE NUMBER OF VIDEO FILES USED IN THE ANALYSIS. BETWEEN ONE AND TEN VIDEO FILES COULD BE SELECTED...... 36 FIGURE 2.3: DIALOG BOX TO SELECT VIDEOS FOR THE ANALYSIS IN THE ORDER THAT THE CAMERAS WERE

SITUATED ALONG THE COURSE. IF THREE VIDEO FILES WERE TO BE USED, THEN THE DIALOG APPEARED THREE TIMES...... 36 FIGURE 2.4: THE SYNCHRONISATION GUI ALLOWS THE OPERATOR TO VIEW EACH VIDEO FILE AND IDENTIFY

SYNCHRONISATION POINTS. ONCE IDENTIFIED AND ENTERED USING THE BUTTONS (SINC1 AND SINC2) THE

OPERATOR CAN REVIEW THEM BY CLICKING ON THE VIDEO FILE IN THE LIST. THE VIDEO ALWAYS JUMPS TO

THE FIST SYNCHRONISATION POINT THUS TO REVIEW THE SECOND POINT THE USER CAN CLICK A BUTTON

(GOTO SINC 2)...... 37 FIGURE 2.5: EXAMPLE OF HOW MULTIPLE VIDEO FILES ARE COMBINED INTO A SINGLE ANALYSIS TIMELINE USING

SYNCHRONISATION POINTS. THE FIRST SYNCHRONISATION POINT (SYNC POINT) OF THE FIRST VIDEO IS

ALWAYS 0 S AND THE LAST SYNC POINT OF THE FINAL VIDEO FILE REPRESENTS THE LENGTH OF THAT VIDEO

FILE IN SECONDS. THE OTHER SYNC POINTS REPRESENT A COMMON TIME POINT BETWEEN TWO VIDEOS WHICH ARE IDENTIFIABLE IN BOTH VIEWS...... 38 FIGURE 2.6: THE ANALYSIS GUI SHOWS THE FINAL LAYOUT OF THE GUI. DURING THE ANALYSIS OF A

COMPETITION RUN, THE OPERATOR CODES EVENTS SUCH AS GATE SPLITS, SECTORS OF UPSTREAM TURNS,

TOUCHED GATES, MISSED GATES, ESKIMO ROLLS. THE OPERATOR CAN ALSO TAG MAJOR OR MINOR

AVOIDANCE STRATEGIES USED TO NEGOTIATE A GATE. FINALLY, THE ENTRY AND EXIT OF EACH STROKE IS MARKED AND CODED INDIVIDUALLY AS RIGHT OR LEFT WITH THE TYPE OF STROKE IDENTIFIED...... 39 FIGURE 2.7: EXAMPLE OF DATA STORED IN THE BACKUP FILE (TEMP CANOE SLALOM ANALYSIS.CSV). THE FIRST

LINE OF THE FILE CONTAINS THE HEADER INFORMATION FOR THE SYNCHRONISATION DATA. THE FOLLOWING

LINES CONTAIN THE SYNCHRONISATION INFORMATION FOR EACH VIDEO FILE IN THE ANALYSIS. THE HEADER

xii INFORMATION FOR THE ANALYSIS DATA AND ANY DATA FROM THE CURRENT ANALYSIS WHICH HAD BEEN

COMPLETED FOLLOWS...... 40 FIGURE 2.8: HOT KEY ASSIGNMENT GUI ALLOWS THE OPERATOR TO ASSIGN WHICH KEYS CONTROL THE PROGRAM

IN THE ANALYSIS GUI. ALL CODING BUTTONS ON THE ANALYSIS GUI HAVE A KEYBOARD SHORTCUT

ASSIGNED TO THEM. THIS ALLOWED RAPID NAVIGATION OF THE SOFTWARE AND INCREASED ANALYSIS SPEED...... 41 FIGURE 2.9: COURSE GATE SEQUENCE INPUT GUI (LEFT – ZERO GATE SELECTED, RIGHT – RUN WITH GATE DEFINITIONS ENTERED)...... 42 FIGURE 2.10: AN EXAMPLE OF A REPORT FILE, SHOWING THE RAW SPREADSHEET. THIS IS THE DATA CONTAINED IN

THE ORIGINAL CSV FILE WHICH WAS CREATED DURING THE ANALYSIS PROCESS. THE SAME STRUCTURE

REMAINS WITH THE SYNCHRONISATION DETAIL CONTAINED IN THE FIRST THREE LINES OF THIS EXAMPLE.

TIMES ARE ALL RELATIVE TO THE START OF THE FIRST VIDEO AND DATA IN THE ORDER IN WHICH IT WAS CODED...... 43 FIGURE 2.11: AN EXAMPLE OF A REPORT FILE, SHOWING THE ACTUAL SPREADSHEET. THE CONVERSION TO THIS

FORMAT REQUIRES ALL DATA TO BE SORTED SEQUENTIALLY AND TIMES MADE RELATIVE TO THE START

ALLOCATED BY THE OPERATOR DURING ANALYSIS. THE DATA ENTERED IN THE COURSE GATE SEQUENCE

GUI HAS BEEN COMBINED WITH THE RAW GATE DATA AND THE PENALTIES AND AVOIDANCE TIED TO THE

GATE AT WHICH THEY OCCURRED. STROKE TIME AND STROKE RATE ARE ALSO CALCULATED IN THIS PROCESS...... 44 FIGURE 2.12: AN EXAMPLE OF A REPORT FILE, SHOWING THE TOTAL SPREADSHEET. THE STROKE INFORMATION IS

PROCESSED TO DETERMINE THE TOTAL NUMBER OF EACH TYPE OF STROKE, AND THE AVERAGE AND

STANDARD DEVIATION OF TIME SPENT IN THE WATER PER STROKE TYPE. THE NUMBER OF STROKES BETWEEN EACH GATE IS ALSO CALCULATED...... 45 FIGURE 2.13: AN EXAMPLE OF A REPORT FILE, SHOWING THE GRAPH SPREADSHEET. THIS SPREADSHEET IS THE

BASE SETUP FOR CREATING THE CUSTOM GRAPHS FOR THE REPORT. NOTE THAT THESE GRAPHS ARE ® PRESENTED IN EXCEL DEFAULT FORMAT AND NOT IN STANDARD SCIENTIFIC FORMAT AS THESE ARE TO

ALLOW THE OPERATOR TO ASSESS THE OUTPUT PRIOR TO REPORTING AND ARE NOT USED IN THE FINAL

REPORT...... 46 FIGURE 2.14: THE FIRST VERSION OF THE RACE TIMELINE. THIS PLOT SHOWS THE STROKE IN WATER TIME (S) ON

THE Y-AXIS AND THE RACE TIME ON THE X-AXIS. EACH STROKE RIGHT (BLUE, CIRCLES) AND LEFT (RED,

SQUARES) IS POSITIONED ALONG THE RACE TIME (X-AXIS) BASED ON THE START (IN) TIME OF THE STROKE.

VERTICAL GREEN AND RED LINES REPRESENT DOWNSTREAM AND UPSTREAM GATES RESPECTIVELY. THE DIVISIONS AROUND UPSTREAM GATES (TURN) WERE DEFINED BY VERTICAL YELLOW LINES...... 48 FIGURE 2.15: THE SECOND VERSION OF THE RACE TIMELINE. THE RACE TIME WAS PRESENTED ON THE X-AXIS.

EACH STROKE POSITION AND DURATION IS REPRESENTED BY A SEPARATE LINE POSITIONED ALONG THE RACE

TIME (X-AXIS), BUT THE RIGHT AND LEFT STROKES ON SEPARATE LINES ON THE Y-AXIS WITH RIGHT = 1 SET

AS BLUE AND LEFT = 2 SET AS RED. VERTICAL GREEN AND RED LINES REPRESENT DOWNSTREAM AND

UPSTREAM GATES RESPECTIVELY. THE DIVISIONS AROUND UPSTREAM GATES (TURN) WERE DEFINED BY VERTICAL YELLOW LINES...... 49

xiii FIGURE 2.16: BAR GRAPH SHOWING THE NUMBER OF EACH TYPE OF STROKE WHICH THE PADDLER TOOK DURING

THE RUN. MULTI-STROKES ARE STROKES THAT COULD NOT BE DEFINED AS ONE OF THE STROKES IDENTIFIED IN THE DEFINITION SET...... 50 FIGURE 2.17: A REPORT TEMPLATE FILE WITH PORTRAIT FIGURES FOR QUICK DATA COLLECTION TO FACILITATE TIMELY COACH AND ATHLETE FEEDBACK...... 51 FIGURE 2.18: THE REPORT TEMPLATE FILE WITH LANDSCAPE FIGURES FOR QUICK DATA COLLECTION TO FACILITATE TIMELY COACH AND ATHLETE FEEDBACK...... 52 FIGURE 2.19: MENU STRUCTURE OF PADDLE FORCE ANALYSIS PROGRAM. THE MENU STRUCTURE DEMONSTRATES

THE PROCESSING CAPABILITIES DESIGNED INTO THE SOFTWARE WHICH INCLUDES, CAPTURING, REPLAYING, DISPLAYING, EDITING, PROCESSING AND CORRECTING DATA...... 62 FIGURE 2.20: THE CAPTURE GUI IS USED TO CAPTURE DATA FROM THE GERMAN (SPERLICH) FORCE SYSTEM. THE FILENAME GENERATED IS TIME AND DATE CODED SO FILES CANNOT BE OVERWRITTEN...... 63 FIGURE 2.21: MAIN DATA MANIPULATION PAGE. THE UPPER PLOT SHOW ALL THE DATA. THE LOWER PLOT

SHOWS A TEN SECOND SUBSECTION ALIGNED WITH THE DASHED WHITE LINE IN THE UPPER PLOT AND THE

SLIDER BAR. THE DASHED WHITE LINE IN THE LOWER PLOT REPRESENTS THE CURRENT DATA POINT, THE

VALUES FOR WHICH ARE SHOWN IN THE TABLE AT THE BOTTOM OF THE GRAPH. THE BLUE LINE REPRESENTS

THE LEFT HAND PADDLE FORCE (UP IS POSITIVE) AND THE RED LINE REPRESENTS THE RIGHT HAND PADDLE FORCE (DOWN IS POSITIVE)...... 65 FIGURE 2.22: PLOT SETTINGS GUI ALLOWS THE USER TO DETERMINE THE CONFIGURATION OF THE DATA VIEWED

ON THE MAIN GUI. THE PLOTS CAN BE RESCALED AND THE LINE WIDTH SELECTED. THE CHANNELS TURNED

ON OR OFF, SWITCHED TO BE UP (POSITIVE UP) OR DOWN (POSITIVE DOWN) AND THE COLOUR OF THE LINE CAN BE ALTERED...... 66 FIGURE 2.23: ZERO CALIBRATION SELECTION (GREEN RECTANGLE IN BOTH PANES OF THE MAIN GUI) AND

ASSIGNMENT (RED CIRCLE) FOR CHANNEL 2. USER LEFT CLICKS TO SET THE START OF THE CALIBRATION

AREA AND THEN RIGHT CLICKS TO SET THE END. ONCE IN THE CALIBRATION GUI THE USER SETS THE CHANNEL AND CLICKS IN THE CIRCLED BOX TO TRANSFER THE MEAN VALUE FOR THE SELECTED SECTION. 68 FIGURE 2.24: SECOND CALIBRATION (200 N) SELECTION (GREEN RECTANGLE IN BOTH PANES OF THE MAIN GUI),

ASSIGNMENT (RED CIRCLE) AND RESULTANT CALIBRATION INFORMATION FOR CHANNEL 2. USER LEFT

CLICKS TO SET THE START OF THE CALIBRATION AREA AND THEN RIGHT CLICKS TO SET THE END. ONCE IN

THE CALIBRATION GUI THE USER SETS THE CHANNEL AND CLICKS IN THE CIRCLED BOX TO TRANSFER THE

MEAN VALUE FOR THE SELECTED SECTION. BY CLICKING CALIBRATE THE DATA FOR THE CHANNEL IS RECALCULATED USING THE CALIBRATION...... 69 FIGURE 2.25: COMPLETED CALIBRATION FOR CHANNELS 2 AND 3. ONCE ALL THE REQUIRED CHANNELS HAVE BEEN CALIBRATED THE USER CAN SAVE THE CALIBRATION DATA FOR LATER USE...... 70 FIGURE 2.26: ZEROING BASED ON A FIXED OFFSET OR AVERAGE OF A SELECTED AREA. CLICKING IN THE TEXTBOX

(RED RECTANGLE) TRANSFERS THE AVERAGE OF THE SELECTED REGION OR THE USER CAN TYPE THE VALUE THEY WANT...... 72 FIGURE 2.27: ZERO SELECTION TAKES FIVE DATA POINTS EITHER SIDE OF EACH GREEN LINE ON THE TOP PANE.

GREEN LINES ARE SET BY CLICKING IN THE BOTTOM PANE AT THE POINT WHERE THE PLOT SHOULD BE ON ZERO...... 73

xiv 2 FIGURE 2.28: POLYNOMIAL MODEL OF POINTS SELECTED, R VALUE AND THE CONTROLS FOR ADJUSTING THE

MODEL. LEFT – ORDER OF THE POLYNOMIAL TO BE FITTED. RIGHT – SHOWS THE RESULTANT DATA

REMODELLED (STRAIGHT LINE) AND THE DETAILS OF THE CORRECTION TRANSFERRED TO THE CALIBRATION PANE (BOTTOM)...... 73 FIGURE 2.29: SELECTION TO DELETE A SECTION AT THE BEGINNING OF THE TRIAL (TOP) AND SELECTION TO DELETE THE END OF THE TRIAL (BOTTOM). THIS ALLOWS THE USER TO TRIM THE FILE TO ONLY RELEVANT DATA... 74 FIGURE 2.30: FIGURE 2.30: BUTTERWORTH FILTER CONTROL GUI SHOWING THE DATA INPUT CHANNEL TO BE

FILTERED AND THE CUT OFF FREQUENCY USED BY THE FOURTH ORDER BUTTERWORTH FILTER WHICH CAN BE

SET BY THE OPERATOR. IN THIS EXAMPLE, 5 HZ WAS USED TO DEMONSTRATE THE FUNCTIONALITY OF THE FILTER...... 75 FIGURE 2.31: EXAMPLE OF THE EFFECT OF A DUAL DIRECTION LOW PASS FOURTH ORDER BUTTERWORTH FILTER

WITH A CUT OFF FREQUENCY OF 5 HZ ON CHANNEL 2 (RED DATA LINE). THE RAW DATA ARE SHOWN ON THE

LEFT. THE SMOOTHED DATA ARE SHOWN ON THE RIGHT. THIS IS AN EXAMPLE TO DEMONSTRATE THE FUNCTIONALITY OF THE FILTER...... 75 FIGURE 2.32: SWITCH CHANNELS GUI – THE USER SELECTS THE TWO CHANNELS TO BE SWITCHED...... 76 FIGURE 2.33: RESULT OF SWITCHING CHANNELS 2 AND 3 (BEFORE – LEFT AND AFTER – RIGHT, RED DATA SWAPPED FOR BLUE DATA AND VICE VERSA)...... 76 FIGURE 2.34: THE TRIAL DETAILS GUI ALLOWS THE USER TO ENTER DETAILS USED IN THE FILE NAME WHEN

EXPORTING THE DATA AND WHEN CREATING REPORTS. IF THESE ARE NOT SET PRIOR TO EXPORTING THE USER IS PROMPTED TO ENTER THEM...... 77 FIGURE 2.35: SELECTING A SECTION OF DATA TO ANALYSE. THE USER LEFT CLICKS TO SELECT THE START OF THE ANALYSIS AREA AND RIGHT CLICKS TO SELECT THE END OF THE ANALYSIS AREA...... 78 FIGURE 2.36: THE ANALYSIS SETTINGS GUI ALLOWS THE USER TO SELECT THE RIGHT AND LEFT DATA CHANNELS AND SET THE THRESHOLD VALUES FOR DETECTING STROKES...... 79 FIGURE 2.37: MEAN RIGHT AND LEFT FORCE CURVES FOR THE SELECTED PERIOD WITH ONE STANDARD DEVIATION EITHER SIDE AND BIOMECHANICAL STROKE CHARACTERISTICS...... 80 FIGURE 2.38: BIOMECHANICAL DATA RELEVANT TO PADDLE FORCE CURVES (FIGURE 2.37) AND INCLUDED IN EACH

REPORT. MEAN DATA FOR THE DOUBLE STROKE CYCLE AS WELL AS RIGHT AND LEFT STROKE CHARACTERISTICS ARE CALCULATED...... 81 FIGURE 2.39: TIME VARIABLES PRESENTED IN THE BIOMECHANICAL TABLE CAN BE DEFINED ON AN AVERAGE FORCE CURVE...... 81 FIGURE 2.40: FORCE AND IMPULSE VARIABLES PRESENTED IN THE BIOMECHANICAL TABLE CAN BE DEFINED ON AN AVERAGE FORCE CURVE...... 82 FIGURE 2.41: PADDLE MEASUREMENTS TAKEN DURING THE SETUP OF THE PADDLE FORCE SYSTEM. LEFT AND

RIGHT CENTRE OF PRESSURE AND MIDDLE FINGER MARKS ARE ALL REQUIRED FOR THE CALIBRATION OF THE

PADDLE FORCE SYSTEM TO THE FLEXIBILITY OF THE SHAFT OF THE PADDLE...... 84 FIGURE 2.42: END VIEW OF THE TRIPOD SETUP FOR THE PADDLE FORCE CALIBRATION PROCEDURE...... 85 FIGURE 2.43: SIDE VIEW OF TRIPOD SETUP (SETUP FOR CALIBRATING LEFT SIDE) FOR THE PADDLE FORCE CALIBRATION PROCEDURE...... 85

xv FIGURE 2.44: SET-UP FOR CALIBRATION OF THE FORCE TRANSDUCERS (LEFT) SHOWING TRIPODS SUPPORTING THE PADDLE WITH THE CALIBRATION WEIGHT HANGING FROM THE HAND GRIP...... 86 FIGURE 2.45: ZERO STABILITY TEST RESULTS FOR ALL SENSORS DEMONSTRATING THE NORMALISED MEAN OF

EACH UNIT AND THE STABILITY OF THE SIGNAL ABOUT THIS POINT. EACH PLOT SHOWS THE MINIMUM AND

MAXIMUM AS THE TAILS (BLACK LINES), AND THE STANDARD DEVIATION OF THE SAMPLE (YELLOW BOX).

THE LEFT PLOT REPRESENTS THE RAW DIGITAL UNITS AND THE RIGHT PLOT REPRESENTS THE PERCENTAGE OF THE FULL SCALE DATA...... 87 FIGURE 2.46: THE LINEAR CALIBRATION OF APPLIED LOAD (Y-AXIS) AND THE PADDLE FORCE SYSTEM MEASUREMENT (X-AXIS), TOGETHER WITH THE REGRESSION EQUATION...... 88 FIGURE 2.47: THE SECTION OF THE UPSTREAM TURN ANALYSED WAS FROM POSITION ONE TO POSITION FOUR.

TURN TIMES ONE AND FOUR WERE DEFINED AS THE POINT WHERE THE CENTRE OF THE BOAT PASSED

PARALLEL TO THE GATE LINE. TURN TIMES TWO AND THREE WERE DEFINED AS THE POINT WHERE THE

CENTRE OF THE BOAT PASSED A LINE PERPENDICULAR TO THE GATE LINE ORIGINATING FROM THE INSIDE

POLE...... 90 FIGURE 2.48: CANOE SLALOM 2005 WORLD CHAMPIONSHIPS FINALS COURSE MAP (AUSTRALIAN CANOEING,

2005). THE RED RECTANGLE HIGHLIGHTS THE MOST COMPLEX SEQUENCE OF GATES ON THE COUSE (GATES

10 TO 15). THIS SECTION REQUIRED PADDLERS TO RAPIDLY CHANGE DIRECTION NUMEROUS TIMES WHICH IS WHY IT WAS CHOSEN FOR THIS ANALYSIS...... 92 FIGURE 2.49: THE CANOE SLALOM COURSE AND EQUIPMENT SETUP USED FOR TESTING, DRAWN TO SCALE. THE

DASHED LINE REPRESENTS THE APPROXIMATE PATH OF A PADDLER COMPLETING THE COURSE FROM RIGHT

TO LEFT. BLACK CIRCLES REPRESENT THE LOCATION OF THE TWO HIGHSPEED CAMERAS WHICH WERE FOCUSED ON GATE 2 (THE DOTTED RECTANGLE)...... 94 FIGURE 2.50: CALIBRATION RIG. THE BASE FRAME (A) WAS PLACED IN THE BED OF THE COURSE AND LEVELLED

USING THE TELESCOPIC SECTIONS. TO CALIBRATE THE VOLUME THE MOVABLE T SECTION (B) WAS ATTACHED TO THE BASE FRAME ABOVE EACH LEG IN TURN AND LEVELLED...... 95 FIGURE 2.51: BOAT MARKER LOCATIONS. THE LOCATION OF THE HEMISPHERICAL MARKERS ATTACHED TO EACH BOAT. THE TOP HALF REPRESENTS A SINGLE KAYAK (K1) AND THE BOTTOM HALF A SINGLE CANOE (C1). . 96 FIGURE 2.52: KEY TO UNDERSTAND THE FORCE AND CAP RUN PROFILE. THIS SHOWS AN EXAMPLE OVERLAY

COMBINING FORCE AND CAP DATA WITH ALL OF THE COMPONENTS AND LINES LABELLED AND DESCRIBED...... 98 FIGURE 3.1: COMPARISON OF AVERAGE PLACING FOR K1M, C1M, K1W AND C2M IN A COMBINED CATEGORY

ACROSS WORLD CHAMPIONSHIPS AND OLYMPIC GAMES BETWEEN 2004 - 2008. DATA SOURCED FROM INTERNATIONAL CANOE FEDERATION (2009B)...... 101 FIGURE 3.2: COMPARISON OF AVERAGE PERCENTAGE OF THE FASTEST TIME (K1M) FOR K1M, C1M, K1W AND

C2M IN A COMBINED CATEGORY ACROSS WORLD CHAMPIONSHIPS AND OLYMPIC GAMES BETWEEN 2004 -

2008. DATA SOURCED FROM INTERNATIONAL CANOE FEDERATION (2009B). LEGEND AS PER FIGURE 3.1...... 102 FIGURE 3.3: COMBINED COURSE TIME IN SECONDS FOR BOTH RUNS (SEMI-FINALS AND FINALS) FOR THE LAST 50

YEARS FOR EACH CATEGORY. MISSING DATA ARE REPRESENTED AS A GAP. A 90 S MINIMUM COURSE TIME RULE WAS INTRODUCED IN 1997...... 103

xvi FIGURE 3.4: IMPROVEMENT OF C1M, K1W AND C2M RELATIVE TO K1M OVER THE PAST 50 YEARS (EXCLUDING

1973 FOR C1M AND K1W DUE TO CHANGES IN RIVER CONDITIONS). MISSING DATA ARE REPRESENTED AS A GAP...... 104 FIGURE 4.1: DIVISION OF TIME AROUND AN UPSTREAM GATE...... 111 FIGURE 4.2: TOTAL STROKE COUNT FOR EACH CATEGORY...... 112 FIGURE 4.3: AVERAGE STROKE TIME FOR EACH CATEGORY...... 113 FIGURE 4.4: LEFT AND RIGHT STROKE COUNT FOR EACH CATEGORY...... 115 FIGURE 4.5: LEFT AND RIGHT AVERAGE STROKE TIME FOR EACH CATEGORY...... 116 FIGURE 5.1: MEAN TRAJECTORY OF THE PADDLERS HEAD IN THE XY-PLANE FOR THE FASTEST AND SLOWEST TWO PADDLERS IN K1M AND C1M...... 124 FIGURE 5.2: CORRELATION BETWEEN TOTAL TIME AND THE BOAT TRAJECTORY FOR EACH TRIAL...... 125 FIGURE 5.3: CORRELATION BETWEEN MEAN TOTAL TIME AND THE BOAT TRAJECTORY FOR EACH PADDLER (± STANDARD DEVIATION)...... 126 FIGURE 6.1: OVERLAY OF COMPETITION ANALYSIS AND PADDLE FORCE DATA FOR RUN 1 OF 6. PLEASE REFER TO FIGURE 2.52 FOR A DESCRIPTION OF EACH COMPONENT OF THIS FIGURE...... 132 FIGURE 6.2: OVERLAY OF COMPETITION ANALYSIS AND PADDLE FORCE DATA FOR RUN 2. PLEASE REFER TO FIGURE 2.52 FOR A DESCRIPTION OF EACH COMPONENT OF THIS FIGURE...... 132 FIGURE 6.3: OVERLAY OF COMPETITION ANALYSIS AND PADDLE FORCE DATA FOR RUN 3. PLEASE REFER TO FIGURE 2.52 FOR A DESCRIPTION OF EACH COMPONENT OF THIS FIGURE...... 133 FIGURE 6.4: OVERLAY OF COMPETITION ANALYSIS AND PADDLE FORCE DATA FOR RUN 4. PLEASE REFER TO FIGURE 2.52 FOR A DESCRIPTION OF EACH COMPONENT OF THIS FIGURE...... 133 FIGURE 6.5: OVERLAY OF COMPETITION ANALYSIS AND PADDLE FORCE DATA FOR RUN 5. PLEASE REFER TO FIGURE 2.52 FOR A DESCRIPTION OF EACH COMPONENT OF THIS FIGURE...... 134 FIGURE 6.6: OVERLAY OF COMPETITION ANALYSIS AND PADDLE FORCE DATA FOR RUN 6. PLEASE REFER TO FIGURE 2.52 FOR A DESCRIPTION OF EACH COMPONENT OF THIS FIGURE...... 134 FIGURE 6.7: PADDLE FORCE DATA FORE EACH RUN NORMALISED BETEEN THE START AND SET UP FOR GATE 2. THIS

DEMONSTRATES THAT THE PADDLER HAD SIMILARITY IN THE EXICUTION OF EACH RUN WITH RESPECT TO THE

MAGNITUDE AND TYPE OF STROKES TAKEN. NOTE THAT IN THE 2ND RUN (GREEN) THE PADDLER UTILISED A DIFFERENT STROKE PATTERN TO MANOUVER AFTER GATE 1...... 135 FIGURE 6.8: PADDLE FORCE DATA FORE EACH RUN NORMALISED BETEEN THE SET UP FOR GATE 2 AND GATE 3 .

THIS DEMONSTRATES THAT THE PADDLER HAD SIMILARITY IN THE EXICUTION OF EACH RUN WITH RESPECT

TO THE MAGNITUDE AND TYPE OF STROKES TAKEN. NOTE THAT IN THE 1ST RUN (RED) THE PADDLER UTILISED A DIFFERENT STROKE PATTERN TO MANOUVER AFTER GATE 2...... 136 FIGURE 6.9: PADDLE FORCE DATA FORE EACH RUN NORMALISED BETEEN GATE 3 AND GATE 5. THIS

DEMONSTRATES THAT THE PADDLER HAD SIMILARITY IN THE EXICUTION OF EACH RUN WITH RESPECT TO THE

MAGNITUDE AND TYPE OF STROKES TAKEN. NOTE THAT IN THE 2ND AND 3RD RUN (DARK BLUE AND LIGHT BLUE) THE PADDLER UTILISED A REVERSE SWEEP STROKE PATTERN TO MANOUVER AFTER GATE 4...... 137 FIGURE 6.10: A) EXAMPLE OF THE NORMALISED FORCE PROFILE OF TYPICAL FORWARD STOKES. B) EXAMPLE OF THE NORMALISED FORCE PROFILE OF A C STROKE...... 138

xvii FIGURE 6.11: A) THE NORMALISED FORCE PROFILE OF A DRAW-FROWARD STROKE PATTERN EXICUTED ON THE

ENTRANCE TO GATE 1. B) EXAMPLE OF THE NORMALISED FORCE PROFILE OF TYPICAL REVERSE-SWEEP STOKES...... 138 FIGURE 6.12: DEMONSTRATION OF HOW THE PADDLER LOST TIME ON A SINGLE UPSTREAM GATE FROM POSITION

ONE TO POSITION FOUR. NOTE THAT THE PADDLER HAS LOST THE MAJORITY OF THE TIME PRIOR TO REACHING THE GATE LINE, BUT STILL LOSES TIME ON THE EXIT AS WELL...... 139 FIGURE 7.1: MEAN STROKE TIME FOR EACH OF THE TEN CONSECUTIVE STROKES FROM THE 12 M SPRINTS...... 144 FIGURE 7.2: AVERAGE BOAT VELOCITY WHILE USING THE SHORT AND LONG PADDLE. THE GREEN OVALS

REPRESENT THE SECTION OF THE TRIAL WHERE THE AVERAGE VELOCITY GENERATED BY THE TWO PADDLES

WAS SIMILAR. THE GREEN RECTANGLES HIGHLIGHT WHERE THE SHORTER PADDLE WAS FASTER...... 144 FIGURE 8.1:SPLIT TIMES BETWEEN EACH GATE OF THE COURSE FOR EACH OF THE ATHLETES IN THE KEY ON THE

RIGHT HAND SIDE. Y-AXIS = SPLIT TIME (S), X-AXIS = GATE NUMBER (RED = UPSTREAM GATE), COLOURED TEXT BOXES ON THE GRAPH REPRESENT AVOIDANCE AT A GATE (COLOUR = ATHLETE ON RIGHT)...... 148 FIGURE 8.2:COMPARISON OF THE SPLIT TIMES TAKEN TO NEGOTIATE A GATE RELATIVE TO THE SUBJECT (SHOWN AS GREEN LINE IN FIGURE 8.1). GATE DESCRIPTIONS AS IN FIGURE 8.1...... 149 FIGURE 8.3: UPSTREAM GATE INTERVAL. Y-AXIS = THE SPLIT TIME BETWEEN THE GATE BEFORE AND AFTER EACH UPSTREAM GATE, X-AXIS = UPSTREAM GATE NUMBER, INDICATES UPSTREAM GATE PERFORMANCE...... 150 FIGURE 8.4: STROKE COUNT BY GATE – COMPARES THE NUMBER OF STROKES EACH ATHLETE TOOK BETWEEN EACH GATE ON THE COURSE...... 150 FIGURE 8.5: STROKE COUNT – THE TOTAL NUMBER OF EACH TYPE OF STROKE TAKEN BY EACH ATHLETE DURING THE RUN. EACH RUN / ATHLETE IS REPRESENTED BY A COLOUR...... 151 FIGURE 8.6: STROKE DURATION FOR EACH ATHLETE, Y-AXIS = TIME IN THE WATER, X-AXIS = STROKE TYPE, SEPARATE PLOT FOR EACH ATHLETE...... 152 FIGURE 8.7: COMPARISON BETWEEN A RIGHT HANDED C1M PADDLER AND A LEFT HANDED C1M PADDLERS FOR

THE ENTIRE RACE. HIGHLIGHTS THE SPECIFIC SECTIONS OF THE COURSE WHICH FAVOUR / REQUIRE THE RIGHT HAND, THOSE THAT FAVOUR THE LEFT HAND AND THOSE WHICH EITHER COULD BE USED...... 153 FIGURE 8.8: THE SPLIT TIMES TAKEN EITHER SIDE OF UPSTREAM GATES. THE POSITIONS ONE TO FOUR LABELLED IN THE FIGURE DIVIDE THE GATE INTO FOUR SECTIONS...... 154 FIGURE 8.9: FULL RACE PROFILE – TIME LINE OF ALL THE EVENTS CODED FOR THE ENTIRE RACE...... 155 FIGURE 8.10: SECTION 1 AND SECTION 2 RACE PROFILES – ZOOMED IN SECTION OF FULL RACE PROFILE FOR FIRST

TWO SECTIONS OF THE RACE. START OF EACH SECTION IS RESET TO THE SAME GATE SO ALL ATHLETES START EQUAL...... 156 FIGURE 8.11: SECTION 3 AND SECTION 4 RACE PROFILES – ZOOMED IN SECTION OF FULL RACE PROFILE FOR FIRST

TWO SECTIONS OF THE RACE. START OF EACH SECTION IS RESET TO THE SAME GATE SO ALL ATHLETES START EQUAL...... 157 FIGURE 9.1: THE MODEL OF THE CRITICAL VARIABLES IN A CANOE SLALOM COMPETITION AS PRESENTED IN

CHAPTER 1, HIGHLIGHTING THE VARIABLES WHICH WERE THE FOCUS OF THIS RESEARCH AND THE TOPICS PRESENTED FOR DISCUSSION IN THIS CONCLUSION...... 165 FIGURE B.0.1: SCREE PLOT PRESENTING THE EIGENVALUE FOR EACH COMPONENT FROM THE PRINCIPAL

COMPONENT FACTOR ANALYSIS...... 183

xviii LIST OF TABLES

TABLE 1.1: THE INTERNATIONAL RIVER GRADING SYSTEM (IRGS) IS A STANDARD ASSESSMENT SCHEME USED TO

RATE THE DIFFICULTY OF RIVERS AND RAPIDS FOR RECREATION AND COMPETITION USE (INCLUDING CANOE SLALOM) (AUSTRALIAN CANOEING, 2009)...... 3 TABLE 1.2: BOAT DIMENSIONS FOR EACH CATEGORY OF COMPETITIVE CANOE SLALOM AS DEFINED BY INTERNATIONAL REGULATIONS (INTERNATIONAL CANOE FEDERATION, 2009A)...... 6 TABLE 1.3: CANOE SLALOM TIME PENALTIES FOR ERRORS RELATED TO INCORRECTLY NEGOTIATING A GATE (INTERNATIONAL CANOE FEDERATION, 2009A)...... 9 TABLE 2.1: STROKE CHARACTERISTICS AND RACE DEFINITIONS...... 31 TABLE 2.2: STROKE DEFINITIONS - PURE STROKES...... 32 TABLE 2.3: STROKE DEFINITIONS - MULTI STROKES...... 33 TABLE 2.4: EXAMPLE OF THE DATA COLLECTED AND STORED WHEN SYNCHRONISING TWO VIDEOS. THE DATA

IDENTIFIES THE VIDEO NAME, ITS STORAGE LOCATION, THE NUMBER OF FRAMES PER SECOND IT CONTAINS

AND THE SYNCHRONISATION DATA. SYNC1 AND SYNC2 REPRESENT THE TIME IN SECONDS WITHIN EACH

VIDEO FILE WHERE THE SYNCHRONISATION POINT OCCURS. THE OFFSET REPRESENTS THE TIME IN SECONDS THAT THE BEGINNING OF EACH VIDEO FILE IS FROM THE START OF THE FIRST VIDEO...... 38 TABLE 2.5: AN EXAMPLE OF THE RAW DATA COLLECTED USING CAP WHEN ANALYSING A SINGLE CANOE SLALOM COMPETITION RUN. EACH DATA POINT ADDS A NEW ROW TO THE SPECIFIC VARIABLES COLUMN OF DATA. . 54 TABLE 2.6: INTRA-OBSERVER VARIATION FOR TEMPORAL DATA WHEN COMPLETING REPEATED ANALYSES USING CAP (ALL VALUES PRESENTED IN SECONDS)...... 56 TABLE 2.7: INTER-OBSERVER VARIATION FOR TEMPORAL DATA WHEN COMPLETING REPEATED ANALYSES USING CAP (ALL VALUES PRESENTED IN SECONDS)...... 56 TABLE 2.8: AVERAGE INTRA-OBSERVER STROKE IDENTIFICATION PERCENTAGE WHEN COMPLETING REPEATED ANALYSES USING CAP...... 57 TABLE 2.9: INTER-OBSERVER STROKE IDENTIFICATION PERCENTAGE WHEN COMPLETING REPEATED ANALYSES USING CAP...... 58 TABLE 2.10: INTRA- AND INTER-OBSERVER AVOIDANCE IDENTIFICATION PERCENTAGE WHEN COMPLETING REPEATED ANALYSES USING CAP...... 58 TABLE 2.11: TEMPORAL PADDLE FORCE VARIABLES DESCRIBING THE TIMING OF EVENTS WITHIN A PADDLE STROKE CYCLE...... 82 TABLE 2.12: FORCE (N) RELATED PADDLE FORCE VARIABLES ANALYSED WITHIN A PADDLE STROKE CYCLE...... 82 TABLE 2.13: IMPULSE (NS) RELATED PADDLE FORCE VARIABLES ANALYSED WITHIN A PADDLE STROKE CYCLE. .. 83 TABLE 2.14: PADDLE MEASUREMENTS (FROM TIP OF PADDLE) USED IN THE EQUIPMENT SET UP STUDY ...... 99 TABLE 3.1: COMBINED RESULTS FOR K1M AND C1M AT THE 2004 OLYMPIC GAMES IN ATHENS. RESULTS SOURCED FROM THE INTERNATIONAL CANOE FEDERATION (2009B)...... 102 TABLE 4.1: CATEGORY COMPARISON FOR MEAN RUN TIMES AND VARIATION...... 107 TABLE 4.2: CATEGORY COMPARISON FOR MEAN SPLIT TIMES...... 108 TABLE 4.3: COMPARISON BETWEEN STRATEGIES USED BY MEN‟S CANOE PADDLERS FOR GATE 10...... 109

xix TABLE 4.4: COMPARISON BETWEEN STRATEGIES USED BY MEN‟S KAYAK AND MEN‟S CANOE PADDLERS FOR GATE 13 AND 14...... 110 TABLE 4.5: DIVISION OF TIME AROUND AN UPSTREAM GATE...... 110 TABLE 4.6: MEAN NUMBER OF ERRORS AND AVOIDANCE AROUND GATES...... 117 TABLE 5.1: PEARSON‟S TWO-TAILED CORRELATIONS TO TOTAL TIME...... 125 TABLE 5.2: PEARSON‟S TWO-TAILED CORRELATIONS AND LINEAR REGRESSION EQUATIONS FOR EACH PARTICIPANT BETWEEN TOTAL TIME AND DISTANCE BOAT TRAJECTORY...... 126 TABLE 8.1: EMAIL OF SUPPORT FROM AUSTRALIAN INSTITUTE OF SPORT (AIS) HEAD COACH FOR CANOE

SLALOM, MIKE DRUCE AFTER PRESENTATION AND FEEDBACK SESSION WITH ATHLETES. THE SESSION AIMED

AT EDUCATING THEM ON HOW TO INTERPRET THE INFORMATION SO IT COULD BE USED IN THEIR TRAINING...... 160 TABLE B.1: TOTAL VARIANCE EXPLAINED BY EACH COMPONENT FROM THE PRINCIPAL COMPONENT FACTOR ANALYSIS...... 184 TABLE B.2: COMPONENT MATRIX FROM PRINCIPAL COMPONENT FACTOR ANALYSIS...... 185 TABLE B.3: CORRELATION MATRIX FROM THE PRINCIPAL COMPONENT FACTOR ANALYSIS...... 186

xx LIST OF APPENDICES

APPENDIX A INFORMED CONSENT – TESTING PROCEDURES FOR CANOE SLALOM SIMULATED COMPETITION ...... 176 APPENDIX B PRINCIPAL COMPONENT FACTOR ANALYSIS FOR VARIABLES RELATING TO UPSTREAM GATE PERFORMANCE ...... 181 APPENDIX C INFORMED CONSENT – TESTING PROCEDURES FOR FLATWATER SPRINT ANALYSIS ...... 188

xxi DEFINITIONS AND NOMENCLATURE

1. Canoe Slalom Rapid – a section of water that has become turbulent as a result of the underlining shape, structures of the river bed, narrowing of the river, sharp bends or sudden drops.

Whitewater – a section of the river that contains many rapids. Due to the turbulent nature of rapids, they cause the water to trap air and as a result take on a white appearance.

Kayak – a decked boat that competitors sit in and is propelled by a double bladed paddle.

Canoe – a boat that competitors kneel in and is propelled by a single bladed paddle. Typically, the boat is non-decked however in competitive canoe slalom canoes are decked.

Canoe Slalom – The sport of slalom canoeing and kayaking which is performed on whitewater. Recreationally this sport involves negotiating natural hazards of a river. Competitively canoe slalom involves negotiating natural and artificial obstacles on a specific whitewater course defined by gates.

International River Grading System – used to assess the degree of difficulty of a rapid and / or river. It is not an absolute scale, does not indicate the full extent of hazards that may be encountered on a river, and may change with river level or remoteness. Rivers using this scale are graded between 1, which are classified as easy, and 6, which means they are extremely technically difficult and unpredictable.

Grade 1 (Easy): Slow to medium flowing water with very small, regular waves or riffles. Relatively few obstacles, with an easy path to find and follow. Suitable for novices.

Grade 2 (Medium): Rapids are straightforward with medium sized, regular waves. The path through rapids can be clearly seen from the water and is often indicated by well-defined chutes or Vs of water. There are some obstacles that require manoeuvring around, but paddlers with a good command of basic strokes can easily miss them.

xxii Grade 3 (Difficult): Rapids have moderate, irregular waves and strong currents. Manoeuvring is required to follow the preferred route. Small to medium sized stoppers may have to be negotiated. The route is difficult for inexperienced paddlers to see and scouting is advisable. Suitable for experienced whitewater paddlers, with the ability to Eskimo roll an advantage.

Grade 4 (Advanced): Rapids have large waves and powerful, confused currents. Drops are big and stoppers can be large and unavoidable. Fast manoeuvres maybe needed to negotiate the rapid. The route is not clear, and scouting may be needed. Suitable only for very experienced whitewater paddlers with consistent skills and reliable Eskimo rolls.

Grade 5 (Expert): Typified by extremely long, obstructed or powerful rapids. Rapids may contain very large unavoidable drops, waves, and stoppers with turbulent and unpredictable currents. Fast and accurate manoeuvring is necessary. Eddies may be very small, turbulent and scarce. The route is complex and scouting is highly recommended. Suitable only for expert paddlers, who are willing to accept the higher level of risk. Rolling in adverse conditions is essential. Swimming is very dangerous.

Grade 6 (Extreme): Rapids are extremely technically difficult, powerful and unpredictable. They are rarely paddled, and if they are paddled successfully they are usually downgraded to Grade 5 plus. The river cannot be paddled without severe risk to life (Australian Canoeing, 2009).

2. Canoe Slalom Competition Gates – consist of one or two poles that are suspended a minimum of 1.2 m apart with the lower end of the pole approximately 20 cm above the water so that the water does not affect pole motion. Downstream gates are painted with green and white rings and upstream gates are painted with red and white rings (Figure 1.1). The bottom ring is always white and each painted ring is 20 cm long. Poles must be of sufficient weight to dampen motion caused by wind. All the gates must be negotiated in numerical order and in accordance with the direction indicated by their colour and the number panels (Figure 1.1) (International Canoe Federation, 2009a).

Gate Line – is the imaginary line between the two poles which make up a gate. With the current one pole system, a second pole on the bank is required to define this line.

xxiii

Undercutting a gate – is when a paddler takes advantage of the difference in height between the bottom of the gate pole and the water by moving underneath it. The most extreme example of this is when a paddler passes the gate with the majority of their body and boat outside of the gate, but with their head (usually flat against the boat) and a small section of boat on the correct side of the gate.

Touch – when any part of the paddler, paddle or boat comes in contact with either gate pole. The paddler receives a two-second time penalty for touching which is added to their time at the end of their run. Each gate can be awarded only one touch penalty regardless of how many times a competitor knocks the same pole.

Incorrect Negotiation – where the paddler goes through the gate in the wrong direction, in the wrong order or misses the gate entirely. A paddler is awarded a 50 s time penalty for incorrect negotiation which is added to their time at the end of their run.

Upstream – opposite to the main flow of the water (uphill). Upstream gates are red and white poles requiring the paddler to paddle around the gate to the right or left (based on positioning relative to the bank) and then pass through from the downstream direction.

Downstream – in the direction of the main flow of the water (downhill). Downstream gates are green and white poles through which a paddler can continue as part of his / her downstream line through the course.

3. Canoe Slalom Techniques Ferry – the paddler moves across the flow of the river by paddling diagonally to the flow. They don‟t move upstream or downstream during the crossing. The paddler faces upstream and paddles forwards to maintain their position relative to the bank

Back-ferry – the same as a ferry but the paddler faces downstream and paddles backwards to maintain their position relative to the bank.

xxiv Offside – is a stroke taken in by a C1 paddler where their lower hand and the blade crosses the centreline of the boat (e.g. left hand at the bottom of the paddle and paddle on the right side of the boat).

Onside – is a stroke taken in by a C1 paddler where their lower hand and the blade remain on the same side of the centreline of the boat as the shoulder they are attached too (e.g. left hand at the bottom of the paddle and paddle on the left side of the boat).

4. Research A number of specific definitions were developed for the purposes of this research. These definitions are listed here but are described in depth in Chapter 2 Section 2.2.1. Australian Institute of Sport slalom coaches worked with the author to develop a definition set that covered performance variables obtainable from video footage of canoe slalom competition: Gate split times (time taken between gates), Touched and missed gates, Turn times, Major and minor avoidance of a gate or poles, Rolls, Paddle in and out of water times, Paddle stroke categorisation.

Pure strokes were identified as strokes that have one predominant phase and included the following strokes: forward, C, draw, sweep, reverse sweep, reverse, tap, brace, punt, side draw and steering. However, many strokes taken during a slalom competition are a mixture of different types of strokes. Therefore, strokes that fell into more than one category but were not multi-strokes were defined by what the predominant action of that stroke was.

xxv paddle did not leave the water, was defined as a „draw-forward‟ and the stroke count for this was one. An exception to this was made for the C1 paddle stroke with each time the pressure of the blade on the water was released was counted as the end of a stroke. If the blade was moved to a new location without pressure on the blade and followed by re-applying of pressure, even if it did not leave the water, it was counted as a new stroke. This was a result of C1‟s natural inability to remove the paddle from the water during the recovery phase on their offside strokes. The strokes that were included as multi-strokes were: draw-forward, reverse sweep-forward, forward-reverse sweep, draw-draw, reverse sweep-draw, draw- sweep, forward-sweep and major / minor strokes.

Turn – a turn was defined as up to four split times recorded around an upstream gate depending on the camera angle and course design. Turn times one and four were defined as the point where the centre of the boat passed parallel to the gate line. Turn times two and three were defined as the point where the centre of the boat passed a line perpendicular to the gate line originating from the inside pole.

xxvi PUBLICATIONS AND PRESENTATIONS BY THE CANDIDATE RELEVANT TO THE THESIS

Peer Reviewed Publications: Hunter, A. (2009). Canoe slalom boat trajectory while negotiating an upstream gate. Sports Biomechanics, 8(2), 105-113.

Hunter, A., Cochrane, J., & Sachlikidis, A. (2008). Canoe slalom competition analysis. Sports Biomechanics, 7(1), 24-37.

Hunter, A., Cochrane, J., & Sachlikidis, A. (2007). Repeatability of canoe slalom competition analysis. Sports Biomechanics, 6(2), 153-167.

Conference Presentations: Hunter, A., Cochrane, J., & Sachlikidis, A. (2009). Tactics employed by elite canoe slalom paddlers. Presented at the Australian Sports Medicine Conference, Batemans Bay, Australia.

Hunter, A., Cochrane, J., Sachlikidis, A., & Böhm, M. (2006). Canoe slalom competition analysis. Presented at the University of Canberra Corroboree, Canberra, Australia

Hunter, A., Cochrane, J., Sachlikidis, A., & Böhm, M. (2006). Canoe slalom competition analysis. Presented at the National Elite Sport Conference, Canberra, Australia.

Hunter, A. (2005). Biomechanical analysis of simulated canoe slalom competition. Presented at the National Elite Sport Conference, Canberra, Australia.

vi CHAPTER ONE

1 INTRODUCTION

This chapter provides the background (Sections 1.1, 1.2 and 1.3) for the statement of the research problem (Section 1.4) and the scope of the study (Section 1.5) to be addressed in this thesis.

1.1 Historical Perspective Canoes and kayaks are among the oldest and most basic forms of transportation in the world. They played a vital role in many early cultures, such as native Americans who utilised birch bark canoes for hunting, migrating and transportation of cargo (furs) for trade (Vaillancourt, 1984). The native people in the Arctic (Greenland and Eastern Canada) used kayaks for seal, caribou and bird hunting as well as fishing (Heath & Arima, 2004). Kayaks are propelled by double bladed paddles. They are decked boats in which the paddler sits (International Canoe Federation, 2009a). Canoes are propelled by a single-bladed paddle. They generally are not decked boats in which the canoeist kneels or sits (International Canoe Federation, 2009a). The earliest canoe and kayak designs have been modified in many ways. The materials used in construction have changed from animal skin or bark as the cover, and bone or timber as the structure, to fibreglass, carbon fibre and epoxy resin.

Canoes and kayaks are widely used today for fishing, transportation and recreation. For example, in Sri Lanka 70,000 canoes are still in use by fishermen (Gulbrandsen, 1990). The majority of these canoes are motorised, but around 12,000 non-motorised boats are still in use (Gulbrandsen, 1990). In developed countries such as Australia, canoes and kayaks are most often used for recreational activities like touring and competition (Australian Canoeing, 1999). Touring exists in a number of forms including flatwater, whitewater and sea kayaking (Australian Canoeing, 1999).

Modern competitive canoe and kayak events take place in a variety of different water conditions, ranging from flatwater, to open ocean, to fast-flowing rivers. Canoe and kayak competitions that are held on flatwater include marathon and sprint racing. Marathon competitors race over distances greater than 10 km, with the traditional distance for the open

1 men‟s category being the same as the running marathon (42 km). In sprint kayaking competitors race over 200 m, 500 m, 1000 m, 5 km and 10 km (500 m and 1000 m races are the only flatwater distances contested at the Olympic Games at present, but a 200 m event will be introduced in the 2012 Olympic Games in London). Canoe polo is a form of flatwater competition (often held in a swimming pool) where paddlers compete as teams in short, agile boats with the aim of getting a ball into a goal, similar to waterpolo. Ocean racing, whitewater river racing and slalom competitions have also become popular with specific boats designed to perform in these challenging conditions. Ocean racing boats include surf skis, sea kayaks and outrigger canoes. Whitewater competitions are as varied as the environment in which they are held and include rodeo and freestyle where paddlers perform tricks and manoeuvres in their kayak on a single rapid to earn points from judges. Down river racing is a time trial over a predetermined whitewater course but paddlers can choose any path down the river from start to finish.

Slalom competition is a time trial that requires paddlers to negotiate a series of gates positioned on a section of river with numerous rapids (Australian Canoeing, 2004). To ensure that each competition is of similar difficulty, rivers are classified using a standard grading system. The International River Grading System (IRGS) provides an assessment scheme and is widely used, for both competition and touring and thus has aspects which are influenced by the isolation of a river (inaccessibility for emergency services) which do not exist in most competition environments. It is not an absolute scale and does not indicate the full extent of hazards that may be encountered on a river. Furthermore, the IRGS scale may change with river level or isolation of a river (Australian Canoeing, 2009). The IRGS is a rating between 1 and 6 (Table 1.1) and international slalom competition occurs usually on a man-made or natural flowing water course of Grade 2 or 3 (Endicott, 2006).

Table 1.1: The International River Grading System (IRGS) is a standard assessment scheme used to rate the difficulty of rivers and rapids for recreation and competition use (including canoe slalom) (Australian Canoeing, 2009). Grade Difficulty Description 1 Easy Slow to medium flowing water with very small, regular waves or ripples. Relatively few obstacles, with an easy path to find and follow. Suitable for novices.

2 Medium Rapids are straightforward with medium sized, regular waves. The path through rapids can be clearly seen from the water and is often indicated by well-defined chutes or Vs of water. There are some obstacles but paddlers with a good command of basic strokes can easily manoeuvre around them.

3 Difficult Rapids have moderate, irregular waves and strong currents. Manoeuvring is required to follow the preferred route. Small to medium sized stoppers may have to be negotiated. The route is difficult for inexperienced paddlers to see and scouting is advisable. Suitable for experienced whitewater paddlers, with the ability to Eskimo roll an advantage.

4 Advanced Rapids have large waves and powerful, confused currents. Drops are big and stoppers can be large and unavoidable. Fast manoeuvres maybe needed to negotiate the rapid. The route is not clear and scouting may be needed. Suitable only for very experienced whitewater paddlers with consistent skills and reliable Eskimo rolls.

5 Expert Typified by extremely long, obstructed or powerful rapids. Rapids may contain very large unavoidable drops, waves, and stoppers with turbulent and unpredictable currents. Fast and accurate manoeuvring is necessary. Eddies may be very small, turbulent and scarce. The route is complex and scouting is highly recommended. Suitable only for expert paddlers, who are willing to accept the higher level of risk. Rolling in adverse conditions is essential. Swimming is very dangerous.

6 Extreme Rapids are extremely technically difficult, powerful and unpredictable. They are rarely paddled, and if they are paddled successfully they are usually downgraded to Grade 5 plus. The river cannot be paddled without severe risk to life.

The first canoe slalom race was a local race held in Switzerland in 1932 on Lake Hallwyl, rather than on a graded whitewater river. A year later the first canoe slalom race held on whitewater was contested on the Aar River, a Grade 2 to 3 river near Ruperswiler in Switzerland. Canoe slalom was described as "a whitewater test" with the idea originating from skiing, where terms like "winter, snow and ski slalom" were modified into "summer,

3 water and canoe slalom" (International Canoe Federation, 2006). In 1934, the Swiss ran a national slalom championship event on the Aar River. Competitors had one run on two separate courses, each about 500 m long with “gates” consisting of buoys and poles rising out of the water around which athletes had to navigate without collision. The person with the fastest time received 100 points (Endicott, 2006). Points were subtracted for slower times and for hitting the buoys or gates using a subjective system where a “light touch” cost one penalty point and a “heavy touch” cost as much as three points. In the event of capsize the competitor was allowed to repeat the run with a 20-point penalty (Endicott, 2006).

The first slalom event to be held in Austria was organised by the Ister Kayak Club of Vienna on the Muhtraisen at St. Georgen (Endicott, 2006). The Austrians suspended coloured poles from ropes across the river similar to the gates used today. They calculated scores by adding penalty points to the running times, a procedure still used today. A two-second penalty was given for touching a gate with the paddle and four seconds for touching it with the boat or body. Competitive canoe slalom events continued through the early years of World War II including an event in 1941 near Paris in which both canoes and kayaks participated in the same event for the first time. By 1944 all whitewater activity had stopped due to the war and the “Internationale Reprasetantshaft fur Kanusport” (IRK) ceased to exist and was replaced with the International Canoe Federation (ICF), which still serves as the governing body today (Endicott, 2006).

After the end of World War II, competitive slalom resumed, driven by the Swiss who organised a large international slalom event in Geneva in 1947, which was advertised as a World Championships. About 70 foldboat (kayak) and canoe competitors from Switzerland, Luxembourg, France, Belgium, Austria and Czechoslovakia raced, but the International Canoe Federation (ICF) refused to sanction the event. In 1949, slalom was considered an internationally competitive event by the ICF and the first slalom world championships were held in Geneva, Switzerland. It was contested by seven nations with 81 competitors across men‟s single kayak (K1M), women‟s single kayak (K1W), men‟s single canoe (C1M) and men‟s double canoe (C2M) categories (Endicott, 2006). The course consisted of 12 gates on a Grade 1 to 2 section of the Rhone River, which was technically quite easy (Table 1.1). At the time, more difficult grades were being paddled recreationally. The winners were Othmar Eiterer in K1M from Austria, Heidi Pillwein in K1W from Austria, Pierre D‟Alencon in C1M from France and Michel Rousseau and Jacques Duboile in C2M from France (Kamber, 2008).

Between 1949 and 1960, slalom courses used rivers of increasing difficulty and greater numbers of gates. Most gates were set up so that paddlers had to negotiate them in a specific way, including upstream and downstream with mandatory forward, upstream and reverse gates. “Free gates” were also used which could be negotiated in any direction. Most courses also included a “360-degree pole” which was a solitary pole coloured red, white or green. The pole was hung in an eddy and paddlers were required to complete a 360-degree turn around the pole. In addition, courses included the “barrier”, a set of yellow parallel poles that formed a line across the river to represent a fallen tree obstructing part of the river. Paddlers were required to perform a manoeuvre called a “back-ferry”. This involved the paddler facing downstream and paddling backwards against the flow to maintain their position. When the paddler angled their boat slightly relative to the flow of the river, this caused the boat to move across the river without travelling up or downstream. The rules required competitors to back- ferry from one side of the course to the other without their bow dropping downstream of the barrier. If the bow of the boat violated this, a 100-second penalty was given. The “barrier” was last used in World Championships in 1957 (Endicott, 2006).

Between 1949 and 1972, fibreglass reinforced plastic boats replaced folding and rigid canvas boats. These structural changes improved the performance of paddlers by reducing the weight, increasing the strength and rounding the shape of the boats. “As a one off” in 1972, canoe slalom was contested at the Olympic Games in Munich. Slalom rules were simplified between 1972 and 1992 with changes to the penalty scoring and the two-run system, where final placings were determined by the cumulative times of two runs. In addition, the C2 boat setup was altered so that the paddlers were repositioned to the middle of the boat and the minimum width of the C1 was changed to 0.7 m making it difficult for a non-slalom paddler to distinguish between a K1 and a C1. Canoe slalom was reintroduced to the 1992 Olympic Games and has been contested at every Olympic Games since then. Further changes to the penalty scoring and the two run system were made in 2009 (International Canoe Federation, 2009a). These changes were made so that only the better score of the two heat runs determines progression to the semi-final. In addition, the semi-final and final runs are judged separately. These changes require competitors to produce one fast heat run to qualify for the semi-finals (20 paddlers progress), then a fast semi-final run to qualify for the final (10 paddlers progress). Ultimately, the final run determines a paddler‟s place in the event. A single pole gate was also introduced, however a second pole must be placed on the bank of

5 the river to create the gate line (International Canoe Federation, 2009a). This is the current format used in competition today.

1.2 Current Slalom Competitive Environment Today, competitive canoe slalom is an event where the emphasis is on speed, but with the added complexities of gates that must be navigated in a specific sequence. Once the gate sequence has been set up competitors cannot practice on the course. Usually a competition consists of heats with two runs, one semi-final run and one final run. The course for the semi- final and final is the same but may be different from the course used in the heats. The fastest 20 boats in each category based on the fastest time including penalties from either run in the heats progress to the semi-finals. The start order of the semi-final run is determined by the times from the heats, with the slowest boat starting first. The top ten competitors in the semi- finals progress to the final where again the slowest boat starts first. The final score for each competitor is determined from the time taken during their final run to which any penalty seconds accumulated during the run are added. The fastest time defines the winner.

The racing categories recognised internationally are K1M, C1M, K1W and C2M (International Canoe Federation, 2009a), with women‟s single canoe (C1W) and women‟s double canoe (C2W) added in January 2007 (Canoe Kayak Canada, 2007). The C1W category was contested during the 2009 international season and there was a demonstration event at the 2009 World Championships in Spain. To prevent excessive bias in performance due to revolutionary equipment development, competitor‟s boats must conform to specific regulations. The international regulations for the dimensions of a modern canoe and kayak are listed in Table 1.2.

Table 1.2: Boat dimensions for each category of competitive canoe slalom as defined by international regulations (International Canoe Federation, 2009a). Boat Minimum Length (m) Minimum Width (m) Minimum Weight (kg) K1 3.50 0.60 9 C1 3.50 0.65 10 C2 4.51 0.75 15

Today, canoe slalom courses consist of natural and artificial obstacles that create complex water flows and are usually classified as Grade 3 to 4 on the IRGS. Gates are placed in and around the obstacles and usually include at least one sequence which offers paddlers several

6 options to negotiate the gates. Good course designs challenge paddler skills by making use of the fluid dynamics of the water and therefore, gates are often located around eddy and wave structures to reward good technical water reading skills. The length of the course can vary between 250 and 400 m, as measured along the centre line of the channel, with 18-25 gates and should be navigable by K1M paddlers in 90 to 100 s. Six or seven gates require approach from the downstream side or against the main current and the last gate must be 15-25 m from the finish line (International Canoe Federation, 2009a). Due to the onside (e.g. left hand closest to the blade and blade on the left side of the boat) and offside (e.g. left hand closest to the blade and blade on the right side of the boat) imbalance in stroke execution imposed by the one bladed paddle of C1 paddlers, courses must provide the same conditions for right and left-handed paddlers.

In modern competition, the gates consist of one or two poles that are suspended a minimum of 1.2 m apart with the lower end of the pole approximately 20 cm above the water so that the water does not affect pole motion. Current competition allows one pole to be on the bank thus, there is only a minimum gate width. Downstream gates are painted with green and white rings (Figure 1.1). Upstream gates are painted with red and white rings (Figure 1.1). The bottom ring is always white and each painted ring is 20 cm long. Poles must be of sufficient weight to dampen any motion caused by wind. All the gates must be negotiated in numerical order and in accordance with the direction indicated by their colour and the number panels (Figure 1.1) (International Canoe Federation, 2009a).

Figure 1.1: Gate 2 represents a downstream gate (green (G) and white pole) through which a paddler can continue as part of their downstream line. Gate 3 is an upstream gate (red (R) and white poles) indicating that the paddler must paddle around the gate to the right or left (based on positioning relative to the bank) and then pass through from the downstream direction.

To negotiate a gate correctly, part of the boat and the paddler‟s head must cross a line connecting the poles at the same time, in the correct direction, without touching the gate poles with the body, the paddle, or the boat (International Canoe Federation, 2009a). If a gate is touched or incorrectly negotiated, time penalties are incurred (Table 1.3). The time penalties are harsh in proportion to the time taken for canoe slalom runs. Typically, penalties exclude the paddlers from progressing to the next level of competition.

Table 1.3: Canoe slalom time penalties for errors related to incorrectly negotiating a gate (International Canoe Federation, 2009a). Penalty Reason for Penalty Zero Penalty- Correct negotiation without fault. Undercutting a gate without seconds touching it and repeated attempts at negotiation are not penalised. The benefit of any doubt is given to the competitor.

Two Penalty- Correct negotiation of the gate, but with a touch of one or both poles. seconds Repeated touches of the same or both poles at the same gate is only (2% of an average course) penalised once.

50 Penalty-seconds Incorrect negotiation of the gate, with or without touching the poles. (50% of an average This is the maximum penalty on any gate. This penalty can also be course) given for: Intentional pushing of a gate to allow negotiation; The head breaking the gate line while entirely under water; The head breaking the gate line in the wrong direction; A gate being left out; The head breaking the line between the poles without part of the boat.

Note: Negotiation of a gate begins when the boat, the body or the paddle touches a gate pole or when part of the competitor‟s head breaks the line between the poles. Negotiation of a gate ends when the negotiation of any subsequent gate begins or the finish line is crossed.

Before a race, a demonstration run is paddled by paddlers who are not competing. The demonstration comprises of up to two boats from each category with preferably one right- handed C1M, one left-handed C1M, two K1M, two K1W and two C2M (one front left and one front right). Once the course is established a training run may be allowed, however provision of training runs is not mandatory. At most competitions, including the Slalom Racing World Championships and the Olympic Games, there is no official training run on the final course setup. Therefore, for most competitions, paddlers must visualise what the course will be like based on what they observed during the demonstration runs, previous experience of the venue and from walking the course.

1.3 Motivation for a Technical Analysis of Canoe Slalom Since the reintroduction of canoe slalom to the Olympic Games in 1992 interest in competitive slalom sports science research has increased with a progressive rise in publications in the area. Early research focused on officiating (Miao & Bi, 2001; Ritter, 1975) and course design (Peters, 1987; Schmidt, 1993), both of which cannot be influenced

9 by the paddler. Fiore & Houston (2001) investigated the incidence of injuries incurred during canoe slalom but focused on the non-competitive side of the sport. More recent research has investigated parameters related to improved performance including visualisation skills (Males et al., 1998; White & Hardy, 1998), mental arousal (Males & Kerr, 1996), energy systems (Zamparo et al., 2006), morphological characteristics of elite paddlers (Norton & Olds, 1996; Ridge et al., 2007), paddle force application (Sperlich & Klauck, 1992) and equipment setup (Ong et al., 2005). However, there remains a paucity of scientific research literature regarding canoe slalom performance in the competition and training environment.

1.3.1 Visualisation Skills While the rules in canoe slalom do not permit participants to paddle the course prior to competition, they can train at the venue in similar water conditions and they have the opportunity to watch up to eight demonstration paddlers negotiate the course (International Canoe Federation, 2009a). The demonstration runs allow paddlers to prepare visualisation plans of how they might best perform their run. Most paddlers use some form of mental imagery to practice the course before paddling it in competition. White & Hardy (1998) found that successful canoe and kayak slalom paddlers use imagery for cognitive and motivational purposes in a variety of ways during their preparation for competition. White and Hardy (1998) found that slalom paddlers used imagery while planning how to complete a course by watching other competitors. White and Hardy (1998) also found that slalom paddlers used imagery while preparing to compete by running through the course in their head (prior to the warm-up, during the warm-up and a couple of minutes prior to their run) and by walking the course. Finally, slalom paddlers used imagery while reviewing errors made by demonstration paddlers, other competitors and during their own runs (White & Hardy, 1998). However, only national level athletes were analysed and it is therefore unknown if the use of mental imagery would differentiate between the competition standard of athletes. Previous research has demonstrated that skilled gymnasts (Ille & Cadopi, 1999), dancers (Starkes et al., 1990) and figure skaters (Deakin & Allard, 1991) are able to accurately replicate observed movements. Unfortunately, no research has investigated the relationship between canoe slalom performance and the ability to replicate observed manoeuvres.

1.3.2 Mental Arousal Sports psychologists (Cox, 2006; Smith & Bar-Eli, 2007) have identified several states of mind that athletes should achieve in order to facilitate quality performances. Males & Kerr

10 (1996) investigated mental arousal and motivational state prior to racing and how this related to paddler performance. Their research showed that elite male slalom paddlers exhibit self- focused control but varied in their focus (present / future) (Males & Kerr, 1996). Paddlers were found to use visualisation, music and breathing to develop their preferred state of mind for competition. Improved performances were more likely to occur when a paddler was focused on themself and the present and had small discrepancies between their felt and preferred arousal levels (Males et al., 1998). Therefore, prior to competition paddlers should try to achieve the level of arousal which matches their preferred arousal level, otherwise they may not perform to their full potential.

1.3.3 Physiology of slalom paddlers The metabolic power distribution of skilled slalom paddlers during a race was reported to be approximately 50% aerobic and 50% anaerobic (Zamparo et al., 2006). Thus, besides training for skill acquisition and improvements in anaerobic power Zamparo et al. (2006) suggested that some high intensity, cardiovascular conditioning should be included in the training programs of athletes specialising in this sport (Michael, Rooney & Smith, 2008).

Zamparo et al. (2006) compared paddling a slalom kayak on flatwater and whitewater revealing that the total metabolic power was about 30% larger during the flatwater all-out test (1.72 ± 0.18 kW) than during the whitewater slalom race (1.35 ± 0.12 kW). This suggests that the flowing water which is a characteristic of a slalom course provides assistance in completing the run. Alternatively, the technical nature of the course may prevent the paddler performing maximally, thus reducing the total energy expenditure for the same time period.

1.3.4 Body composition of elite slalom paddlers It is common for successful athletes to possess specific physical attributes that facilitate performance in their sport. Consequently, physiological and anthropometrical profiling of athletes is common and frequently studied (Norton & Olds, 1996). In canoe slalom, while some research has been conducted on the morphology and anthropometry of paddlers (Ridge et al., 2007; Sidney & Shephard, 1973; Vaccaro et al., 1984) this has been quite limited in scope. At the 2000 Olympic Games, age, body mass, skinfolds, height, weight and sitting height, together with anthropometric measures such as limb length, breadth and girth were measured for 31 male and 12 female canoe slalom paddlers (Ridge et al., 2007). The results suggested that the top ten finishers in K1M were more compact with a smaller proportional

11 hip girth than that of other competitors in the same event. Ridge et al. (2007) suggested that being short provides slalom kayak paddlers with a lower centre of gravity, thereby increasing stability in a complex and unstable environment. However, sitting height is perhaps more important in lowering the centre of gravity of a seated or kneeling paddler. Furthermore, a shorter sitting height allows paddlers to negotiate more easily under gates set up at a standard height off the water (20 cm). Interestingly, the morphological dimensions of these elite paddlers were not significantly different to those of a reference population of non-athletes (Ridge et al., 2007) suggesting that (1) such measurements are not useful for talent identification and (2) that paddler morphology is less important than mastery of the technical aspects of the sport.

1.3.5 Technique It is possible that technique influences drag more profoundly than anthropometry in canoe slalom, as has previously been found in swimming (Komogorov & Duplischeva, 1992). Pendergast et al. (2005) reported that in swimming, rowing and flatwater kayaking, technique modifications through training reduced drag and improved overall performance. Given that canoe slalom paddlers negotiate gates in flowing water, it would be expected that technique would play a large role in performance. Slalom paddlers need to adapt rapidly their strategies in a constantly changing environment. These adaptations demand not only excellent paddling and boat control skills but also rapid problem assessment and solving abilities. Thus, an investigation into the different techniques used when negotiating a slalom course and how these relate to overall performance would provide invaluable information to coaches and athletes at all levels of the sport.

1.3.6 Force Application Paddle force analysis in flatwater sprint kayaking has advanced the sport by providing coaches and sports scientists with an objective measure for prescribing and assessing technique modifications aimed at enhancing performance (Aitken & Neal, 1992; Ho et al., 2000; Sperlich, 1995a, 1995b; Sperlich & Klauck, 1992; Stothart et al., 1987). Application of the same procedures to canoe slalom could provide a powerful tool for providing objective data about paddling technique and strategies employed (Sperlich & Klauck, 1992).

Sperlich & Klauck (1992) found that in canoe slalom, paddle force curves were related to paddler actions but suggested that obtaining objective data about the kinematics and dynamics

12 of turning movements is difficult. Thus, Sperlich & Klauck (1992) utilised video footage to determine gate splits during the run and overlayed it on the paddle force data. The results from Sperlich & Klauck (1992) suggested that objectively linking competition analysis data derived from video footage and paddle force data would provide substantial information about canoe slalom performance.

1.3.7 Repeatability of Athletic Performance Canoe slalom competitions vary considerably between venues and courses. Each venue has its own fluid dynamics that dictates to some extent, the designs of the courses at that event. The conditions are a function of water flow rates and can therefore vary between competitions at the same venue. Thus, paddlers are exposed to a wide range of conditions with respect to flow rate, obstacle location, length (250 m to 400m), number of gates (18 to 25) and location of every gate (almost anywhere on the course) (International Canoe Federation, 2009a). Environmental conditions such as wave and eddy specifics and wind conditions vary continually during competition. Consequently, consistency of performance can be severely impaired by external events within a competition and is therefore a crucial aspect of successful performance in canoe slalom. Thus, to succeed in winning, an athlete must deliver three top-level performances regardless of external environmental conditions.

1.3.8 Equipment Setup Ong et al. (2005) questioned the method that coaches use to decide optimum boat set-up for their sprint or slalom paddlers. Most paddlers take a trial-and-error approach to setting up their equipment and for many paddlers comfort takes priority over mechanical advantage (Ong et al., 2005). Zumerchik (1997) stated that choosing the correct paddle setup (length of shaft, hand position and length of blade) depends on the kayak (length, width and mass) as well as the anthropometry of the paddler (height and reach). Ong et al. (2005) suggested that to a large extent the equipment should be fitted based on the participant‟s body dimensions but cautioned that equipment set-up decisions could affect the ability of athletes to apply propulsive force to the boat.

The mechanical advantage of the paddle is of particular importance in canoeing and kayaking as it influences the ability to propel the boat. For a given paddle length, the paddler may alter the mechanical advantage of the propulsive system by changing the position of the hands when gripping the paddle shaft (Ong et al., 2005). The standard setup for hand spacing is

13 determined by holding the paddle shaft above the head with the upper arms horizontal and forearms held vertical with a right angle formed at the elbow (Rademaker, 1977). Thus altering the paddle configuration could achieve a better mechanical advantage, which in turn would improve the ability to accelerate the boat. In canoe slalom the ability to accelerate rapidly is important due to the stop-start nature of the sport as the paddler manoeuvres the boat around the course.

1.3.9 Augmented information Changes over the past 20 years in the accessibility and convenience of technology has allowed feedback from video that coaches cannot observe themselves (Bartlett, 2001). However, augmented information from computerised notational analysis such as counts of events (e.g. number of passes in a game) is unlikely to revolutionise performance (Bartlett, 2001). Hughes (1995) suggested that in the development of a time-motion analysis system for racket sports, the augmented information should focus on tactical evaluation, technical evaluation, movement analysis, creating databases and modelling because these were identified as important for success in these sports.

There are now a large number of commercial notational analysis systems available that considerably enhance the power of feedback, although most objective quantitative analysis systems have been developed for research purposes with coach and athlete feedback considered as a bonus (Hughes et al., 2007). Research has shown that the more quantitative and objective the feedback the greater effect it has on performance (Franks et al., 1983). Murray et al. (1997) found that feedback from computerised notational analysis improved squash performance by increasing the number of wining strokes and decreasing the number of errors.

Presentation of computerised feedback at the correct time and in the correct quantity plays a large part in successful learning of new skills (motor skill acquisition) and enhancement of performance (Liebermann et al., 2002; Murray et al., 1997). Advances in technology have made it possible to augment and improve the feedback coaches and athletes receive during training and competition (Liebermann et al., 2002). However much of this feedback is based on video that takes time to analyse. This limits immediate feedback to the video images themselves and restricts prompt feedback to simple kinematic and temporal data (Bartlett, 1999).

14 1.3.10 Summary of Key Intrinsic Attributes Several intrinsic paddler attributes play critical roles in determining successful canoe slalom performance. Athletes need good visualisation skills to develop successful negotiation strategies prior to their competition runs. Self-focus, achieving preferred arousal levels and an ability to cope with the metabolic power demands of canoe slalom competition (50% aerobic and 50% anaerobic) are critical for optimal performance. Slalom paddlers also need to produce high quality, successive runs to ensure progression from heats to finals. A poor run will result in elimination from competition and therefore repeatability and consistency are critical attributes. Morphologically, slalom paddlers who have a low sitting height are advantaged because this lowers their centre of gravity. This improves stability and allows them to negotiate under gates more easily. As with swimming, rowing and flatwater kayaking it is expected that in canoe slalom, technique modifications through training would reduce drag and improve overall performance. Given that canoe slalom paddlers negotiate gates in flowing water, it is expected that technique would play a large role in performance, with rapid acceleration a crucial component due to the large amount of manoeuvring required to negotiate the course. However, a paddler‟s equipment should be fitted based on their body dimensions as this could affect the ability of athletes to apply propulsive force to the boat.

1.4 Statement of the Research Problem for this Thesis Despite what is known about the attributes of successful canoe slalom paddlers, many aspects of competitive canoe slalom have not been investigated. The overall performance outcome for an athlete in a canoe slalom competition is their final placing. Due to the „knock out‟ style of qualification rounds, the performance tolerances decrease until the final, where the outcome is based on performance of a single run. Results for each round of competition are determined by the summation of the time which the paddler achieves for the course (run time) and the number of penalties accumulated during the course (penalties). There are many compounding variables that influence the time a paddler achieves and the number of penalties they accumulate for a course. These factors can be subdivided into intrinsic components over which the paddler has control and extrinsic factors that cannot be controlled by the paddler.

Figure 1.2 provides a schematic model of critical intrinsic and extrinsic variables that influence the result of a canoe slalom competition. The performance outcomes, highlighted in blue, are a direct result of all variables beneath them in the model. Variables in red and

15 orange represent external factors that neither the coach nor the athlete can control. These include course design, wind and judging errors. Course design is comprised of the flow rate of the course, the positioning of permanent and movable obstacles and the number and placement of upstream and downstream gates. Although extrinsic factors may contribute to the outcome of the race or influence the way the paddler negotiates the course they are beyond the paddler‟s control. Thus, the coach and athlete must formulate the best strategy to manage these extrinsic factors and achieve the fastest run time with the least penalties for each run.

The model highlights the variables which the paddler can influence (green). Quantitative information relating to these areas is beneficial to coaches and athletes because each can be adjusted to improve overall performance. With the exclusion of physiological, psychological and experiential factors which have been documented in the sport science literature, the green areas define the scope of this research. In particular, my research in canoe slalom concentrated on (1) defining the attributes of a successful run and how this relates to elite paddler technique with respect to strategies used to negotiate a course in competition; (2) canoe slalom specific skills such as upstream gate negotiation and the application of force to the paddle; and (3) equipment setup, particularly paddle length.

Prior to investigating the research questions, it was necessary to quantify any major differences in performance between the slalom categories to ensure correct application of subsequent research (Chapter 3).

The research questions that were addressed in this thesis are: What are the strategies employed by elite canoe slalom paddlers to negotiate a course in terms of split times, stroke rates, stroke types, turns and gate avoidance? (Chapter 4) How does the path taken by an elite canoe slalom paddler influence the time taken to negotiate an upstream gate? (Chapter 5) Can paddle force data be used to explain some of the temporal variability and events identified from competition analysis? (Chapter 6) Does the length of the paddle affect paddle force, boat acceleration and sprint time from a standing start? (Chapter 7)

Canoe Slalom Competition Result

Competition

Result from Heat One Result from Result from Semi Final Final Result from Heat Two

External improvements possible Performance outcomes Paddler dependent variables and attributes Fixed environment Each Run Run Time Penalties

Judging Path Taken Mean Velocity Errors Errors

Course Design Paddler Attributes Wind Appeal

Flow Rate Gates Obstacles Strokes Used Technique Judge Experience Variability Experience Downstream Upstream Gates Gates Physiology Psychology

Figure 1.2: A model of the critical variables in a canoe slalom competition, their interaction and their general grouping based on type. The model highlights the variables which the paddler can influence (green).

1.5 Scope of the Thesis The schematic model of the critical variables in canoe slalom competition (Figure 1.2) demonstrates that there are numerous extrinsic variables which influence performance and that the attributes of a paddler are linked directly to all aspects of competition performance (green area in Figure 1.2). The main scope of this thesis is defined by the green area with particular attention to paddler stroke usage and technique as well as paddler variability.

17 Limitations in the application of existing time-motion analysis software to canoe slalom led to the development of a competition analysis program (Chapter 2 – Competition Analysis Program (CAP) for Canoe Slalom). An operational definition set for the qualitative performance variables was developed to ensure standardisation and repeatability of the data generated by the system and the system was assessed to determine the intra-observer and inter-observer reliability (Chapter 2 – Reliability Assessment for CAP). This was important because the provision of valid, accurate and reliable performance data to coaches and athletes is critical for effective coaching and ultimately in improving performance.

The requirement to collect and analyse paddle force information during the research highlighted a lack of functionality with the existing software. Therefore, a paddle force curve analysis program for kayaking capable of collecting, processing, analysing, reporting and exporting force data from the existing and redeveloped paddle force system was developed (Chapter 2 – Paddle Force Curve Analysis Program for Slalom).

To ensure that the research outcomes of this thesis are widely applicable in the sport of canoe slalom the distribution of athletes within each category and relative to the other categories of canoe slalom were investigated. This provided information about the similarity of categories with respect to performance as well as the depth of each category through the distribution of the finalist in each category. In addition, the progressive performance of each category relative to the fastest category gives an indication of the improvement of each group over time. This can provide insight into which categories could achieve the greatest gains in the future (Chapter 3 – Analysis of Canoe Slalom Category Clustering).

Time-motion analysis of canoe slalom competition could be utilised to understand further what causes differences between paddle classes based on the strategies they use. This would provide information about the tactics and strategies employed for the course and subsections of the course (Chapter 4 – Canoe Slalom Competition Analysis).

Due to the large amount of manoeuvring required to negotiate an upstream gate, the upstream gate presents one of the most complex gate sequences and occurs six or seven times in every competition. Thus, improvement in the ability of a paddler to negotiate upstream gates could have a substantial impact on the overall race time. Therefore, investigation into how the trajectory of the boat while negotiating an upstream gate influences the time taken complete

18 the gate was completed (Chapter 5 – Canoe Slalom Boat Trajectory while Negotiating an Upstream Gate).

The tactics a paddler uses to complete the course including the strokes, manoeuvrers and path which they select influence the time taken to complete the course. Each of these movements requires the paddler to control the boat through the paddle. Therefore, the ability to utilise paddle force information and video time-motion analysis to provide canoe slalom coaches and athletes with beneficial information about performance was assessed. In addition, stroke identification, consistency between trials and analysis of how a paddler‟s errors impact on performance were assessed using paddle force information and video time-motion data (Chapter 6 – Interactions between Strategy and Paddle Force).

Many aspects of a paddler‟s equipment can be altered and it is therefore important that it is setup in the best way for optimal performance based on the paddler‟s anatomical and physiological attributes. Paddlers spend hours setting up equipment through trial and error to get the optimal performance for themselves. One critical component of the performance of the paddler is the mechanical advantage of the paddle. One factor which alters the paddle mechanical advantage is the shaft length (Sperlich, 1991). This was investigated to determine its influence on paddle force, boat acceleration and boat velocity (Chapter 7 – Equipment Setup – Canoe Slalom Paddle Length Comparison).

Information transfer to the coaches and athletes is the critical link in any analysis process as no beneficial gains can be made if the information stops with the scientist. Thus, the way this information was presented for coaches and athlete is presented in Chapter 8 – Dissemination of Information to Coaches and Athletes.

19 1.6 Thesis Layout and Structure This thesis is comprised of two methodology development sections (Development of Analysis Systems for Canoe Slalom – Chapter 2 Section 2.2), five research studies (Chapters 3 to 7), one example report (Chapter 8) and the conclusions (Chapter 9).

Methodology 1 – Details the Competition Analysis Program (CAP) for Canoe Slalom (Chapter 2 Section 2.2.1). This includes development of the CAP software, an operational definition set and the reliability analysis for CAP.

Methodology 2 – Describes the development of a Paddle Force Curve Analysis Program for Slalom (Chapter 2 Section 2.2.2).

Study 1 – Presents the findings from the Analysis of Canoe Slalom Category Clustering (Chapter 3).

Study 2 – Presents findings from Canoe Slalom Competition Analysis (Chapter 4).

Study 3 – Analyses Canoe Slalom Boat Trajectory while Negotiating an Upstream Gate and how this influences the split time around the gate (Chapter 5).

Study 4 – Utilises CAP and paddle force data to assess the Interactions between Strategy and Paddle Force (Chapter 6).

Study 5 – Presents the findings from the Equipment Setup – Canoe Slalom Paddle Length Comparison Analysis (Chapter 7).

Example Report – Dissemination of Information to Coaches and Athletes (Chapter 8).

Conclusions – Conclusions (Chapter 9).

20 CHAPTER TWO

2 METHODOLOGY

This chapter investigates available methodologies for the analysis of these paddling strategies based on what has been utilised in other sports and identifies areas where slalom specific methodologies are required; (1) customised software for time-motion analysis of canoe slalom competition and (2) paddle force curve analysis program. These methodologies were developed as part of this research and are discussed in detail in Section 2.2. Finally, the new methodologies used to analyse canoe slalom in coaching and racing environments are described.

2.1 Previous Research Methodology Competition analysis of canoe slalom remains neglected in the scientific literature. However, some methods used for analysis of other sports can be transferred to canoe slalom. This section reviews those methods considered to contain features that would provide beneficial information to elite canoe slalom athletes and coaches, and be feasible in training and racing environments.

2.1.1 Time-motion analysis Time-motion analysis uses video footage to study the modes of motion used throughout a sporting event. There is a long tradition of time-motion studies in Australia (Lyons, 2002), especially in sports such as football (McKenna et al., 1988) and squash (Flynn, 1998) where time-motion analysis has been used to identify the relationship between performance and particular skills, methods of locomotion and sections of the game or competition.

Hughes et al. (2007) stated that technique and tactics in racket sports are inter-dependent. Thus defining a player‟s technical strengths and weaknesses also defines the player‟s tactical decision-making skills (Hughes et al., 2007). This information is of vital importance to coaches in their quest to improve their athletes. Time-motion analysis has evolved from notational analysis, a process whereby a researcher / coach notes down numbers of events, timing of events, qualitative descriptors and subjective thoughts about the state of play during

21 a game to determine the technical strengths and weaknesses of their team or player. With the addition of video cameras the opportunity to analyse the footage after the event as well as slow the footage down (lapsed-time) allowed events which could not be notated in situ to be analysed after the competition. Current time motion analysis utilises video cameras and computers to provide improved repeatability (quality assurance and control), automation of the analysis process improving analysis time and longevity of data through standard collection and storage methods (Bloomfield et al., 2005).

One of the main advantages of time-motion analysis is that it offers a non-intrusive method of analysing performance during competition (Bloomfield et al., 2005). Therefore, in all sports and activities information about the movement patterns can be obtained using time-motion analysis without interfering with the competition. Accurate and reliable performance data have been identified as essential for effective coaching (Davies, 2003), therefore for time- motion analysis to be used effectively it must also meet these requirements. It has been suggested that this type of quantitative analysis with demonstrated and adequately reported reliability and validity measures should supplement subjective judgements of performance (Schokman et al., 2002).

Hughes & Bartlett (2002) found that 70% of time-motion analysis papers did not report the reliability of any variables and reported statistics were derived using questionable processes. For example, large discrepancies have been reported in the distances covered by football players as a result of different approaches to time-motion analysis (Bloomfield et al., 2005). Before beneficial comparisons can be made using time-motion analysis, standardised operational definitions identifying skills, methods and phases of the sport need to be developed (Pearce, 2005). The purpose of definition standardisation is to enable investigators to operationalise standard methods to produce comparable results across athletes and events. Accordingly, for canoe slalom to gain valuable information from time-motion analysis, a definition set needs to be developed and the methods for data collection need to be assessed for reliability.

Many major international and national sporting competitions, as well as some local and youth competitions, are broadcast live or recorded post-event either in their entirety or as edited highlights. Krosshaug et al. (2005) suggested that this provides an opportunity to collect video footage of sports injuries and analyse their mechanisms. Davies (2003) recognized that

22 quick, accurate and measurable performance analysis from video could be used for player selection, talent identification, player development and instant competition feedback and analysis. Many team sports use event coding systems to provide detailed individual, unit and team analysis, as well as to detect strengths and weaknesses of opponents (Lyons, 2002). Sports scientists, coaches and athletes record those sporting events that are not televised for qualitative and quantitative performance review. Given the correct circumstances, this footage provides competition specific information that can be used for research purposes.

Time-motion analysis has also been used to answer specific research questions such as the relationship between unforced errors and the outcome of a squash game (Flynn, 1998). Spencer, et al. (2004) utilised time-motion analysis to characterise changes in movement patterns with specific attention to repeated-sprint ability of elite male field-hockey players during an international game. Application of this type of research to canoe slalom could lead to the identification of variables relating to performance.

2.1.2 Need for Customised Analysis Software for Canoe Slalom Time Motion Analysis Messersmith was an early twentieth century pioneer of real-time recording of performance (Lyons, 1997). His work in basketball used innovative methods to record the distance travelled by basketball players in games (Messersmith & Corey, 1931). Hughes & Franks (1997) reported that the early real-time recording of performance used pen and paper systems. Downey‟s (1977) recording of badminton play is an excellent example of this approach. Early adopters of video technology developed lapsed-time analysis. Sanderson & Way (1979) used video to analyse successful and unsuccessful patterns of play in squash. However it took more than 5 hours to learn how to use this system and a further 40 hours to analyse the data from a single match (Hughes et al., 2007). The availability of microcomputers transformed notational analysis into modern lapsed-time time-motion analysis (Hughes et al., 2007). Hughes (1985), for example, used this approach to improve the notational analysis of squash pioneered by Sanderson & Way (1979).

In the 1980s with increased access to computers and graphical software, there were significant developments in the visualisation of processed data (Hughes et al., 2007). Reports evolved from two-dimensional representations which were not easy to understand, to coloured three- dimensional histograms that could be rotated and viewed from different angles making the results easier to understand (Hughes et al., 2007).

Since the early development of software for notational analysis, a large number of commercial systems have been developed such as Dartfish® (Fribourg, Switzerland), Silicon Coach® (Dunedin, New Zealand) and SportscodeTM (Warriewood, Australia). Many of these offer a large number of features for video manipulation and analysis but because they are generalised for use across a number of different sports, they are not amenable to specific reporting or data analysis methods. Prozone® (Leeds, England) and Amisco® (Nice, France) have been developed for team sports such as soccer and rugby. These systems offer a wide variety of sport specific statistical and analysis tools. However, for lower profile sports such as canoe slalom, no specific software and no sports specific definition set has been developed to suit the demands of the sport.

2.1.3 Motion Analysis Three-dimensional motion analysis from digitisation of video involves setting up a minimum of two stationary cameras focused on the same area (Shapiro, 1978). A calibration frame is placed in view of all cameras to provide reference coordinates for the calculation of the direct linear transformation. Once the calibration is completed the camera views cannot be altered without recalibration. Reference markers are attached to the participant or equipment and then trials are recorded. Although video based technology is information rich, substantial amounts of time are required to analyse the information and convert it to a format that allows inter- and intra-athlete comparisons. Obtaining positional information from video is also limited to a defined and calibrated space. The accuracy of the data are dictated by the resolution of the cameras.

The conversion of video footage to a three-dimensional (3D) frame defined by the calibration coordinates is now done electronically. Numerous commercial software packages such as Ariel Performance Analysis System (APAS, CA, USA), Vicon Motus (Vicon, Oxford, UK) and KWON3D (VISOL Inc, Korea) are available to automate this process. All these packages use the same methodology; digitising control (known) points on the calibration frame, followed by digitising reference markers in each still frame of video. Direct Linear Transformation (DLT) resolves the relationship between the two-dimensional (2D) digitised coordinates and the 3D image space coordinates (x, y, z) on the calibration frame using simultaneous equations (Abdel-Aziz & Karara, 1971). This relationship can then be used to

24 determine the 3D coordinates of the reference markers attached to the object of interest (Ariel, 2002).

Technological advances have led to three-dimensional motion analysis becoming more commonly used to analyse sports performance (Davids et al., 2000). In sports with similar manoeuvring characteristics to canoe slalom such as slalom skiing, snowboarding and snow biking (Jelen & Jandová, 1999), motion analysis has been used to obtain insight on the biomechanics of an athlete negotiating a turn around an obstacle. Jelen and Jandová (1999), for example, used a cinematographic method to determine the kinematic parameters of the snowbiker and snowbike through a series of slalom gates using high frequency cinematography cameras. This footage was digitised and analysed to determine the turn phases using a model developed for slalom skiing (Jelen & Pribramsky, 1987). The turn was broken down into initiation (11% of the time for the entire turn), steering I (36%), steering II (31%), end (18%) and transitional phases (4%). The radius, velocity and centrifugal force of each phase of the turn were calculated. Results revealed that snowbikers experience more than 4487 N or 4 G (G in this context represents g-force which is a measure of the acceleration occurring normalised to gravitational acceleration). The researchers suggested that these forces may influence subsequent reactions of the snowbikers and their ability to execute the next turn in an optimal way which in turn will impact on race time (Jelen & Jandová, 1999).

Currently there is a paucity of research regarding the boat kinematics (i.e. the path taken by the boat) in canoe slalom. Canoe slalom is similar to snowbiking with regards to athletes manoeuvring through a series a gates. Therefore, similar 3D analysis methods to those used for snowbiking could prove beneficial in determining biomechanical factors which impact on canoe slalom performance, even though the velocities involved in canoe slalom are much lower.

2.1.4 Global Positioning Systems Global Positioning Systems (GPS) provide positional and time information using doppler shift of signals from satellites in a fixed orbit (Salmon, 1998). The number of satellites which the GPS simultaneously tracks influences the accuracy of its positioning estimate. Typically, 12 satellites can be tracked at the same time, but restricted sky visibility limits tracking performance.

In addition to the 12 satellites being tracked there are a number of different types of receiver configurations which integrate data from alternate signals to correct the satellite signals. These include non-differential GPS (distance accuracy (da) = 15 m), differential GPS (da = 3-5 m), Wide Angle Augmentation System (WAAS) enabled GPS (da < 3 m) and carrier wave differentiation using two neighbouring receivers (da = cm accuracy) (Witte & Wilson, 2004). These configurations improve accuracy but all systems apart from WAAS are large and not practical for measurement of human performance (Witte & Wilson, 2004) in short distance or low velocity activities. Small portable non-differential GPS units are now available allowing GPS technology to become more commonly used in sport. Sports have adopted GPS for athlete tracking, performance monitoring and distance measurement (Randers et al., 2010).

Sensor-based technology, such as GPS, has the potential to provide positional information without the extensive time required for processing and can be used at any location without calibration. However, the sample rate used on commercial GPS receivers is usually 1 Hz. Although the satellites provide signal continuously these are sampled at fixed intervals which results in some variability in the distance and speed estimate. This sample rate along with positional accuracy causes resolution problems when considering short or fast events. For example, 1 Hz over a 100 m running race that takes 10 s to complete only provides ten samples each with > 10 m (10%) accuracy. The same system used for a marathon that takes 2 hours to complete would provide more than 7200 data points over 42000 m with > 100 m (0.2%) accuracy. Thus, the accuracy and sample rate are a much smaller percentage for marathon competition compared to sprint.

Many studies using GPS units have not assessed the accuracy of the receiver for the particular application. Rather they use the manufacturer‟s accuracy specifications which may not be applicable to the environment in which their testing is being conducted. Witte & Wilson (2004) assessed the accuracy of a non-differential GPS unit with regards to velocity and found that 45% of values were within 0.2 ms-1 and the accuracy of speed determination was preserved even when positional information was degraded due to poor satellite numbers or geometry. Over a 100 s canoe slalom event, this equates to 20 m or 5-8% of the total course (250 m – 500 m long course).

26 Errors in GPS information increase on circular paths, especially those with small radii of curvature, due to a tendency to underestimate distance and therefore speed. These errors suggest that GPS would not be effective for positional or velocity information for canoe slalom. In addition Australia does not have an active WAAS system and therefore must rely on non-differential GPS. The positional errors associated with non-differential GPS units can be as high as 15 m (Witte & Wilson, 2005). For canoe slalom, an accuracy error of 15 m would prevent discrimination between paddlers, gates and the path taken down the course. Hence, video based analysis is still the most accurate and applicable method for extracting information on the path taken by canoe slalom paddlers.

2.1.5 Force Curves Research using paddle force systems has been important for advancing the sport of flatwater sprint kayaking (Aitken & Neal, 1992; Ho et al., 2000; Sperlich, 1995a, 1995b; Sperlich & Klauck, 1992; Stothart et al., 1987). Paddle force data acquisition systems have given coaches and sports scientists working with flatwater kayaking an objective measure for prescribing and assessing technique modifications aimed at enhancing performance (Aitken & Neal, 1992). Although the paddle force systems used for sprint kayaking can be utilised in canoe slalom without modification, the data are not comparable because of the nature of the each sport. Flatwater sprint kayak involves racing in a straight line, using only forward strokes and is performed in a relatively controlled, repeatable environment. In contrast, canoe slalom involves manoeuvring through gates, using a number of different types of strokes and is raced on whitewater with a course that changes with each competition.

Paddle force systems measure the force acting on the paddle through the deflection of the shaft which provides a unidirectional measure of the net force applied to the blade face (Aitken & Neal, 1992). In flatwater sprint the paddles do not have a flat surface and the applied force is usually assumed perpendicular to the blade face. In canoe slalom the blades are flat therefore the unidirectional measurement of shaft deflection would provide a better measure of the quantity of force generated by the paddler (Marhold & Herrmann, 1981). However, the generation of force by the paddler could produce movement of the boat in any direction. The complexity of accurately measuring boat position and orientation, paddle position and orientation and water flow conditions limits the viability of measuring every aspect of performance. Even so, many insights can be gained through the unidirectional

27 forces if these are combined with descriptive information from video footage which gives context to the stroke and course (Sperlich & Klauck, 1992).

2.2 Development of Analysis Systems for Canoe Slalom This section details the Competition Analysis Program (CAP) for canoe slalom, the associated operational definitions and the repeatability analysis for this system as well as development of the paddle-force analysis program including the calibration and repeatability of the force system. These analysis, measurement and data collection packages were necessary, as none were commercially available when this research was initiated.

2.2.1 Competition Analysis Program (CAP) for Canoe Slalom 2.2.1.1 Development of CAP Commercial time-motion analysis systems were not found to offer customised user-specific reporting or data analysis methods. Development of specific software allows for customised data collection and analysis, and reporting methods that make the information collected more relevant to the sport and easier to understand whilst maintaining a standardised format in which the data are collected (Hughes et al., 2007). In canoe slalom competition, no such specific software exists that provides specific information for coaches to compare paddlers or runs for the same paddler. A customised canoe slalom program would enhance the feedback provided to coaches and athletes which has been shown to aid performance (Murray et al., 1997).

A customised software program was developed as part of the research reported in this thesis. The program made it possible to undertake competition analysis of an entire canoe slalom run in terms of qualitative data on strokes taken and temporal data such as split times between gates. It was necessary for the system to be capable of playing multiple video files, synchronising them, coding different stroke types and event times and outputting the specific information relative to a common timeline. The competition specific information obtainable from time-motion analysis that would allow coaches to compare paddlers or runs for the same paddler included gate split times, turn times, penalties, avoidance around gates, the types of strokes and strategies used.

28 2.2.1.1.1 Platform for CAP Visual Basic for Applications (VBA) contained within Microsoft Excel® software (Microsoft, Washington, United States) was used as the platform on which the customised canoe slalom competition analysis program was developed. This platform was used because of its familiarity to coaches and athletes, widespread accessibility, ability to utilise the graphing and data analysis functions of Microsoft Excel®, ease of creating operator interfaces, and large number of built-in functions. Video control was achieved using the Windows Media Player® (Microsoft, Washington, United States) application programming interface (API). Additional operator controls were developed to record temporal information from the video and qualitative descriptors.

2.2.1.1.2 Operational Definition Set The process to develop the operational definition set for canoe slalom strokes and run involved the international canoe slalom coaches and three performance analysts employed by the Australian Institute of Sport (AIS). The development of the definition set involved creation of a list of potential variables, describing each variable then reviewing and refining these descriptions. During this process, C2M presented many problems, as one paddler was inevitably occluding the view of the other paddler which made it difficult to undertake time- motion analysis. In addition, C2M would require two separate analyses (1 for each paddler) for each run and then the net effect of both to be combined. The athlete pool for C2M is small and therefore C2M were excluded from further analysis. Thus, the resultant definition set applies to K1M, K1W and C1M and includes the following measures: gate split times (time taken between gates), touched and missed gates, turn times, major and minor avoidance (contortion of body around a gate), rolls, paddle in and out of water times, and stroke categorisation. Strokes were recorded as being on the left or right hand side of the boat from the paddler‟s perspective. For the C1M category, the preferred (dominant) side was noted so that left and right could be changed to onside (preferred) and offside (non-preferred) strokes post analysis. These stroke characteristics and race definitions are detailed in Table 2.1. From these data the between gates split times during different sections of the race could be compared along with the number of errors made, the magnitude of these errors and impact on other aspects of the run. Strokes were defined as the time period between the paddle in and the paddle out of water point and these were broken down into two categories, pure (single phase) strokes (Table 2.2) and multi (dual phase) strokes (Table 2.3). A description of the

29 main characteristics of each stroke, the effect the stroke had on the boat in long and short descriptions as well as a diagram of the stroke were provided.

Pure strokes were identified as strokes that have one predominant phase and included the following strokes: forward, C, draw, sweep, reverse sweep, reverse, tap, brace, punt, side draw and steering (Table 2.2). However, many strokes taken during a slalom competition are a mixture of different types of strokes. Therefore, strokes that fell into more than one category but were not multi-strokes were defined by the predominant action of that stroke.

Multi-strokes were identified as strokes where the paddler did not remove their blade from the water before performing a second type of stroke. In some situations the paddler sliced the blade through the water to get to the starting position of the next stroke, whereas in others the starting position of the second stroke was the end of the first. Identifiable components of multi-strokes were labelled (e.g., steering-forward) but the combination was treated as one stroke. For example, a stroke that consists of a steering stroke followed by a forward stroke, where the paddle did not leave the water, was defined as a „multi-stroke: steering-forward‟ and the stroke count for this was one. An exception to this was made for the C1 paddle stroke with each time the pressure of the blade on the water was released was counted as the end of a stroke. If the blade was moved to a new location without pressure on the blade and followed by re-applying of pressure, even if it did not leave the water, it was counted as a new stroke. This was a result of C1‟s natural awkwardness to remove the paddle from the water during the recovery phase on their offside strokes. The strokes that were included as multi-strokes were: draw-forward, reverse sweep-forward, forward-reverse sweep, draw-draw, reverse sweep- draw, draw-sweep, forward-sweep and major / minor strokes (Table 2.3).

The definition for the starting time for a turn used in this study was the corresponding time when the boat was 90 degrees to the water flow and then continued to turn beyond this point. The end time of a turn was noted when the boat was at 90 degrees to the water flow again. For example, turning upstream for an upstream gate and then return to pointing downstream was two turning points, one before and one after the gate.

Table 2.1: Stroke Characteristics and Race Definitions. Stroke Characteristic Description Quick Reference Gate Split Times - Use the lower torso as it passes the poles as the reference for the gate split. -To get consistent timing: - Front view: use a wave which can be identified in every run and then use when the nose or certain 1. Use lower torso against poles or section of the boat breaks through the wave. This should achieve greater consistency than trying to 2. Boat relative to waves. pick the gate. Touched Gates - Any part of paddle, paddler or boat makes contact with either pole - Gate pole was hit - 2 second penalty - 2 s penalty Missed Gates - The head of the paddler didn‟t go between the poles of the gate - Gate missed by paddler - 50 second penalty - 50 s penalty Turn Times - Turn Times can be defined as the time when the boat was 90 degrees to the water flow and then - Boat turning upstream or downstream (major continued to turn beyond this point. change in direction) - eg. Turning upstream for an upstream gate and then return to pointing downstream = 2 turning - Point where the boat is 90 degrees to the main points, one before and one after the gate. water flow. Major Avoidance - This occurs when the paddler contorts their body to get around / through a gate and this contortion - Causes a negative impact on propulsion, balance results in their normal paddling technique, balance and / or propulsion being negatively affected. and / or stroke. - Propulsion of the boat becomes reduced due to major avoidance. - For example: hesitation before taking the next stroke, cut a stroke short, using an ineffective stroke. Minor Avoidance - This occurs when the paddler contorts their body to get around / through a gate but, this contortion - Does not impact on propulsion, balance and / or does not result in their normal paddling technique being affected. stroke. - Propulsion of the boat remains unaffected during minor avoidance. Rolls - When the boat goes upside down and the paddler has to perform a roll to right themselves again. - Boat upside-down - To indicate the start and end of the period during which the boat was inverted. Paddle In Time - Closest point to where the paddle begins to grip the water. - In side of gripping the water. - If the majority of the blade can be seen then it isn't in the water yet. - First point the paddle causes an effect on the boat. - If between frames then pick the frame on the side of definitely gripping the water. - First point the paddle causes an effect on the boat. Paddle Out Time - Closest point to where the paddle begins to be no longer effective in the water. - In side of losing grip with the water. - If the majority of the blade can be seen then it has already left the water. - Last point the paddle causes an effect on the boat. - If between frames, then pick the frame on the side of still gripping the water. - Last point the paddle causes an effect on the boat. Stroke - A stroke is the period between the „Paddle In time‟ and the „Paddle Out Time‟. - Paddle In to Paddle Out - Defined as meaningful use of the paddle. - Meaningful use of paddle Onside - C1 paddling on their preferred side. - Preferred side. - Top hand crosses over deck. - Top hand crossed over deck. - If left hand is the bottom hand on the paddle then the paddlers onside is their left. OffSide - C1 paddling on their non preferred side. - Non preferred side. - Bottom hand cross over deck. - Bottom hand crossed over deck. - If left hand is the bottom hand on the paddle then the paddlers offside is their right.

Table 2.2: Stroke Definitions - Pure Strokes. Stroke Name Description Effect on Boat Stroke Effect Stroke Diagram Forward - Propulsive stroke. - Propulsion of the boat forward. Forward >90% - Normal paddle stroke (paddle pulls straight through water). - Stern following bow. - Top hand moves straight forward. - No significant change in direction. - Similar to old flat-water technique before blade changes. - Boat tracks straight. C - Propulsive draw (smaller opening angle than draw). - Turns boat while propelling it forward. Turn 50% - One continuous action. - Smooth movement in both turning and Forward 50% - Blade moves in a C shaped path relative to the boat. propulsion. or combination - Combines both actions together. Draw - Blade facing inwards, parallel to boat (more open). - Significant change in direction. Turn 100% - Top hand kept high. - Causes the boat to rotate (just turns the boat). Some forward run - Blade drawn in towards the bow of the boat. - No real propulsion during stroke. from previous strokes. Sweep - Blade moves in an arc around the paddler starting at the - Significant change in direction. Turn >90% bow. - Not much propulsion during stroke. Forward <10% - Top hand moves low across body. - Bow moves away from the blade. - Blade facing outwards. Reverse Sweep - Blade moves in an arc around the paddler starting at the - Significant change in direction. Turn >90% stern. - Does not slow the boat. Reverse <10% - Top hand moves low across body. - Bow moves towards the blade. - Blade facing outwards. Reverse - Braking stroke. - Braking Braking >90% - Can propel boat backwards. - Propulsion of the boat backwards. Turning <10% - Blade moves in a forward direction starting at stern. - No significant change in direction. - Often occurs behind the body. Tap - Pure in and out of the water without much pressure on the - No observable effect on Boat Ineffective stroke blade. Timing stroke - Very short stroke.

Brace - Face of blade facing the sky or river bed, shaft flat to water. - Usually, no observable effect on Boat. Balance stroke - Supporting stroke. - Can be used to angle the boat - Paddlers use this stroke for stability. - Top hand mid to low, paddle pressing on water. Punt - Tip of blade in contact with the bank or other solid obstacle. - Usually used as a turning but, also propulsive. Forward varied - Pushing action along length of blade. Turn varied

Table 2.2: Stroke Definitions - Pure Strokes (Continued). Stroke Name Description Effect on Boat Stroke Effect Stroke Diagram Side Draw - Blade facing inwards, parallel to boat (more open) in line - No real propulsion. Sideways 100% with body of paddler. - No real change in direction. - Top hand kept high. - Boat moves sideways. - Blade drawn towards the middle of the boat. Steering - Position is the same as the start of a draw or the end of a - Steering / guiding boat. Guiding Stroke sweep (blade parallel to boat). - No real propulsion. Turn 100% - Blade not moved, used like a rudder. - Can be used to turn the boat or used to keep - No propulsion. the boat tracking straight.

Table 2.3: Stroke Definitions - Multi Strokes. Stroke Name Description Effect on Boat Stroke Effect Stroke Diagram Draw - Forward - Combination of a draw stroke followed by a forward stroke. - Turns the boat and then propels it forward. 1. Turn 100% - 1. Draw 2. Forward 100% - 2. Forward

Reverse Sweep - - Combination of a reverse sweep stroke followed by a - Turns the boat and then propels it forward. 1. Turn >90% Forward forward stroke. 2. Forward 100% - 1. Reverse Sweep - 2. Forward

Forward - - Combination of a forward stroke followed by a reverse - Propels the boat and then turns it. 1. Forward 100% Reverse Sweep sweep stroke. 2. Turn >90% - 1. Forward - 2. Reverse Sweep

Draw / Draw - Combination of two draw strokes with a cutting slicing action - Turns the boat a large number of degrees. 1. Turn 100% between to return the blade to the starting position. 2. Turn 100% - 1. Draw - 2. Draw

Reverse Sweep - - Combination of a reverse sweep stroke followed by a draw - Turns the boat a large number of degrees. 1. Turn 100% Draw stroke. 2. Turn <90% - 1. Reverse Sweep - 2. Draw

Draw - Sweep - Combination of a draw stroke followed by a sweep stroke. - Turns the boat one direction and then turns it 1. Turn 100% - 1. Draw back again. 2. Turn >90% - 2. Sweep - Used mainly for gate where the paddler travels across the flow.

Table 2.3: Stroke Definitions - Multi Strokes (Continued). Stroke Name Description Effect on Boat Stroke Effect Stroke Diagram Forward - Sweep - Combination of a forward stroke followed by a sweep stroke - Propels the boat and then turns it. 1. Forward 50% When the blade comes in line with body, the blade is pulled 2. Turn 50% towards the boat (squeeze). - 1. Forward - 2. Sweep Major / Minor - A combination of two strokes which cannot be described as - Causes a variety of movements depending on Varied one of the existing multi strokes. For example: the phases being combined and their relative - 1. Brace contribution to the stroke. - 2. Sweep

Key to Stroke Diagrams Arrow indicates the path Reverse-Sw eep Component Dashed arrow s indicate path of paddle of the boat during the stroke Forw ard Component in the w ater during the stroke. This example is a multistroke made up from a forw ard stroke and a reverse-sw eep. Arrow in the bottom left points in the direction of the nose of boat. Key: Boat as viewed from above

2.2.1.1.3 Navigating CAP The main menu graphical user interface (GUI) directs the operator to the various controls and functions available in CAP (Figure 2.1). To create and change analysis files, New Analysis starts a new analysis, Load Analysis opens an existing analysis and Crash Recovery loads the backup file. To personalise the analysis shortcuts, Set Hot Keys allows the operator to allocate their own preferred coding buttons. Finally, to analyse and report the data, Individual Report One performs calculations on a single file, Individual Report All performs calculations all files in a folder and Summary Report exports the data into a summary file for statistical analysis.

Figure 2.1: Main Menu GUI (Graphical User Interface) allowed the operator to navigate the functionality of the competition analysis program (CAP). New Analysis is for starting analysis of a new competition run. Load Analysis opens an existing analysis for review or editing. Crash Recovery opens the current analysis if the program was inadvertently closed. Set Hot Keys allows the user to setup their personalised keyboard shortcuts for the Analysis GUI. Individual Report One and All process the analysed data and produce basic statistics on each file. Summary Report sends the data to a single summary file for statistical comparison.

35 2.2.1.1.4 New Analysis Since the number of cameras used at a slalom event may vary from competition to competition, the user has to define the number of video files / cameras used in each analysis. This is done by clicking the New Analysis button on the main menu. A popup selection box is displayed that requires the operator to select the number of video files to combine into the analysis file (Figure 2.2). The program prompts the operator to load each Audio Video Interleave (AVI) file through a dialog box (Figure 2.3). The program loads each video file and determined the duration and frame rate.

Figure 2.2: Dialog Box for the number of video files used in the analysis. Between one and ten video files could be selected.

Figure 2.3: Dialog box to select videos for the analysis in the order that the cameras were situated along the course. If three video files were to be used, then the dialog appeared three times.

36 To combine multiple video files into a single analysis each video files must be synchronised with each of the others. A time point common to all video files is identified and marked using the Synchronisation GUI (Figure 2.4), thus allowing the analysis timeline to extend across all video files (Figure 2.5). The synchronisation process collects and stores each of these common time points. The data stored during the synchronisation process includes the video name, frame rate, sync point 1, sync point 2 offset and file name including path. Sync point 1 in the first video is always zero (the time of the first frame) and in the second video sync point 1 was the frame in which the video matches sync point two in the first video. In the example provided (Table 2.4) the synchronisation point occurred 45.28 seconds from the beginning of the first video file and 1.56 seconds from the beginning of the second video file, therefore the beginning of the second video file would occur 43.72 seconds from the beginning of the first video file.

Figure 2.4: The Synchronisation GUI allows the operator to view each video file and identify synchronisation points. Once identified and entered using the buttons (Sinc1 and Sinc2) the operator can review them by clicking on the video file in the list. The video always jumps to the fist synchronisation point thus to review the second point the user can click a button (Goto Sinc 2).

37 Sync Point 2 Sync Point 2 Video 1 Sync Point 2 Video 2 3 Video 3 Video 1 3 Video 2 3 Video 3

Sync Point 1 Sync Point 1 Video 1 Sync Point 1 Video 2 3 Video 3 3 3 Figure 2.5: Example of how multiple video files are combined into a single analysis timeline using synchronisation points. The first synchronisation point (sync point) of the first video is always 0 s and the last sync point of the final video file represents the length of that video file in seconds. The other sync points represent a common time point between two videos which are identifiable in both views.

Table 2.4: Example of the data collected and stored when synchronising two videos. The data identifies the video name, its storage location, the number of frames per second it contains and the synchronisation data. Sync1 and Sync2 represent the time in seconds within each video file where the synchronisation point occurs. The offset represents the time in seconds that the beginning of each video file is from the start of the first video. Video Name Hz Sync1 Sync2 OffSet File Name VideoFile1 25 0 45.28 0 C:\VideoFile1.avi VideoFile2 25 1.56 64.6 43.72 C:\VideoFile2.avi

Once the video views are synchronised the operator can progress to the Analysis Graphical User Interface (GUI) (Figure 2.6). As the operator clicks the buttons across the bottom or uses the hot keys to code, the data are added to one of the five lists on the right hand side of the GUI (Figure 2.6). Data added to the lists can be edited, deleted or updated. By clicking on a value in the list, the program steps automatically to the corresponding time reference in the video.

Figure 2.6: The Analysis GUI shows the final layout of the GUI. During the analysis of a competition run, the operator codes events such as Gate Splits, Sectors of Upstream Turns, Touched Gates, Missed Gates, Eskimo Rolls. The operator can also tag major or minor avoidance strategies used to negotiate a gate. Finally, the entry and exit of each stroke is marked and coded individually as right or left with the type of stroke identified.

39 2.2.1.1.5 Load Analysis After a run is completed and the file saved, the operator can review the file at a later date through Load Analysis. If the video files remain in the same location the videos along with all data are loaded. Any data can then be reviewed and changed where necessary.

2.2.1.1.6 Crash Recovery As data are inputted, a backup file is simultaneously stored in comma separated values (csv) format (Figure 2.7). The backup file (TEMP Canoe Slalom Analysis.csv) is updated when the operator changes values and enters data. The backup file enables the operator to recover data if the program inadvertently closes or crashes. The files saved by the backup program are in the same format as the data output files.

Figure 2.7: Example of data stored in the backup file (TEMP Canoe Slalom Analysis.csv). The first line of the file contains the header information for the synchronisation data. The following lines contain the synchronisation information for each video file in the analysis. The header information for the analysis data and any data from the current analysis which had been completed follows.

2.2.1.1.7 Set Hot Keys Keyboard short cuts are built in for all video controls and coding operations. Initially short cuts were programmed to specific keys but subsequently the operator is able to define the keys that control the program and inputted data (Figure 2.8). This information is stored in a comma separate values (csv) format in the Slalom Keys.key file that resides in the same folder as the program file. The operator can save various setup versions as different files which can be loaded later to allow multiple users to code using different customised keyboard shortcuts. Although there is an option to customise the coding controls, default keys are allocated specifically to maximise coding efficiency. This is set up so the operator can control the video with the right hand on the number pad and code stroke and timing information with the left hand using keys located on the left side of the keyboard.

Figure 2.8: Hot Key Assignment GUI allows the operator to assign which keys control the program in the Analysis GUI. All coding buttons on the Analysis GUI have a keyboard shortcut assigned to them. This allowed rapid navigation of the software and increased analysis speed.

2.2.1.1.8 Individual Report One and All Once an entire run analysis is completed, the operator can generate summary information and reports from each raw file. After saving the file, the operator chooses to generate either one report for the file selected in the popup dialog box by clicking Individual Report One or to generate a report for each file in the same folder as the file selected in the popup dialog box by clicking Individual Report All. However, before these files are converted to a summary report the operator must input the gate definitions (upstream or downstream) for the course (Figure 2.9). The gate sequence file can be saved as a csv file for later use. This GUI appears only once at the beginning for the Individual Report All function, so all the files have to be from the same course. Clicking OK runs the program to generate the report.

Figure 2.9: Course Gate Sequence Input GUI (Left – zero gate selected, Right – run with gate definitions entered).

The report program generates a Microsoft Excel® file (xls) with the same name as the raw csv file. It does not overwrite the original file. The resultant Microsoft Excel® file contains four spreadsheets: Raw which contains the same data as the raw csv file; Actual which contains the data adjusted for the video start time, arranged in time and gate sequence with direction, penalties and avoidances associated with the each gate. In addition the stroke time and stroke rate are also reported (Figure 2.11); Graphs which contains data and the initial development of the graphs used in the final reports (Figure 2.13). This spreadsheet contains three graphs, two of which would be impractical to create through the operator interface of Microsoft Excel® (Figure 2.13). Totals which provides counts, mean times and standard deviation of times for each stroke type for left, right and total strokes (left and right combined). In addition this spreadsheet contains the number of strokes taken between each gate which is again divided into left, right and total.

Figure 2.10: An example of a report file, showing the Raw spreadsheet. This is the data contained in the original csv file which was created during the analysis process. The same structure remains with the synchronisation detail contained in the first three lines of this example. Times are all relative to the start of the first video and data in the order in which it was coded.

Figure 2.11: An example of a report file, showing the Actual spreadsheet. The conversion to this format requires all data to be sorted sequentially and times made relative to the start allocated by the operator during analysis. The data entered in the course gate sequence GUI has been combined with the raw gate data and the penalties and avoidance tied to the gate at which they occurred. Stroke time and stroke rate are also calculated in this process.

Figure 2.12: An example of a report file, showing the Total spreadsheet. The stroke information is processed to determine the total number of each type of stroke, and the average and standard deviation of time spent in the water per stroke type. The number of strokes between each gate is also calculated.

Figure 2.13: An example of a report file, showing the Graph spreadsheet. This spreadsheet is the base setup for creating the custom graphs for the report. Note that these graphs are presented in Excel® default format and not in standard scientific format as these are to allow the operator to assess the output prior to reporting and are not used in the final report.

46 The „Race Timeline‟ was developed to display the time the paddler spent on each stroke during a race and when these occurred relative to the gates and turns. The time at which the stroke started relative to the run start is recorded on the x-axis and the length of the stroke in seconds is recorded on the y-axis (Figure 2.14). This plot was replaced by a second race timeline graph in which each stroke is a separate line on either one (right, blue) or two (left, red) and the length of the line represents the time the stroke was in the water. Each of these strokes could be identified by naming each separate series (Figure 2.15). This graph would be impractical to create manually due to the number of series required - one for each stroke. The final graph generated is a simple bar graph with the total stroke count for each type of stroke (Figure 2.16).

Once the individual reports are generated for each raw file a report file is initiated. The report file starts as a Microsoft Excel® template file which the program opens and into which data are inserted (Figure 2.17). A number of figures are produced (Figure 2.18). This file can be saved and printed by the operator. At the end of these processes a report can be generated for the coach and athlete describing their performance.

2.2.1.1.9 Summary Report At the end of these processes a summary of all information can be added to a summary report file. The file was setup for statistical analysis in comma-separated format (csv) with one row per trial.

Figure 2.14: The first version of the race timeline. This plot shows the stroke in water time (s) on the y-axis and the race time on the x-axis. Each stroke right (blue, circles) and left (red, squares) is positioned along the race time (x-axis) based on the start (in) time of the stroke. Vertical green and red lines represent downstream and upstream gates respectively. The divisions around upstream gates (turn) were defined by vertical yellow lines.

Figure 2.15: The second version of the race timeline. The race time was presented on the x-axis. Each stroke position and duration is represented by a separate line positioned along the race time (x-axis), but the right and left strokes on separate lines on the y-axis with right = 1 set as blue and left = 2 set as red. Vertical green and red lines represent downstream and upstream gates respectively. The divisions around upstream gates (turn) were defined by vertical yellow lines.

Figure 2.16: Bar graph showing the number of each type of stroke which the paddler took during the run. Multi-strokes are strokes that could not be defined as one of the strokes identified in the definition set.

Figure 2.17: A report template file with portrait figures for quick data collection to facilitate timely coach and athlete feedback.

Figure 2.18: The report template file with landscape figures for quick data collection to facilitate timely coach and athlete feedback.

2.2.1.2 Reliability Assessment for CAP Repeatability studies of time-motion analysis in sports with similar manoeuvring characteristics to canoe slalom have reported that correlation coefficients greater than 0.8 for intra-observer comparisons indicated acceptable reliability levels for the analysis (McKenna et al., 1988; Schokman et al., 2002). Schokman et al. (2002) assessed intra-observer and inter-observer reliability using voice recognition analysis to classify accurately gait transitions and quantify gait durations from video. Both reliability measures were consistently high ranging from r = 0.87 to r = 0.99. McKenna et al. (1988) reported that an acceptable level of reliability for a single observer recording the time a player spent performing gait and game specific activities was r ≥ 0.92 for total time and r ≥ 0.83 for a count of the activities. Note that this approach was developed before the availability of digital records of performance.

The competition analysis program (CAP) described in the previous section depends on the ability of the operator to accurately and reliably relate stroke definitions to what they see the paddler doing on the video footage. This section investigates the performance of CAP by analysing the intra- and inter-observer reliability of three operators using the operational definition set and the competition analysis program described in the previous section. This section is taken directly from Hunter et al. (2007) published in Sports Biomechanics.

2.2.1.2.1 Methodology One competition run from four national level canoe slalom paddlers in an Australian national selection race were filmed using two Sony® (Minato, Tokyo, Japan) digital video cameras (DSR-PDX10P PAL). The four runs included one men‟s kayak (K1M), one women‟s kayak (K1W) and two canoe (C1M) runs (one right handed and one left handed). One camera was positioned to capture the top half of the course and the second camera was positioned to capture the bottom half of the course. Each camera was situated in the best position to view its respective section of the course, whilst overlapping with the other section. These cameras recorded onto Sony® (Minato, Tokyo, Japan) digital video cassettes (DVM60) and were time stamped to an accuracy of 1/50th of a second. To aid the analysis process the camera operator continually framed the paddler using the zoom and pan so they filled the frame. However, some of the filming for this study was not as tightly zoomed as recommended for the optimal analysis. The shutter speed of the cameras was dependent on light condition but both cameras were set to faster than 1/1000th of a second to prevent blurring of the image. This footage

was subsequently captured onto computer using video editing software capable of producing a sample rate of 25 Hz without blurring due to incorrect de-interlacing of the video file.

In order to test the reliability of the system and observer coding, each of the four different runs were analysed three times by three observers using the CAP described in the previous section. There was no predetermined time limit for analysis of the runs. The repeat analyses were randomised for each observer to reduce any direct learning effect on a particular run. Nevertheless, initial analyses took around 3 hours whereas later analyses took around 1 hour to complete suggesting that operators were learning the system with time.

The reliability of detecting penalties through video was considered important because human error in judging at competitions results in some penalties being missed and others being awarded when no error was made. Judging errors can occur due to environmental condition such as wind causing gate movement or paddler and wave interaction causing water to spray up and move the gate. These interactions can cause judges to award penalties when one should not have been and not award a penalty when one should have been awarded. Therefore, the video was used to assess penalties independently and many times over so that comparison of each paddler‟s true performance could be made.

From each raw file (Table 2.5) generated from the analysis of each run, gate split times, turn times and stroke information were extracted into three separate tables for each variable. The tables were arranged so that intra-observer, trial and inter-observer error could be calculated.

Table 2.5: An example of the raw data collected using CAP when analysing a single canoe slalom competition run. Each data point adds a new row to the specific variables column of data. Gate Turn Penalty Right Gate Splits Times Points Penalty Time Left Strokes IN OUT Strokes IN OUT 0 3.2 21.84 Minor Avoidance 29.12 Forward 3.08 3.84 Forward 4.12 4.68 1 6.84 24.44 50 Miss 44.72 R. Sweep 4.8 5.56 Forward 5.8 6.16 2 9.76 37.16 2 Touch 68.52 Brace 6.36 7.04 Reverse 7.24 7.88 3 14 45.16 Sweep 8.08 8.72 Forward 8.88 9.48 4 16.48 61.56 Forward 9.76 10.16 Forward 10.44 10.96

Each time datum point within- and between-runs could not be compared directly without normalising the data to a common mean. Therefore, the mean time for each observer was calculated for equivalent data points across their three repetitions. The mean time for all

observers across the nine repetitions (three for each observer) was calculated. The time differences to the means were then used to compare data for synchronisation time, gate split times, turn times, stroke in times and stroke out times. However, before stroke in and stroke out of water times could be analysed, both left and right were combined into one list based on time and then strokes were matched for time in and out for each trial.

To determine the reliability of the stroke identification for both intra-observer and inter- observer analysis the number of matching stroke identification responses were expressed as a percentage of total number of stroke identification responses. If the strokes matched exactly they were classified as correct strokes. However, it was also possible to have a half correct identification by either identifying only one part of a multi-stroke or combining a pure stroke into a combination stroke. The percentage of half-correct strokes was then divided by two to weight them lower than correct strokes. Correct strokes and half-correct strokes were then added to give a total percentage of correct strokes.

A similar method to that used for stroke identification was employed to assess the reliability of penalties and identification of avoidance, but without half correct identifications. Using two categories „avoidance‟ and „no avoidance‟ the rate of correct identification was assessed based on the majority of identifications for each gate. The percentage of major and minor avoidance was compared for avoidance and no avoidance.

For all time related variables the mean, minimum, maximum, range, standard deviation, error of measurement and limits of agreement were calculated for each variable and indicated the variability / reliability of the data. Limits of agreement provide an indication of the minimum difference between analysis trials that were not considered errors of measurement process. Normalisation of the data resulted in the mean always being calculated as zero.

2.2.1.2.2 Results Analysis of all variables revealed that when trials were analysed by a single observer (intra- observer analysis) there was less variation than when multiple observers analysed the trials (inter-observer analysis). This trend was observable in the range, standard deviation, total error of the mean and the limits of agreement for each variable. Results of a two tailed paired T-test for all variables for intra and inter- observer variability demonstrated that intra-observer variability was significantly (p < 0.05) less than inter-observer analysis variability. Intra-

observer difference was 69% of the variability of the inter-observer analysis. The data for the video synchronisation point showed the least variation. The limits of agreement were 0.03 and 0.04 seconds for the intra-observer and inter-observer analysis respectively (Table 2.6 & Table 2.7). Gate split times and turn times showed similarities in variation. In the intra- observer analysis the limits of agreement for both gate split times and turn times was 0.21 seconds. For inter-observer analysis the limits of agreement for both gate split times and turn times increased to 0.39 and 0.37 seconds respectively (Table 2.6 & Table 2.7). Stroke in and out of water times revealed similar variation. The limits of agreement increased from 0.07 and 0.08 seconds for stroke in and stroke out in the intra-observer analysis to 0.13 and 0.11 seconds respectively for inter-observer analysis (Table 2.6 & Table 2.7).

Table 2.6: Intra-Observer Variation for temporal data when completing repeated analyses using CAP (all values presented in seconds). Synchronisation Gate Split Turn Times Stroke In Stroke Out Mean 0.00 0.00 0.00 0.00 0.00 Minimum -0.05 -0.45 -0.63 -0.40 -0.43 Maximum 0.03 0.53 0.59 0.35 0.59 Range 0.08 0.99 1.21 0.75 1.01 Standard Deviation 0.01 0.11 0.11 0.04 0.04 Total Error of the Measurement 0.01 0.08 0.08 0.03 0.03 Limits of Agreement 0.03 0.21 0.21 0.07 0.08

Table 2.7: Inter-Observer Variation for temporal data when completing repeated analyses using CAP (all values presented in seconds). Synchronisation Gate Split Turn Times Stroke In Stroke Out Mean 0.00 0.00 0.00 0.00 0.00 Minimum -0.06 -0.92 -0.76 -0.53 -0.76 Maximum 0.03 0.80 0.80 0.60 0.84 Range 0.08 1.72 1.56 1.13 1.60 Standard Deviation 0.02 0.20 0.19 0.06 0.05 Total Error of the Measurement 0.01 0.14 0.13 0.05 0.04 Limits of Agreement 0.04 0.39 0.37 0.13 0.11

Intra-observer analysis of stroke identification for all trials revealed that C, draw-draw, forward-reverse sweep, forward-sweep, reverse sweep, reverse sweep-draw, reverse sweep- forward, reverse, taps and steering strokes recorded ten or less occurrences each out of the 355 strokes analysed (Table 2.8). This compared to inter-observer analysis where 364 strokes were identified, but three fewer categories were identified, that is forward-reverse sweep, forward-sweep and reverse strokes were not included. The percentage of correct identifications for strokes identified fewer than ten times was generally low, but ranged from

22% for reverse sweep to 100% for reverse sweep-draw and reverse strokes (Table 2.8 and Table 2.9).

The correct identification of strokes identified greater than ten times for intra-observer analysis was 72% for multi-strokes, 82% for braces, 80% for draws, 81% for sweeps and 86% draw-forwards (Table 2.8). For inter-observer analysis the correct identification of these strokes were 51% for multi-strokes, 52% for braces, 65% for draws, 73% for sweeps and 60% draw-forwards (Table 2.9). Including half-correct strokes improved the identification of all strokes.

Forward strokes made up greater than 60% of the total number of strokes and these were identified correctly between 96% of the time for intra-observer analysis and 85% of the time for inter-observer analysis including half correct strokes. Intra-observer analysis recorded 355 strokes which were identified correctly 89% of the time, this increased to 91% with the inclusion of half correct strokes. Of the 364 strokes recorded for inter-observer analysis, 78% of the time the strokes were identified correctly, which increased to 81% with the inclusion of half correct strokes (Table 2.9).

Table 2.8: Average Intra-Observer Stroke Identification percentage when completing repeated analyses using CAP. Number of % of Correct % of Half Correct % of Correct Stroke Type Strokes Strokes Strokes Strokes + Half / 2 Brace 26 82% 4% 84% C 3 70% 0% 70% Draw 17 80% 11% 85% Draw-Draw 2 72% 11% 78% Draw-Forward 34 86% 5% 89% Forward 217 95% 2% 96% Forward-Reverse Sweep 1 33% 33% 50% Forward-Sweep 4 58% 8% 63% Multi 16 72% 18% 81% Reverse Sweep 2 61% 6% 64% Reverse Sweep-Draw 1 100% 0% 100% Reverse Sweep-Forward 7 95% 3% 97% Reverse 1 100% 0% 100% Steering 5 61% 0% 61% Sweep 27 81% 4% 83% Tap 6 80% 0% 80% All Strokes 355 89% 4% 91%

Table 2.9: Inter-Observer Stroke Identification percentage when completing repeated analyses using CAP. Number of % of Correct % of Half Correct % of Correct Stroke Type Strokes Strokes Strokes Strokes + Half / 2 Brace 12 52% 9% 56% C 1 33% 0% 33% Draw 18 65% 15% 72% Draw-Draw 1 44% 44% 67% Draw-Forward 43 62% 14% 69% Forward 240 85% 4% 88% Multi 11 51% 28% 65% Reverse Sweep 1 22% 0% 22% Reverse Sweep-Draw 1 33% 33% 50% Reverse Sweep-Forward 8 78% 21% 88% Steering 1 67% 0% 67% Sweep 22 73% 5% 76% Tap 5 69% 0% 69% All Strokes 364 78% 7% 81%

Only one penalty (touch) was detected in the analysis and this was identified correctly in all trials by all observers. Detection of when no avoidance occurred was greater than 90% correct. Minor and major avoidance were detected correctly greater than 80% of the time for intra-observer analysis and 70% of the time for inter-observer analysis. When no avoidance was detected, 83% of errors resulted from minor avoidance being detected and the other 17% of the time because major avoidance was detected (Table 2.10).

Table 2.10: Intra- and Inter-Observer Avoidance Identification percentage when completing repeated analyses using CAP. Penalty Intra-Observer (average) Inter-Observer Minor Avoidance 91% 83% Major Avoidance 81% 69% No Avoidance 94% 91% No Avoidance - Minor Avoidance 83% 83% Errors due to Major Avoidance 17% 17% Touched Gates 100% 100% Missed Gates None Recorded None Recorded

2.2.1.2.3 Discussion This study determined the intra-observer and inter-observer reliability for time-motion analysis from a canoe slalom competition using the CAP and the operational definition set developed in the previous section. The dependent variables used in this research to assess reliability were related to the time at which each event occurred in addition to the classification of each of these events.

Analysis of stroke identification revealed that the ability of observers to identify strokes correctly was greater than 78% (the mean inter-observer value), which compared favourably to previous literature on gait (Schokman et al., 2002) and association football (McKenna et al., 1988) that involved the identification of far fewer categories.

Half-correct strokes indicated the amount of multi-strokes that were not identified correctly but still contained part of the correctly identified stroke in them. Results indicated that there was a small amount of confusion about the identification of multi-strokes due to their complexity. This may have been accentuated due to long multi-strokes being broken at slightly different points and thereby presenting as different strokes in the final analysis. Further training may assist in improving this area along with consultation to clarify strokes that are difficult to categorise.

Comparison of the number of stroke identifications and the percentage of correct stroke identifications indicated a trend existed low identification counts and low reliability in identification. It was thought that if a trial was to include a large number of these strokes then the identification reliability would more closely match the identification reliability for all strokes.

The level of detection for penalties (touched gates and missed gates) was exceptional. However, this would be expected to decrease for analysis in windy conditions, when gates are splashed and when touches are less distinctive. Observers were successfully able to detect when avoidance had and had not occurred and were able to categorise the type of avoidance which had occurred. This allows useful information about the paddlers‟ lines (i.e. the path the boat took) to be assessed.

The variation recorded for stroke in and out times was less than the variation for gate split times and turn times due to the ability to use the paddle position relative to the water as a reference for stroke in and out times. The accuracy and variation in gate split times and turn times measured off each gate was dependent on the camera angles for each gate. The reliability for the times recorded for stroke in and out positions, gate splits and turn times was believed to be lower than the land based equivalent such as running or skiing due to the number of times that waves obscured critical measurement components. Also, some of the filming for this study was not as tightly zoomed as recommended for the optimal analysis,

therefore further improvement in analysis could also be expected as a result of improved filming.

Results from all variables revealed that intra-observer analysis was more reliable than inter- observer analysis. This likely resulted from slightly different interpretations of the definitions and differences in the knowledge of canoe slalom among the observers. All three observers in this study were relatively inexperienced in the use of the system and definition set used for analysis. This was due to the definitions, analysis method and analysis system specific to canoe slalom being developed immediately prior to this study. Although some trials were completed prior to the reliability study, it was recognised that learning was still occurring. Therefore, the reliability for both intra-observer and inter-observer analysis would improve further for all variables with continued use.

Applications for this analysis technique include the comparison of an individual‟s performance to top paddlers at the same competition and comparison of multiple runs from an individual to determine the impact of various techniques or strategies on their run time. This analysis method can also be applied to characterise top performers at a competition and compare performances from competition to competition to determine the effect of various courses on an individual‟s performance and strategies.

Application of this method of competition analysis will result in performance improvements through the discovery of methods, strategies and techniques that top performers use. Although each paddler has their own strategy, comparison between individuals using this analysis method will highlight where an individual can gain time through the use of different strategies. Linking strategies and techniques to run time allows paddlers to quantify the effect of different strategies which in turn allows them to determine objectively the fastest technique.

Further research topics relating to the current investigation could include detailing the characteristics of top canoe slalom paddlers from international competition and determining the influence of venue and competition on the techniques, performances and strategies of top canoe slalom performers. This analysis could also be used to assess the strategies an individual uses, determine areas which could be improved through the adoption of alternate strategies and the impact of these changes for an individual.

For performance analysis to be effective it must be capable of detecting differences between high-level athletes and longitudinal changes in high-level athletes. This research is the first step in demonstrating that competition analysis is effective for canoe slalom as it shows that the analysis technique for canoe slalom is reliable. Further research that applies this methodology is required to determine the ability of the method to detect differences in high- level athletes.

2.2.1.2.4 Conclusions The reliability of all the variables analysed in this study were considered acceptable and compared favourably to previous research on gait and association football. The definitions, analysis method and analysis system specific to canoe slalom were novel and it was recognised that learning was still occurring in the performance analysts. Therefore, it is anticipated that the reliability of these performance analysis techniques would improve further with practice after the completion of this study.

Intra-observer analysis for all variables assessed in the current investigation demonstrated greater reliability than inter-observer analysis. This most likely resulted from individual differences in interpretation of definitions in addition to experience of the operators. Therefore, to gain the greatest accuracy and repeatability from such an analysis, a single observer should complete all analyses. This concludes the section taken directly from Hunter et al. (2007) published in Sports Biomechanics.

2.2.1.3 Summary of CAP A customised application (CAP) was successfully implemented using Visual Basic for Applications (VBA) attached to Microsoft Excel® (Microsoft, Washington, United States) as a development platform. Repeatability analysis demonstrated that the definition set and method developed for competition analysis of canoe slalom was reliable and compared favourably to previous research on gait (Schokman et al., 2002) and association football (McKenna et al., 1988) containing fewer variables. Thus, the data gathered through this system was assessed as being systematic and reliable allowing researchers to analyse and interpret the data with confidence in the accuracy.

2.2.2 Paddle Force Curve Analysis Program for Slalom This section describes the development of a program capable of capturing, analysing and reporting paddle force data for canoeing applications. The program initially was required to capture and process raw data from an existing paddle force system. The program initially was required to capture and process raw data from the paddle force system (Jurgen Sperlich, Germany) but be flexible to provide customised reports and allow integration with other devices such as GPS. This paddle force system was assessed for accuracy prior to use as described in the second part of this section.

2.2.2.1 Development The original program was written in VBA (Microsoft, Washington, United States), but had major limitations especially when attempting to store ten channels of 200 Hz data. As a result the program was converted to Visual Basic (Microsoft, Washington, United States) due to the similarity in coding methods. The introduction of the new force system required some major alterations in how data were imported into the program, but once in the program the manipulation remained the same as in the original VBA version (Figure 2.19).

Figure 2.19: Menu structure of paddle force analysis program. The menu structure demonstrates the processing capabilities designed into the software which includes, capturing, replaying, displaying, editing, processing and correcting data.

2.2.2.1.1 Capturing The ability to record data from the pico-11 analogue to digital converter was built in the menus File – Capture File (Figure 2.19). This allows the user to capture the incoming data from this device at 200 Hz. Due to many difficulties in collecting ten channels of data at 200 Hz, as well as providing a graphical image of the incoming data, the program has no visual display while capturing apart from a green box that says Capturing (Figure 2.20). All data are written directly to file using csv format with the filename Kayaking Date and Time .log.

Figure 2.20: The Capture GUI is used to capture data from the German (Sperlich) Force system. The filename generated is time and date coded so files cannot be overwritten.

2.2.2.1.2 Displaying the Data The .log files generated by the capture section of the program can be loaded (File – Load) in the main GUI (Figure 2.19). In addition to the .log files, binary and ASCII files (.txt) generated by the new force system and exported csv files (.csv) can all be loaded into the main GUI. This GUI consists of two sub-panes. The top pane displays an image of the entire data file scaled so it fills the length of the pane (x-axis = time). The scale on the y-axis is changeable but defaults to 500 units above and below a central zero point. Alternate channels are assigned as being up or down to improve visualisation of the data. By clicking on the top pane or dragging the slider below the top pane the user can rapidly navigate to a section of the data which is then displayed in the bottom pane. For finer navigation clicking either side of the slider jumps the entire length of the bottom pane forwards or backwards and clicking the arrows at the end of the slider steps the value of one gridline in the bottom pane. Moving the cursor over the bottom pane displays the value of the current data point (signified by a line on both panes) in a box below the bottom pane (Figure 2.21).

The user has the ability to control how they wish to display the information on the main GUI including which channels are shown, the direction which the visible channels are drawn, the colour and width of the lines for each channel, the maximum value of the y-axis, the length of time viewed in the bottom pane and the interval of the gridlines in both directions. All these features can be changed on the plot settings GUI (Figure 2.22).

Figure 2.21: Main Data Manipulation Page. The upper plot show all the data. The lower plot shows a ten second subsection aligned with the dashed white line in the upper plot and the slider bar. The dashed white line in the lower plot represents the current data point, the values for which are shown in the table at the bottom of the graph. The blue line represents the left hand paddle force (up is positive) and the red line represents the right hand paddle force (down is positive).

Figure 2.22: Plot Settings GUI allows the user to determine the configuration of the data viewed on the main GUI. The plots can be rescaled and the line width selected. The channels turned on or off, switched to be up (positive up) or down (positive down) and the colour of the line can be altered.

2.2.2.1.3 Calibration A calibration is required for each setup of the paddle. The user is required to record and load a calibration data file. Once the file is loaded the user can extract the required calibration information. To perform the calibration Edit – Select – Mark Calibration (Figure 2.19) must be selected which allows the user to left click to start a region and right click to end a region. This process involves a two point calibration, whereby the user selects a section of the trial where the data should be equal to zero, then opens the calibration GUI (Process – Calibrate) and clicks in the zero calibration box, below the zero drop down (Figure 2.23) after checking that the channel is correctly selected. This process is repeated for a second section of known value, currently the default is 200 N, but the user can select or type any value. Once the second value is assigned in the calibration GUI the user can then click the Calibrate button. This calculates the equation of a linear line joining these two points, draws it in the left-hand pane of the calibration GUI and details it in the bottom pane (Figure 2.24). At the same time the entire data set for this channel is recalculated based on this calibration equation so when the user returns to the main GUI the data is scaled correctly. This process is repeated for each channel of data which requires calibration.

Once all channels are calibrated the calibration information can be saved to file for later use by clicking Save on the calibration GUI. This writes the information contained in the bottom pane to a .cal file named the same as the open file name with a calibration tag and the date and time. At this point if a mistake is made all calibration information has to be cleared (Clear All) and started again. Clear All resets the file to its original scaling and length by reloading it. Calibration files can be loaded once saved. Once a calibration is open or completed any file loaded into the main view has these equations applied to it, this includes exported files which are calibrated. Therefore, to view these correctly the calibration has to be cleared (Figure 2.25). The validity and reliability of the calibration of paddle force was reported in section 2.2.2.2 (Paddle force calibration and error assessment).

Figure 2.23: Zero calibration selection (green rectangle in both panes of the main GUI) and assignment (red circle) for channel 2. User left clicks to set the start of the calibration area and then right clicks to set the end. Once in the calibration GUI the user sets the channel and clicks in the circled box to transfer the mean value for the selected section.

Figure 2.24: Second calibration (200 N) selection (green rectangle in both panes of the main GUI), assignment (red circle) and resultant calibration information for channel 2. User left clicks to set the start of the calibration area and then right clicks to set the end. Once in the calibration GUI the user sets the channel and clicks in the circled box to transfer the mean value for the selected section. By clicking Calibrate the data for the channel is recalculated using the calibration.

Figure 2.25: Completed calibration for channels 2 and 3. Once all the required channels have been calibrated the user can save the calibration data for later use.

2.2.2.1.4 Drift or mechanical shift correction Due to drift and offset issues with the original force system, an ability to zero the data set once calibrated was included. This consists of a fixed zero offset where the user can type in the amount to add or subtract from the data, or the user can use the select region (Edit – Select Mark Calibration – Green box) and then click in the fixed offset box to transfer the average value of the section (Figure 2.26). In addition, drift can be corrected through the polynomial modelling built into the software. To enable this function Edit – Select – Mark Zeros is selected. The user can then select as many points through the data set as required to model the zero line (Figure 2.27). The resultant model can be viewed in the top right-hand pane of the calibration GUI and the Pearson‟s correlation coefficient (r2) below it. The model can be changed through the menu in the top right of the calibration GUI Polynomial Model. Selecting First Order fits a linear equation through five points before and after each selected data point. If only a section of the data set needs to be zeroed or there is a sudden shift which cannot be modelled, then by clicking the Section Only button on the calibration GUI, just the data between the first and last selection lines are modelled and adjusted. Once satisfied with the model the user clicks Zero Polynomial or Zero Fixed and the program corrects all the data based on the equation and writes the details in the calibration pane at the bottom of the calibration GUI (Figure 2.28).

Figure 2.26: Zeroing based on a fixed offset or average of a selected area. Clicking in the textbox (red rectangle) transfers the average of the selected region or the user can type the value they want.

Figure 2.27: Zero selection takes five data points either side of each green line on the top pane. Green lines are set by clicking in the bottom pane at the point where the plot should be on zero.

Figure 2.28: Polynomial model of points selected, r2 value and the controls for adjusting the model. Left – Order of the polynomial to be fitted. Right – shows the resultant data remodelled (straight line) and the details of the correction transferred to the calibration pane (Bottom).

2.2.2.1.5 Trimming the Trial Often the start and end of the data set contains no information or are not required. Therefore a trimming function was built in whereby the user selects the data length. Through the trim function the start (Edit – Select – Trim Start) or end (Edit – Select – Trim End) can be removed (Figure 2.19). The data to remove are highlighted with a red box in both panes (Figure 2.29). To finalise selection the user clicked Edit – Delete (Start or End) (Figure 2.19).

Figure 2.29: Selection to delete a section at the beginning of the trial (top) and selection to delete the end of the trial (bottom). This allows the user to trim the file to only relevant data.

2.2.2.1.6 Filtering Many of the trials captured through the original force system include a component of noise from external sources. As a result, a low pass fourth order Butterworth filter with a selectable cut off frequency which filters in both directions to prevent any time shift was built into the program to smooth the data (Figure 2.19 and Figure 2.30). The results of the filter setup with a cut off frequency of 5 Hz can be observed in Figure 2.31.

Figure 2.30: Figure 2.30: Butterworth filter control GUI showing the data input channel to be filtered and the cut off frequency used by the fourth order Butterworth filter which can be set by the operator. In this example, 5 Hz was used to demonstrate the functionality of the filter.

Figure 2.31: Example of the effect of a dual direction low pass fourth order Butterworth Filter with a cut off frequency of 5 Hz on Channel 2 (red data line). The raw data are shown on the left. The smoothed data are shown on the right. This is an example to demonstrate the functionality of the filter.

2.2.2.1.7 Switching Channels of Data Occasionally the channels need to be switched due to errors in the setup of the paddle by the user or for data processing and reporting purposes. The program provides a direct swap of two channels through the Switch Channels GUI (Figure 2.19 and Figure 2.32) the results of this on channels 2 and 3 can be observed in Figure 2.33.

Figure 2.32: Switch channels GUI – the user selects the two channels to be switched.

Figure 2.33: Result of switching channels 2 and 3 (before – left and after – right, red data swapped for blue data and vice versa).

2.2.2.1.8 Exporting and Saving The user enters details of the data set in the Trial Details GUI (Figure 2.19). If no data is entered into the First Name textbox of this GUI before exporting (saving a calibrated data file) the program will prompt the user asking if they would like to enter them before exporting. If the user selects to enter details when prompted the program aborted the export, so once the user enters the details they reinitiate the export. The trial details are not limited to the descriptive labels to their left, but this information is used in the final average reports (Figure 2.34). Exported files store the trial details and all calibration information in the header along with the calibrated data. The user can select to export the entire file, a range (selected area) or the entire file with no header information (this is used for importing into Logan, a customised software program for another device).

Figure 2.34: The Trial details GUI allows the user to enter details used in the file name when exporting the data and when creating reports. If these are not set prior to exporting the user is prompted to enter them.

2.2.2.1.9 Analysis Analysis of the data set is setup to produced average curves based on a region selected in the main GUI (Figure 2.19 and Figure 2.35). Once an area is selected (must be at least eight strokes) the user sets up the analysis parameters, including: the right and left channels, the minimum force required to identify a stroke, the minimum change required to identify the start and end of the stroke as well as the time for the final graph (left at a standard 0.7 seconds) (Figure 2.36).

Figure 2.35: Selecting a section of data to analyse. The user left clicks to select the start of the analysis area and right clicks to select the end of the analysis area.

Figure 2.36: The Analysis settings GUI allows the user to select the right and left data channels and set the threshold values for detecting strokes.

The product of the analysis is an average force curve for right and left with the standard deviation dashed either side of each data point. The line thickness, colour and y-axis can be changed through the Plot Settings GUI prior to analysis. Moving the cursor (vertical line) over the average curve displays the current time and data points. Above the figure the details of the trial entered by the user previously are displayed. These details can be changed by the user if something different to the trial detail is required. Below the figure all the biomechanical parameters relating to the figure are presented, these are described further in Section 2.2.2.1.10 which describes the reporting methods (Figure 2.37). To save this report the user clicks the button at the bottom. This takes a snap shot of this report, the main GUI and exports the average data to a csv file. These files are saved in the same location and named the same as the data file which was open, but with the stroke rate at the beginning.

Figure 2.37: Mean right and left force curves for the selected period with one standard deviation either side and biomechanical stroke characteristics.

2.2.2.1.10 Reporting The creation of the report is described in the previous section (Analysis, 2.2.2.1.9). This section describes the biomechanical data contained in the reports. Each variable (Figure 2.38) is defined on a figure (Figure 2.39 and Figure 2.40) so the numbers become meaningful to coaches and athletes.

Figure 2.38: Biomechanical data relevant to paddle force curves (Figure 2.37) and included in each report. Mean data for the double stroke cycle as well as right and left stroke characteristics are calculated.

Figure 2.39: Time variables presented in the biomechanical table can be defined on an average force curve.

Table 2.11: Temporal paddle force variables describing the timing of events within a paddle stroke cycle. Variable Description Total Time for Double Stroke The time in seconds from entry of the left blade to next entry of the left blade Stroke Rate How many individual strokes a paddler would take in one minute (right and left being separate strokes) Pull Time as % of Stroke Time „Pull Time‟ as a percentage of „Stroke and Glide Time‟ Glide Time as % of Stroke Time „Glide Time‟ as a percentage of „Stroke and Glide Time‟

Figure 2.40: Force and Impulse variables presented in the biomechanical table can be defined on an average force curve.

Table 2.12: Force (N) related paddle force variables analysed within a paddle stroke cycle. Variable Description Newtons Measure of force (100 Newtons (N) = 10.2 Kg approximately) Mean Force The average force applied to the paddle while it is in the water Peak Force The maximum force applied to the paddle while it is in the water Peak Rate of Force Development The maximum change in force during between blade entry and peak force

Table 2.13: Impulse (Ns) related paddle force variables analysed within a paddle stroke cycle. Variable Description Impulse Force multiplied by time (area under the force curve) presented in Newton Seconds Total Impulse Right and Left impulses added together % Total Impulse on Left „Left Impulse‟ as a percentage of the „Total Impulse‟ % Total Impulse on Right „Right Impulse‟ as a percentage of the „Total Impulse‟

2.2.2.1.11 Conclusions The improved force curve program proved successful in the analysis of kayaking performance. The program allows detailed information to be extracted rapidly and provides a tool for the analysis of future canoe slalom as well as sprint kayaking. The analysis software was initially developed for flatwater paddling where paddlers perform forward strokes. Flatwater paddling was initially targeted due to the repetitive nature of the paddling motion and thus the ease of curve detection. In future algorithms could be developed for identification of strokes for canoe slalom. At present the reporting methods used in the software are specific for kayaking on flatwater. However due to the flexibility of customised software it will be possible to manipulate the reporting component of the software to produce information relevant to slalom coaches and athletes. An investigation into how video information may augment the interpretation of paddle force information in canoe slalom is presented in Chapter 6 – Interactions between Strategy and Paddle Force.

2.2.2.2 Paddle Force calibration and error assessment The use of shaft bend as a measure of paddle force requires that the system be calibrated to the flexibility of the shaft (Aitken & Neal, 1992) as there can be considerable variation in the elasticity of different paddle shafts. These paddle force systems make a number of assumptions about the loads applied to them. Firstly the calibration assumed that the point of the pulling force at the paddle shaft (paddlers hand) remained constant and the point of force application at the centre of the blade (centre of pressure of the blade) also remained constant (Marhold & Herrmann, 1981). Deviations must be expected because the paddle varies in depth at entry and has different movements in the water but when the blade is fully submerged in the water the centre of pressure would not alter substantially. If interpretation of the measured values of individual athletes is to be performed, these influences may be

considered as systematic errors (Marhold & Herrmann, 1981). However, caution is necessary when inter-individual comparisons are made (Marhold & Herrmann, 1981). Many of these assumptions may not hold true for canoe slalom. Therefore, the ability to gain meaningful data in whitewater needs to be assessed.

2.2.2.2.1 Methods The paddle force system used in this thesis was made in Germany by Jurgen Sperlich. The system was a wireless system which consisted of two cantilevered optical movement sensors connected to a radio transmitter mounted on the paddle. The data transmitted were collected using a datalogger (minidisc recorder) in the boat and / or a computer based receiver nearby. If the data were collected on the minidisc recorder it had to be replayed in real-time and captured onto the computer through the receiver. Between the receiver and the computer was a 10 channel 10-bit analogue to digital converter (ADC) providing 1024 points of digital resolution. To perform the error assessment the system was setup on the shaft of a paddle owned by the Australian Institute of Sport (AIS). The paddle force system setup procedure was as follows:

Measurements – the paddler was asked to hold their paddle with their hands in the normal paddling position. The paddler then lifted their middle finger on each hand so the locations could be marked on the shaft of the paddle. The centre of pressure of each blade was also marked using a standard distance of 225 mm from the tip of the paddle (Baker, 2008). The outboard measurements were taken as the distance from the paddle centre of pressure to the middle finger mark. The inboard distance was measured as the distance between the middle finger marks (Figure 2.41). The centre of the shaft was also marked for locating the sensors and transmitter (half the paddle length).

Figure 2.41: Paddle measurements taken during the setup of the paddle force system. Left and right centre of pressure and middle finger marks are all required for the calibration of the paddle force system to the flexibility of the shaft of the paddle.

Equipment Attachment – To attach the transmitter, the paddle was held in one hand at the centre of the shaft and allowed to rotate until the paddle stopped rolling. With the paddle in that position the transmitter was attached on the top of the shaft oriented with the left and right sensor plugs at the correct ends. Both sensors were attached to the paddle by first unscrewing the adjustment bolt so that it did not touch the shaft of the paddle, and then clamping it to the shaft lightly so the sensors could still be moved.

Tripod Setup – Two tripods with special paddle calibration heads were setup, so that the long head was perpendicular to gravity (using a spirit level) (Figure 2.42). The paddle was placed so that the centre of pressure for the left blade sat on the tripod. The other tripod was positioned so that it was located under the right hand middle finger mark. The tripod was adjusted so that the paddle shaft was perpendicular to gravity (using a spirit level) (Figure 2.43)

Figure 2.42: End view of the tripod setup for the paddle force calibration procedure.

Figure 2.43: Side view of tripod setup (setup for calibrating left side) for the paddle force calibration procedure.

Setup Sensor – The left hand sensor was adjusted so that it was perpendicular to gravity (on top of the shaft) then the clamp bolts tightened so the sensor was held firmly on the paddle. The adjustment bolt was then screwed in until it just touched the shaft and then tightened a

half turn further (this was checked live in the software ensuring that a value 200 bits was achieved from the ADC), then the locking screw was tightened to secure the setting. The shaft of the paddle was then flexed a few times by pushing down on the left hand grip to ensure the sensor was properly seated.

Other Side – Steps 3 (Tripod Setup) and 4 (Setup Sensor) were repeated for the right hand side of the paddle.

Static Calibration – To calibrate the left side of the paddle, the shaft of the paddle was supported at the right middle finger mark and the left blade centre of pressure as described in step 3 (“Tripod Setup”) of the “Setup”. This calibration was performed in the software on the computer. The physical procedure to get the data for calibration involved a Zero where the paddle is left for 10 – 20 seconds to record an unloaded, zero offset measurement followed by a Scale where a weight (20 kg, 196 N) was suspended from the left middle finger mark (Figure 2.44). This weight was left for 10 – 20 seconds to record a loaded measurement. The software assumed linearity of the sensors then used these points to convert the bend in the shaft to an adjusted unit value (Newtons) (Oka et al., 1983) of the force being produced at the handgrip. These steps were then repeated for the right hand side of the paddle.

Figure 2.44: Set-up for Calibration of the Force Transducers (left) showing tripods supporting the paddle with the calibration weight hanging from the hand grip.

To assess the accuracy of the system a zero stability test and a linearity test were performed. The zero stability of the system was assessed by leaving the system for a 1-minute period. Then the stability of the system was assessed through the mean, standard deviation, maximum and minimum. The linearity of the system was assessed through application of a series of weights from 1.25 kg up to 45 kg. Difficulty in maintaining the positioning of the paddle in the calibration limited the success of weights greater than 45 kg. The output of the paddle force system was then correlated to the mass of the weights applied.

2.2.2.2.2 Results The zero stability tests showed that the mean of all sensors was varied based on the pretension which was applied thus all datasets were normalised. For all sensors the typical minimum was -7.84 digital units which represented -0.77% of the full scale (FS) and the maximum was 6.56 digital units representing 0.64% of FS. The typical deviation was 0.53 digital units which represented 0.05% of FS (Figure 2.45).

Figure 2.45: Zero stability test results for all sensors demonstrating the normalised mean of each unit and the stability of the signal about this point. Each plot shows the minimum and maximum as the tails (black lines), and the standard deviation of the sample (yellow box). The left plot represents the raw digital units and the right plot represents the percentage of the full scale data.

The linearity of the system demonstrated high r2 values between the paddle force system‟s measured weight and the actual weight applied. There was some deviation between the paddle force system and the applied loads (Figure 2.46).

Figure 2.46: The linear calibration of applied load (y-axis) and the paddle force system measurement (x- axis), together with the regression equation.

2.2.2.2.3 Conclusions The zero stability of the system was found to be acceptable with the range of less than 2% of the full scale and the typical deviation of 0.05% of full scale. The linearity of the paddle force system demonstrated strong correlations to the actual weight applied and the residual absolute variance in the linearity data was on average 5.1 N and the standard deviation in the residual variance was 7.6 N. The repeatability of the system was deemed suitable for providing paddle force information to paddlers and coaches.

2.3 Slalom Analysis in coaching and Racing Environment 2.3.1 Analysis of Canoe Slalom Category Clustering Race time in canoe slalom is the most commonly used performance indicator. However, in canoe slalom the course is different at every event, thus for comparisons between events, the results were normalised to the fastest paddler at the competition. Given that K1M, C1M, K1W and C2M are the most well established categories in competitive canoe slalom, it was necessary to understand the similarities and differences between each category to ensure that the research outcomes of this thesis are applicable across the sport as a whole.

Rankings were calculated from the competition results data obtained from the International Canoe Federation (2009b) website. The ten final times including penalties from the three world championships between 2005 and 2007, the 2004 and 2008 Olympic Games were used in this analysis. The average time for each athlete in each canoe slalom category relative to the fastest paddler from each competition (K1M) was calculated. The average ranking of each paddler when all categories were combined was then determined. To highlight similarities which have occurred between K1M and C1M, the results for the two categories at the 2004 Olympic Games in Athens were presented in combined rank order.

To assess how each canoe slalom category has improved over time, competition results data from the International Canoe Federation (2009b) website and the historical competition results data from Endicott (2006) published on the International Canoe Federation (2009b) website were used. The winning time from K1M, C1M, K1W and C2M from every World Championship and Olympic Games between 1961 and 2008 were collated from the international canoe federation website excluding the 1991, 1993 and 1995 World Championships whose data were not available. The winning times for each year were compared to determine trends in race performance and course design over time. The K1W and C2M data from 1973 were found to be substantially slower than K1M and C1M due to river conditions changing during the event, therefore these were removed from the trend analysis. The issue of course difference was removed by normalising to the fastest paddler at the competition.

2.3.2 Canoe Slalom Competition Analysis Program (CAP) Utilising the competition analysis program (CAP) along with the definition set developed for canoe slalom, which were analysed for reliability in Chapter 2, Section 2.2.1, an analysis of an international competition was performed. This section describes the methodology which was used to collect and analyse these data.

For this investigation a turn was defined as up to four split times recorded around an upstream gate depending on the camera angle and course design. Figure 2.47 illustrates the four times taken. Turn times one and four were defined as the point where the centre of the boat passed parallel to the gate line. Turn times two and three were defined as the point where the centre of the boat passed a line perpendicular to the gate line originating from the inside pole.

Figure 2.47: The section of the upstream turn analysed was from position one to position four. Turn times one and four were defined as the point where the centre of the boat passed parallel to the gate line. Turn times two and three were defined as the point where the centre of the boat passed a line perpendicular to the gate line originating from the inside pole.

Thirty competition runs from the semi-finals and finals of the 2005 World Championships (Figure 2.48) were filmed using three Sony® (Minato, Tokyo, Japan) digital video cameras (DSR-PDX10P PAL). Each camera was situated in the best position to view its respective section of the course, whilst overlapping with the other sections. These cameras recorded onto Sony® (Minato, Tokyo, Japan) digital video cassettes (DVM60) and were time stamped

to an accuracy of 1/50th of a second. To aid analysis, the paddler was continually framed using the zoom and pan so they filled the frame and the shutter speed for both cameras was set to faster than 1/1000th of a second. This footage was subsequently captured onto computer using video editing software capable of producing a sample rate of 25 Hz.

The ten fastest runs from K1M, C1M and K1W were selected for analysis. Each trial was analysed by one of the three trained observers using CAP, the customised canoe slalom time- motion analysis program described in Chapter 2 Section 2.2.1. A single observer checked the data from each run following the analysis of the run to ensure that there were no errors.

Gate split times, turn times, total stroke information, left and right stroke information and gate errors were extracted into separate tables so that statistical analysis could be performed on the data. Two turning methods were identified for analysis including the spin, which was defined as when the paddler turned away from the next gate, turn angle is greater than 180 degrees and pivot, which was defined as when the paddler turns towards the next gate, turn angle is less than 180 degrees were used to assess strategy. The strategy paddlers used between gates 9 and 15 (Figure 2.48) were used to group paddlers into one of eight categories which included: Spin-Spin-Spin, Pivot-Pivot-Pivot, Pivot-Pivot-Spin, Spin-Spin-Pivot, Spin-Pivot- Pivot, Pivot-Spin-Spin, Spin-Pivot-Spin and Pivot-Spin-Pivot. For gates 12 and 15 paddlers were grouped into four different strategies: Spin-Spin, Pivot-Spin, Spin-Pivot, Pivot-Pivot.

Figure 2.48: Canoe Slalom 2005 World Championships Finals Course Map (Australian Canoeing, 2005). The red rectangle highlights the most complex sequence of gates on the couse (Gates 10 to 15). This section required paddlers to rapidly change direction numerous times which is why it was chosen for this analysis.

Descriptive statistics and a one-way ANOVA with a Scheffé post hoc test were used to analyse split times with respect to category and strategy, time spent in each quarter of an upstream gate with respect to category and the number of touches and major and minor avoidances with respect to category. To assess variance for split times over the course, descriptive statistics and the coefficient of variation were used. Gate by gate analysis of strategy and comparison between Pearson‟s correlations for split time / run time and turn time / run time were assessed using a two-tailed t-test in which equal variance was not assumed. Comparison of right and left sides in men‟s kayak / women‟s kayak and preferred (onside) and non-preferred (offside) for men‟s canoe with respect to stroke counts, stroke types and categories were assessed using descriptive statistics. To assess the relationship between

stroke time / run time and stroke count / run time a Pearson‟s correlation was used. The significance level was set at α < 0.05.

2.3.3 Canoe Slalom Boat Trajectory while Negotiating an Upstream Gate out of Competition A six-gate canoe slalom course designed in consultation with elite canoe slalom coaches was constructed on the top quarter of a whitewater course with five pumps generating whitewater flow at 14 m3/s (Penrith, Australia). The flow conditions were those used for the 2005 Canoe Slalom World Championships. The course involved four downstream gates and two upstream gates (Figure 2.49); the analysis in this investigation focused on Gate 2 which was a left hand upstream gate.

Seventeen males participated in this study (11 single kayak (K1M) and six single canoe (C1M) paddlers) after giving informed consent (Appendix A). All procedures were approved by the Human Ethics Committee of the University of Canberra and the Australian Institute of Sport. Of the C1M paddlers five were right handed and one was left handed. The data from the left handed were included with the right handed paddlers because analysis revealed no differences between the strategies used with respect to the variables analysed. Each participant used their own boat and paddle which conformed to International Canoe Federation regulations. All participants were ranked in the top 40 for a World Cup or the World Championships in 2005 for canoe slalom.

Paddlers performed six trials on the course, each of which took approximately 20 seconds. Five minutes were allocated to perform the trial and walk back to the start to ensure full recovery between trials. Trials where the paddler touched either of the gate poles were excluded from the analysis.

Video footage of Gate 2 was captured using two high speed cameras (Phantom V4.2 colour running firmware version 299; Vision Research Inc., United States of America) fitted with a 16-100 mm f/1.9 zoom lens. This footage was captured through a computer running the Phantom software V630. The cameras captured at 100 Hz with a resolution of 512 x 348 pixels and a shutter speed between 1/1000th and 1/500th of a second. The cameras were positioned 2.8 m above the water and oriented as defined in Figure 2.49.

Figure 2.49: The canoe slalom course and equipment setup used for testing, drawn to scale. The dashed line represents the approximate path of a paddler completing the course from right to left. Black circles represent the location of the two highspeed cameras which were focused on Gate 2 (the dotted rectangle).

A custom portable calibration rig was used to calibrate the motion analysis space; (1) a base frame that was levelled in the bed of the course when no water was running and (2) a movable „T‟ section which was attached to each corner of the base frame and levelled (Figure 2.50). The volume calibrated for the high-speed cameras was 6 m x 5 m x 1.5 m. Twenty calibration points distributed throughout the volume were entered into the commercial software (Ariel Performance Analysis System, United States of America) which uses a three dimensional direct linear transformation to convert digitised image-based coordinates into real world coordinates.

Figure 2.50: Calibration Rig. The base frame (a) was placed in the bed of the course and levelled using the telescopic sections. To calibrate the volume the movable T section (b) was attached to the base frame above each leg in turn and levelled.

Soft, fluorescent orange hemispherical markers, 40 mm in diameter were attached to the boats and helmet of each paddler to create visible landmarks on the boat and head which could be digitised easily. The location of markers five and six were 0.05 m wider for C1 compared to K1 (Figure 2.51) because of the difference in boat width.

Figure 2.51: Boat marker locations. The location of the hemispherical markers attached to each boat. The top half represents a single kayak (K1) and the bottom half a single canoe (C1).

Results from a frequency analysis (Fast Fourier Transformation) on a trial revealed that the majority of the signal was below 1 Hz. The video footage was manually digitised at 25 Hz using commercial software (Ariel Performance Analysis System, United States of America) to reconstruct three-dimensional kinematic information about the boat and paddler‟s head during the upstream gate.

Repeatability analysis revealed that the mean absolute difference of every data point from all markers across the three digitising repetitions to the mean was 0.01 m, the maximum was 0.04 m and the maximum absolute difference of every data point from all markers across the three repetitions to the mean was 0.07 m. The raw x, y, z coordinates were then filtered using a 5th order dual pass Butterworth filter and a cut off frequency of 2 Hz, which was selected from the results of the fast Fourier transformation and the Nyquist rate (twice the band width).

A principal component factor analysis of all variables extracted from the analysis was performed to determine the variables which were most relevant to upstream gate performance. The methods, results and discussion of this factor analysis are described and presented in Appendix B.

Total time was measured as the time taken for the head to travel from positions one to four and boat trajectory was defined as the mean distance between the paddler‟s head and the inside pole between positions one and four (Figure 2.49). Only data between positions one and four were analysed because this is where the majority of the turn occurs and reacceleration of the boat occurs after position four. This minimised the influence of individual paddling velocity on results while accentuating the effect of the strategy used for the actual turn. To compare the line taken around the gate by the two fastest and the two slowest K1M and C1M paddlers the average of each data point was determined.

A Pearson‟s two-tailed correlation was used to determine the strength of the relationship between total time and boat trajectory. To determine if this relationship remained across individuals a Pearson‟s two-tailed correlations was performed on each participant and the linear regression equations parameters calculated. The variance (coefficient of variation) and standard deviation of total time and boat trajectory across all analysed trials were correlated to total time to determine if trial variability increased as total time increased. A one-way analysis of variance was performed (ANOVA) on the four groups of paddlers (the fastest vs. the slowest two paddlers in each class (K1M and C1M)) with total time and boat trajectory as dependent variables to determine if differences exist between the fastest and slowest strategies in this population. A post-hoc Scheffé was used to assess differences between the groups.

2.3.4 Interaction between Strategy and Paddle Force This study involved combining data from video based time-motion analysis using the competition analysis software developed as described in Section 2.2.1 – Competition Analysis Program (CAP) for Canoe Slalom with data from paddle force sensors as described in Section 2.2.2 – Paddle Force Curve Analysis Program for Slalom. The data used for this comparison were collected on the simulated course described in Section 2.3.3 – Canoe Slalom Boat Trajectory while Negotiating an Upstream Gate (Figure 2.49).

To determine the split times, turn times, stroke in and out times, stroke side and type of stroke each trial was recorded using a digital video camera and then coded manually using the CAP (Section 2.2) at 25Hz. To categorise the types of strokes which the paddler took during the trial the definition set was used (Section 2.2). The two data sets were overlaid to generate a timeline of force application, trial events and stroke types (Figure 2.52).

Figure 2.52: Key to understand the Force and CAP Run Profile. This shows an example overlay combining force and CAP data with all of the components and lines labelled and described.

2.3.5 Equipment Setup – Canoe Slalom Paddle Length Comparison One K1M international standard canoe slalom paddler was involved in this investigation. The two paddle lengths were 1960 mm (Short) and 2040 mm (Long) which were -4cm and +4cm from the paddlers original paddle. Both paddles (Table 2.14) were from the same manufacturer (Solution Paddle Gear, East Perth, Australia) and were the same design with a joiner in the middle to adjust the length of the shaft. The distance between the paddler‟s hands and the tip of the blade were standardised for both paddles. The paddler used both paddles in alternate training session for two weeks prior to the testing. The boat and other paddling equipment used during the trials were the paddler‟s normal equipment and were not altered during testing. All testing was completed on flatwater to control for as many factors as possible.

Table 2.14: Paddle Measurements (from tip of paddle) used in the Equipment Set up Study Paddle Blade Hand Sensors Type Length Type Length Width 1 2 1 2 Short 1.96 m Solution 0.48 m 0.208 m 0.655 m 0.655 m 0.84 m 0.84 m Long 2.04 m Solution 0.48 m 0.208 m 0.655 m 0.655 m 0.84 m 0.84 m

After giving informed consent (Appendix C) and completing their typical warm up procedure the athlete performed 12 m sprints with 2 minutes recovery. The paddler performed 12 sprints alternating between the long (n = 6) and short paddle (n = 6). The athlete was then allowed to perform their cool down. All procedures were approved by the Human Ethics Committee of the University of Canberra and the Australian Institute of Sport.

The force applied to the paddle by the paddler was measured using the Jurgen Sperlich force system setup and calibration as described in Section 2.2.2.2.1. To compare trials, each trial was aligned based on the first major increase in force and then the sum of the average impulse for each condition (short and long paddle) was calculated.

To determine the split times and stroke times each trial was recorded using a 50 Hz camera positioned on a tripod on the bank in line with the finishing pole. Gates were setup for the start and finish. The video was coded manually using the competition analysis software (Section 2.2) at 25 Hz. Split times were measured from the first movement of the first major forward stroke, regardless of any movement occurring prior to this point, to the frame where the paddler‟s torso was in an equivalent position relative to the finish pole as it was relative to

the start pole at the first movement. Stroke times, measured as the time the blade spent in the water, were recorded for ten strokes for each trial.

In addition to the time measures, the forward acceleration and velocity of the boat were measured using a MinimaxX unit (Catapult, Scoresby, Australia) attached to the deck in front of the cockpit of the kayak. This unit measured velocity using GPS signal at 1 Hz and acceleration through 3D accelerometers at 100 Hz. Data from the accelerometers were calibrated and then scaled to real units (9.8 m.s-2). Once real units had been applied a zero offset for the boat was found for the forward sensor and then this offset was applied to all trials. Velocity was calculated from the acceleration data. Each trial was aligned based on the first major increase in velocity and then plotted separately for each condition. Trial 1 was removed due to the large difference between this trial and the mean of either condition (short or long paddle). The average velocity for each condition (short and long paddle) was calculated allowing comparison of the ability to get the boat moving with the short and long paddle. The difference between the two paddles for split-times, stroke times and velocity were tested using and independent 2-tailed T-test with significant set at p = 0.05.

100

CHAPTER THREE

3 ANALYSIS OF CANOE SLALOM CATEGORY CLUSTERING

3.1 Results Combining K1M, C1M, K1W and C2M categories showed that C1M paddlers would consistently place between 5th and 10th in K1M competition (Figure 3.1). Based on time taken to negotiate the course, paddlers from K1W and C2M categories are evenly matched but only overlap at the tail end of the K1M and C1M categories (Figure 3.1).

C1 K1 C2 K1W

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 Place in Combined Category

Figure 3.1: Comparison of average placing for K1M, C1M, K1W and C2M in a combined category across world championships and Olympic Games between 2004 - 2008. Data sourced from International Canoe Federation (2009b).

When the data are normalised by the winning times at each competition, eight of the ten K1M paddlers are tightly grouped between 100% and 105% with the slowest two K1M paddlers recording times greater than 112% of the fastest paddler (Figure 3.2). Paddlers in C1M are all grouped between 103% and 112% of the fastest time with little increase in distribution in the slower half of the group. The top half of K1W and C2M paddlers based on percentage are evenly matched with times between 111% and 122% of the fastest time, but the last two C2M and K1W are separated from the rest of the group with times greater than 130% of the fastest time (Figure 3.2).

101

C1 K1 C2 K1W

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 100% 105% 110% 115% 120% 125% 130% 135% 140% 145% 150% Percentage Placeof fastest in Combined time in Combined Category Category

Figure 3.2: Comparison of average percentage of the fastest time (K1M) for K1M, C1M, K1W and C2M in a combined category across world championships and Olympic Games between 2004 - 2008. Data sourced from International Canoe Federation (2009b). Legend as per Figure 3.1.

Comparison of K1M and C1M in any of the international competitions analysed for this study showed that the fastest C1M paddler came to the fastest K1M paddler was 100.6% (1.2 seconds slower) at the 2004 Olympic Games in Athens. At this event, C1M paddlers would have placed second and third in a combined race with K1M paddlers, and the spread of results (Table 3.1) suggests that at the elite level, C1M paddles are now evenly matched with K1M paddlers, even though they use only one blade. However, a confounding factor is that the finals in Athens were run two days apart so different conditions may have impacted on the times achieved by each category.

Table 3.1: Combined Results for K1M and C1M at the 2004 Olympic Games in Athens. Results sourced from the International Canoe Federation (2009b). Category Paddler (Nationality) Total Time (s) Difference from fastest (s) K1M Benoit Peschier (Fra) 187.96 C1M Tony Estanguet (Fra) 189.16 + 1.20 C1M Michal Martikan (Svk) 189.28 + 1.32 K1M Campbell Walsh (Gbr) 190.17 + 2.21 K1M Fabien Lefevre (Fra) 190.99 + 3.03 C1M Stefan Pfannmoeller (Ger) 191.56 + 3.60 K1M David Ford (Can) 192.58 + 4.62 C1M Robin Bell (Aus) 192.83 + 4.87 K1M Thomas Schmidt (Ger) 192.93 + 4.97 K1M Scott Parsons (USA) 194.76 + 6.80 C1M Tomas Indruch (Cze) 195.28 + 7.32 K1M Grzegorz Polaczyk (Pol) 196.57 + 8.61 K1M Sam Oud (Ned) 197.28 + 9.32 K1M Warwick Draper (Aus) 197.43 + 9.47 C1M Simeon Hocevar (Slv) 199.78 + 11.82 C1M Jordi Sangra (Spa) 200.41 + 12.45 K1M Uros Kodelja (Slv) 201.61 + 13.65 C1M Stuart Mcintosh (Gbr) 211.19 + 23.23

102

The International Canoe Federation World Championships results from the last 50 years show that race times became progressively shorter until the introduction of the 90 s minimum course time rule in 1997. After this, there was a sudden increase in course time and then stabilisation around 200 s for the combine run time (Figure 3.3).

Comparison of results relative to K1M highlighted that in 1973 there were dramatic differences between categories due to changes in course conditions. Data demonstrated that all categories have improved relative to K1M over the last 50 years, particularly C1M paddlers who now complete courses within 105% of the K1M time (Figure 3.4). The greatest improvement in C1M relative to K1M occurred between 1975 and 1980. However, for the last ten years the change relative to the percentage of the winning time (K1M) for C1M, K1W and C2M are similar with steady improvements (Figure 3.4).

Figure 3.3: Combined course time in seconds for both runs (semi-finals and finals) for the last 50 years for each category. Missing data are represented as a gap. A 90 s minimum course time rule was introduced in 1997.

103

Figure 3.4: Improvement of C1M, K1W and C2M relative to K1M over the past 50 years (Excluding 1973 for C1M and K1W due to changes in river conditions). Missing data are represented as a gap.

3.2 Discussion Comparison of the top ten paddlers from each category showed that K1M paddlers had the lowest variability in run time as a percentage of the fastest run time for any given course. This suggests that small changes which result in a reduced run time would allow a paddler to improve their rank. Conversely, any errors or time penalties would cause the paddler to be out of contention.

The run times of K1M and C1M were not substantially different which indicated that C1M were not limited by only having one blade for the majority of the course. At the 2004 Athens Olympic Games the fastest C1M paddler was just 1.2 seconds slower than the fastest K1M paddler and C1M paddlers would have placed second and third in a combined category. There are a number of possible factors which may have contributed to this close match between K1M and C1M such as, the competitions were two days apart so the condition may have varied, the course design was more sympathetic to the C1M than other course, and the ability of the athletes competing in C1M was better suited to the course and conditions. Even with though there were unique characteristics about this course which allowed C1M to achieve times similar to K1M the overall trend has been that C1M have been consistently

104

close to K1M. Thus due to differences in paddle configuration it would be expected that C1M paddlers are slower for sections of a course that require strokes on their offside. Therefore, C1M paddlers must utilise onside strokes very efficiently or practice superior strategic skills to K1M paddlers to compensate for the limitations on their offside. However, a confounding factor is that the finals in Athens were run two days apart so different conditions may have impacted on the times achieved by each category.

The normalised times for K1W and C2M categories were significantly slower than K1M and C1M. K1W paddlers may have slower run times because of strength differences which limit their ability to perform technical manoeuvres used by K1M paddlers. The slower run times of C2M may result from difficulties of getting two paddlers and a larger boat around and through the gates. Furthermore, the two C2M paddlers have to match different strokes to achieve the desired boat direction while being restricted by the position of the other crew member. The variability of run times and the rates of improvement over the last 50 years for K1W and C2M also demonstrate that these categories are not as competitive as K1M. This suggests that K1W and C2M possess greater potential to improve in the future, both in terms of the percentage of the winner‟s time and the distribution of times within each category. However the race time differences may also be driven by the population of athletes participating in each of these categories. In C2M the requirement to have one left and one right-handed paddler limits the combinations of boats, thereby restricting the population of available paddlers. The slower run times of C2M may result from difficulties of getting two paddlers and a larger boat around and through the gates. Additionally, having two paddlers and a larger boat also causes C2M to have greater momentum than the other categories, impacting on their ability to change direction and negotiate upstream gates.

Every canoe slalom competition uses a different course, therefore comparisons can only be made for athletes on the same course. To extend the number of strategies that could be looked at, a coach and paddler should consider the other categories. The bronze medallist from K1M (Fabien Lefèvre) at the 2004 Athens Olympic Games was described in 2006 as having technique similar to a C1M paddler, such as the way he pushed off the walls and his ability to perform an upstream gate on one stroke (Endicott, 2006). The reason for the similarity between his technique and a C1M paddler‟s technique stems from the fact that he copied what they were doing. Fabien Lefèvre was quoted as saying “I studied the C1s a lot and saw they turned faster than we did. So I asked myself why and I tried to copy them.”

105

(Endicott, 2006, pp. 1, Section VI). Not only was Fabien Lefèvre trying to copy the technique of C1 paddlers but he changed his boat to achieve some of the characteristics of a C1 “I tried to evolve my boat to approach the C1 shape but I knew I could never have it exactly like theirs. It's a mix between C1 technique and what you can do in a kayak.” (Endicott, 2006, pp. 3, Section VI). One prominent feature of Fabien Lefèvre‟s C1 based technique was the ability to negotiate an upstream gate in one stroke. This required the paddle to be used in a different manner. “It's not there (lower hand) but here (upper hand) that‟s important. Because canoes use their hands a lot here (upper hand). I tried to work with that and then redistribute the work load between the lower hand and the upper hand. So, mobility here (upper hand). C1s use that a lot and I tried to do it in K1, too, but it wasn't always easy” (Endicott, 2006, pp. 3, Section VI).

3.3 Conclusions Over the past 50 years C1M have demonstrated the greatest improvement in race time relative to K1M and are now achieving run times within 3% of K1M despite limitations associated with only having one blade. In contrast, K1W are within 10% of both K1M and C1M. To understand why someone within a particular category consistently outperforms paddlers in other groups, a more detailed analysis is required that can provide greater information regarding a paddler‟s performance during competition. The K1W category shows the greatest potential for improvement, both in terms of race times as well as variance in ability in the group. An analysis of K1M, C1M, K1W and C2M in competition performance may provide an insight on those performance variables that would lead to improvements for all four categories, especially K1W and C2M. This would then focus the attention of coaches to these variables when correcting technique and developing racing tactics.

106

Chapter 4

This chapter has been removed due to copyright restrictions.

This chapter is available as:

Hunter, A., Cochrane, J., & Sachlikidis, A. (2008). Canoe slalom competition analysis. Sports Biomechanics, 7(1), 24‐37.

Links to this chapter:

Print http://webpac.canberra.edu.au/record=b1594164~S4 Online http://ezproxy.canberra.edu.au/login?url=http://search.ebscohost.com/logi subscribed n.aspx?direct=true&db=s3h&AN=33763897&site=ehost‐live content (UC community) Online general http://www.tandfonline.com/doi/abs/10.1080/14763140701683155 public DOI 10.1080/14763140701683155

Chapter 5

This chapter has been removed due to copyright restrictions.

This chapter is available as:

Hunter, A. (2009). Canoe slalom boat trajectory while negotiating an upstream gate. Sports Biomechanics, 8(2), 105‐113.

Links to this chapter:

Print http://webpac.canberra.edu.au/record=b1594164~S4 Online http://ezproxy.canberra.edu.au/login?url=http://search.ebscohost.com/logi subscribed n.aspx?direct=true&db=s3h&AN=41539585&site=ehost‐live content (UC community) Online general http://www.tandfonline.com/doi/full/10.1080/14763140902934837#previe public w DOI 10.1080/14763140902934837

CHAPTER SIX

6 INTERACTIONS BETWEEN STRATEGY AND PADDLE FORCE

6.1 Introduction Chapters 4 and 5 demonstrated that the time taken to complete a course was influenced by the tactics a paddler used. Each tactical manoeuvre requires the paddler to control the boat through the paddle, thus paddle force information could provide insights into how this control is executed (Aitken & Neal, 1992; Sperlich & Klauck, 1992). However, the use of paddle force data alone does not indicate if a C1M paddler is performing an onside (preferred side) or offside (non-preferred side) stroke. The paddle force data lacks event information, thus lacks the relative timing of strokes compared with the run, as well as the description of the type of stroke that the paddler performing. In contrast, video recording provides temporal information about the trial but does not describe the paddler‟s effort.

This chapter determines if paddle force combined with stroke and course data would provide informative feedback for canoe slalom coaches and athletes. In addition, the way in which a paddler loses time during a course relative to repeated runs on the same course is assessed to quantify paddler variability. To address these research areas a case study involving six repeated runs of a five gate course was conducted with a C1M paddler of international standard. Competition analysis and paddle force data from each run were combined and the results presented in a timeline-based graphical format. For methods and definitions of variables, refer to Chapter 2 Section 2.3.4.

6.2 Results The combined results demonstrate that paddle force data and competition analysis data sets complement each other. Paddle force data can be described by competition analysis data. Comparison between Figure 6.1 through to Figure 6.6 illustrates the consistency with which the paddler completed the six runs. Normalisation of sections of the course demonstrates how the timing and characteristics of strokes are dependent on the course (Figure 6.7 to Figure 6.9).

130

Comparing Figure 6.1 with Figure 6.2 to 6.6, the paddler came into the second upstream gate more slowly than on other runs, moved more quickly through the first quarter (position one- two) but took much longer to manoeuvre the second quarter (position two to gate) (Figure 6.1). A similar entry situation was found in Run 5 (Figure 6.5) with rapid movement through the first quarter (position one-two). The difference is that in Run 5, force was maintained on the paddle whereas in Run 1, the paddle was withdrawn. It can be observed that in the first run the paddler lost approximately one second around the second upstream gate.

Clear trends between paddle force and the type of stroke identified were observed in relation to each force curve and the associated time-motion analysis data. For example, forward, C, draw-forward, reverse and brace strokes could be identified in the paddle force curves. Forward strokes were typically a single peak curve with an amplitude of 500 N, a linear rate of force development (straight line from no force to maximal force) and a duration of ~ 0.6 s (Figure 6.10 a). C strokes were typified by a magnitude of 450 N and a duration of ~ 1 s. The initial rate of force development in the C strokes had irregularities compared to forward strokes (Figure 6.10 b). Draw-forward strokes presented three different profiles dependent of the section of the course which it was being performed. However, typically all had two or more peaks and were generally long with a duration of ~ 2 s (Figure 6.11 a). The paddler recorded negative forces up to -200 N. This was a result of the paddler using the back of the blade (Figure 6.11 b) and was probably a reverse or reverse sweep stroke. This type of stroke was consistently miscoded. One brace stroke was identified which used the back of the blade briefly (Run 3, Figure 6.3 at 2 s).

In Run 1 (Figure 6.1) the paddler was 1 s slower (with a run time of greater than 18 s) than on other runs (all less than 17 s). During the execution of Gate 2 the paddler made a poor choice of path and stroke which can be observed through the split times around the gate. Prior to entering Gate 2, the paddler withdrew the paddle from the water to take a second stroke, whereas for the faster runs, the force was maintained during this period. In Run 5 (Figure 6.5) similar entry conditions to Gate 2 as in Run 1 were used, however the paddler kept the paddle in the water and the boat continued to turn around the gate. Comparison of Gate 1 in Run 1 and Run 2 also demonstrates that the paddler lost time on the execution of the upstream gate from position one to position four (Figure 6.12).”

131

Paddle force (N) force Paddle

Time (s)

Figure 6.1: Overlay of competition analysis and paddle force data for Run 1 of 6. Please refer to Figure 2.52 for a description of each component of this figure.

Paddle force (N) force Paddle

Time (s) Figure 6.2: Overlay of competition analysis and paddle force data for Run 2. Please refer to Figure 2.52 for a description of each component of this figure.

132

Paddle force (N) force Paddle

Time (s)

Figure 6.3: Overlay of competition analysis and paddle force data for Run 3. Please refer to Figure 2.52 for a description of each component of this figure.

Paddle force (N) force Paddle

Time (s) Figure 6.4: Overlay of competition analysis and paddle force data for Run 4. Please refer to Figure 2.52 for a description of each component of this figure.

133

Paddle force (N) force Paddle

Time (s)

Figure 6.5: Overlay of competition analysis and paddle force data for Run 5. Please refer to Figure 2.52 for a description of each component of this figure.

Paddle force (N) force Paddle

Time (s) Figure 6.6: Overlay of competition analysis and paddle force data for Run 6. Please refer to Figure 2.52 for a description of each component of this figure.

134

Paddle(N) force

0% 50% 100 % Percentage of Start to Gate 2

Figure 6.7: Paddle force data fore each run normalised beteen the start and set up for gate 2. This demonstrates that the paddler had similarity in the exicution of each run with respect to the magnitude and type of strokes taken. Note that in the 2nd run (green) the paddler utilised a different stroke pattern to manouver after gate 1.

135

Paddle(N) force

0% 50% 100% Percentage of Gate 2 to Gate 3

Figure 6.8: Paddle force data fore each run normalised beteen the set up for gate 2 and gate 3 . This demonstrates that the paddler had similarity in the exicution of each run with respect to the magnitude and type of strokes taken. Note that in the 1st run (red) the paddler utilised a different stroke pattern to manouver after gate 2.

136

(N)

Paddleforce

0% 50% 100% Percentage of Gate 3 to Gate 5

Figure 6.9: Paddle force data fore each run normalised beteen gate 3 and gate 5. This demonstrates that the paddler had similarity in the exicution of each run with respect to the magnitude and type of strokes taken. Note that in the 2nd and 3rd run (dark blue and light blue) the paddler utilised a reverse sweep stroke pattern to manouver after gate 4.

137

Paddle force Paddle (N) Paddle force Paddle (N)

0% 100% 0% 100% Percentage of Stroke Percentage of Stroke

Figure 6.10: a) Example of the normalised force profile of typical forward stokes. b) Example of the normalised force profile of a C stroke.

Paddle force Paddle (N) Paddle force Paddle (N)

0% 100% 0% 100% Percentage of Stroke Percentage of Stroke

Figure 6.11: a) The normalised force profile of a draw-froward stroke pattern exicuted on the entrance to gate 1. b) Example of the normalised force profile of typical reverse-sweep stokes.

138

Figure 6.12: Demonstration of how the paddler lost time on a single upstream gate from position one to position four. Note that the paddler has lost the majority of the time prior to reaching the gate line, but still loses time on the exit as well.

139

6.3 Discussion The results demonstrated the value of combining force and video data to provide a powerful visual description of how a paddler performs a trial (Figure 2.52) (Sperlich & Klauck, 1992). Through normalisation of sections of the course, it was possible to assess the strategies the paddler utilised to negotiate the course. A comparison of 6 runs from a male single canoe (C1M) paddler (Figure 6.1 to Figure 6.6) allowed for comparison of not only the time that the paddler passed through each gate, but also temporal information on when the paddle was placed in the water and on which side of the boat. In addition, the type of stroke executed and the amount of force (curve) produced were compared.

Comparison of the 6 runs demonstrated a number of interesting points. Firstly, the timing of specific strokes relative to events in the course suggested that the paddler implemented a specific strategy based on his location. That is, the paddler performed strokes and applied force based on where he was along the course. In this repeated trial scenario, the similarity of the trials could be observed, not only in time but also in the force curves and types of strokes which the paddler performed. Through normalisation of three sections of the course, it was evident that the strokes, timing of strokes and force required were dependent on the location of the paddler on the course. The greatest variation in stroke execution occurred when the paddler was manoeuvring the boat around technical gates requiring large changes in direction. Some of this variation may have occurred due to the unstable nature of the water as the paddler‟s blade moved through the water. For example, the draw-forward as the paddler entered gate 1 showed greater variation for the first half of the stroke compared to the second. This variation may have resulted from the paddle / paddler passing through the eddy line (an unstable section of water that defines the edge of the water circulating in the opposite direction to the main flow of water).

Clear trends were identified between stroke type and paddle force profiles. Forward, C, draw- forward, reverse sweep and brace strokes were all identifiable based on their paddle force profile included:. The consistency with which the paddler applied strokes during the various trials demonstrated that paddle force data could be utilised for stroke identification if appropriate algorithms were create to identify each of the unique characteristics for each type of stroke. Paddle and oar force profiles have been used in flatwater kayaking (Aitken & Neal, 1992; Sperlich, 1995b; Stothart et al., 1987) and rowing (Schneider et al., 1978; Wing & Woodburn, 1995) extensively to perform technique analysis. In these sports, it has been

140

identified that each paddler or rower has a signature force profile which remains consistent (Schneider et al., 1978; Wing & Woodburn, 1995). If a signature, profile exists in canoe slalom it may require adaptive algorithms to identify strokes as this may detract from the identifiable patterns that exist for each stroke.

Paddle force data demonstrated it may have greater potential for future stroke identification than video due to the additional detail which it can provide. That is, the reverse or reverse sweep strokes were consistently miscoded as a draw in the time-motion analysis, potentially due to the visual similarity in the movement of the paddle and boat during the draw and reverse sweep strokes. Thus, in this instance, the ability to detect when the front of the blade (positive forces) and the back of the blade (negative forces) were being used provides substantial detail which was not discernable from video.

The first trial differed substantially from the next 5 trials. This was likely a result of the paddler not having paddled the course prior to his first run. It is likely that the paddler altered his strategy based on his experiences during the first run. In the first run the paddler moved quicker through the first quarter of the second upstream gate but took longer to manoeuvre through the second quarter (Figure 6.1). This resulted in the paddler losing approximately 1 s on this gate. This error was observed in the time-motion analysis data but also in the timing of the force curves. A similar gate entry was found in Run 5 (Figure 6.5) with rapid movement through the first quarter, but the paddler sustained the force on the paddle and maintained the turn. Comparing these runs with the results of Chapter 5, which analysed the path of elite canoe slalom paddlers negotiating an upstream gate, it is clear that paddlers who cut the line between position one and position two (Figure 5.1), end up being pushed downstream and laterally across the gate, effectively increasing the line distance and therefore the time.

6.4 Conclusions The use of combined competition analysis and paddle force data provides valuable information on many aspects of canoe slalom performance. This study demonstrated that the strokes a paddler utilises to negotiate a course are linked to the resultant time; forces on the paddle during different strokes (forward, C, draw-forward, reverse sweep and brace) have distinct characteristics (shapes of curve) which make them uniquely identifiable; and the

141

timing of strokes and forces during different runs correspond for the same section of the course. In the future, this combination of video and force information may provide valuable descriptions of forces generated during particular strokes as well as provide tactical and repeatability information about a paddler.

142

CHAPTER SEVEN

7 EQUIPMENT SETUP – CANOE SLALOM PADDLE LENGTH COMPARISON

7.1 Introduction In canoe slalom, the ability to accelerate rapidly is important because of the stop-start nature of the sport as paddlers manoeuvre their boat around the course. As demonstrated in Chapter 6 the ability to apply paddle force effectively is critical to a paddler negotiating a complex gate sequence such as an upstream gate. Thus, improvements in acceleration from altering paddle configuration may be an important factor in determining overall race time. To test this, a case study was undertaken on flatwater with a K1M paddler of international standard using two identical paddles of different lengths. A series of 12, 12 m maximal sprints from a standing start were conducted as described in Chapter 2 Section 2.3.5. Force curves were measured as per Chapter 2 Section 2.2.2.

7.2 Results The time taken to complete each 12 m sprint showed that the trials with the short paddle (n = 6, 4.68 ± 0.16 s) were significantly (p = 0.012) faster than the trials with the long paddle (n = 6, 4.87 ± 0.13 s) (Figure 7.1). In particular, the first two strokes with the short paddle were significantly faster (p = 0.03 & p = 0.04 respectively) than the first two strokes with the long paddle (Figure 7.1). The most powerful indicator that the stroke time was faster with the shorter paddle was the comparison between the average stroke times for each trial for each condition. The shorter paddle (0.30 ± 0.013 s) had significantly (p = 0.003) faster stroke times compared to the long paddle (0.33 ± 0.009 s) (Figure 7.1).

The average trial velocity showed that the shorter paddle accelerated the boat significantly faster overall (p = 0.003) (Figure 7.2). Further analysis showed that the average starting velocities for the two paddle lengths were similar (ovals in Figure 7.2). However, after two seconds the short paddle was faster (rectangles in Figure 7.2).

143

The sum of the average impulse (area under the force curve) for the shorter paddle was 3% greater on the left and 5% greater on the right side compared with the longer paddle. Thus, the mean improvement in impulse was 4%.

0.8 Long Paddle p = 0.030 Short Paddle 0.7

0.6

0.5

p = 0.042 p = 0.039 0.4

Stroke Time Time Stroke (s) . 0.3

0.2

0.1

0 1 2 3 4 5 6 7 8 9 10 Ten Sequentual Strokes

Figure 7.1: Mean Stroke Time for each of the ten consecutive strokes from the 12 m sprints.

6 Short Long

3 Boat Velocity (m/s)BoatVelocity 2

0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 Time (s) Figure 7.2: Average Boat Velocity while using the short and long paddle. The green ovals represent the section of the trial where the average velocity generated by the two paddles was similar. The green rectangles highlight where the shorter paddle was faster.

144

7.3 Discussion The paddle with the shorter shaft length allowed the paddler to complete the 12 m sprints on average 4% faster (4.68 s) than when using the longer paddle (4.87 s). In a canoe slalom competition a 4% gain is equivalent to a 4 s reduction in the overall time for a run and certain improvement in the overall standing. In the 2005 canoe slalom world championships (Penrith, Australia) first place in C1M achieved a total time of 209.26 s with second and third +0.21 s (0.1%) and 1.38 s (0.7%) behind respectively, demonstrating the small difference between competitors at this level of competition. Any competitor in the top 7 could have won with a 4% (8.0 s) improvement in performance (International Canoe Federation, 2009b). The stroke times for the shorter paddle were significantly faster for the first two strokes as well as on average. Furthermore, the stroke times for the short and long paddles indicate that the paddler was able to change stroke rate more rapidly using the shorter paddle. The rotational inertia of the shorter paddle is less than that of the longer paddle, allowing the paddler to increase or decrease their stroke rate more rapidly (Sperlich & Klauck, 1992). In addition to the reduction in rotational inertia, by maintaining the distance between each hand and the tip of the blade while altering shaft length, the mechanical setup of the paddle was altered, in that the paddler‟s hands were 8 cm closer to each other on the shorter paddle (Oka et al., 1983; Ong et al., 2005). The net result is that the paddler‟s hands and arms travel through a smaller range of motion during the stroke and cane therefore achieve higher stroke frequencies if strong enough.

The times taken to complete the 12 m sprints were substantially faster with the short paddle even though the time spent with the blade in the water was significantly reduced. Comparison of the paddle force data indicates that the impulse the paddler produced from the shorter paddle was 4% higher than that of the longer paddle. This improvement in force mirrors the improvement in the split times for the 12 m sprint. The implication is that increased acceleration of the paddle in the water results in higher forces on the paddle that in turn results in faster boat speed, at least over 12 m.

145

7.4 Conclusions For the paddler involved in this study using a shorter paddle while maintaining the distance between each hand and the tip of the blade, allowed for greater acceleration and faster sprint times over 12 m with a K1 slalom boat on flat water compared with a longer paddle. This was achieved even though the stroke time (time with the blade in the water) was reduced. Interestingly mean impulse was proportional to the increase in boat speed and decreases in stroke time.

146

CHAPTER EIGHT

8 DISSEMINATION OF INFORMATION TO COACHES AND ATHLETES

8.1 Introduction Research suggests that quantitative and objective feedback have greater positive effects on performance than qualitative and subjective feedback (Franks et al., 1983; 1997). However, coach and athlete feedback are often not considered in the process of developing objective quantitative analysis systems (Hughes et al., 2007). Advances in technology have made it possible to augment and improve the feedback athletes receive during training and competition (Liebermann et al., 2002). However, much of this is based on video that takes time to analyse. This limits immediate feedback to the video images themselves and restricts prompt feedback related to simple kinematic and temporal data (Bartlett, 1999). Unfortunately, the timing of quantitative feedback is important in influencing the acquisition of skills and the enhancement of performance (Liebermann et al., 2002; Murray et al., 1997). Despite this, the turnaround time of the video based analysis offers large quantities of information which can be used prospectively in training and preparation for the next competition.

Providing timely and easily disseminated feedback played a major role in the development of CAP and the force curve analysis system as part of this thesis. This chapter provides an example of the kind of reports actually generated for coaches and athletes who participated in this research. The coach and athlete report is presented as if the reader were the athlete receiving feedback on their performance at a competition.

147

8.2 Example of a Race Report sent to the Coach and Athlete 8.2.1 Comparison of gate interval times The runs analysed in this report are: o Your final run – 103.16 s (2nd fastest run) o Your semi-final run – 106.10 s o Slalom Athlete 2 final run – 101.58 s (fastest run)1 o Slalom Athlete 3 final run – 103.99 s (3rd fastest run)1

The report will be discussed in terms of comparing your fastest run (final) to that of your semi-final run and Slalom Athlete (SA) 2 and SA 3 runs.

In your finals run you were faster to the first gate than SA 2 and SA 3. You also achieved the fastest time from gate 3 to 4 but lost time between gates 4 to 5 and 6 to 7. Between gates 7 to 17 you were faster than SA 3. Your time was quicker than SA 2 between gates 7 to 10 and 13 to 15, however he was slightly faster between gates 10 to 13. You lost a bit of time between gates 17 to 18, 19 to 20 and 21 to 22. Between gates 20 to 21, you were 0.6 s faster than SA 2, and SA 3 was less than 0.1s ahead of you (Figure 8.1 and Figure 8.2).

Min = minor avoidance Maj = major avoidance

Figure 8.1:Split times between each gate of the course for each of the Athletes in the key on the right hand side. Y-axis = Split time (s), X-axis = gate number (red = upstream gate), coloured text boxes on the graph represent avoidance at a gate (colour = athlete on right).

1 Usually chosen comparisons with the fastest paddler as well as paddlers just a bit faster and slower than the subject.

148

Slower = how much slower your final run was compared to other runs.

Faster = howm much faster your final run was compared to other runs.

Figure 8.2:Comparison of the split times taken to negotiate a gate relative to the subject (shown as green line in Figure 8.1). Gate descriptions as in Figure 8.1.

The upstream gate interval graph displays the time from the gate above the upstream gate to the gate below the upstream gate. The gate 4 upstream interval had similar times for both of your runs and was similar to SA 3. SA 2 was only slightly faster for this upstream gate split. You were faster than both SA 2 and SA 3 in both your semi-final and final runs at upstream gates 8 and 9. You were also slightly faster at upstream gate interval 16. For both the upstream gate 20 and 21 intervals your final run was slower than that of SA 3 but faster than that of SA 2, but your semi-final run time was longer on these gates compared to the other runs (Figure 8.3).

The upstream gate intervals indicate that you were particularly good at gates 8 and 9. At most upstream gates you were similar or slightly faster than SA 2. In comparison to SA 3, you were similar at gates 4 and 16, faster at 8 and 9 and he was faster at 20 and 21 (Figure 8.3).

149

Figure 8.3: Upstream Gate Interval. y-axis = the split time between the gate before and after each upstream gate, x-axis = upstream gate number, indicates upstream gate performance.

8.2.2 Stroke count Your final run, which was quicker, had a lower total number of strokes compared to your semi-final run. This was also apparent for your forward stroke count with more forward strokes executed in the semi-final run. The semi-final run also included two brace / forward strokes and an extra forward / steering stroke. In comparison to the other two paddlers, SA 2 used more strokes compared to your fastest run and SA 3‟s fastest run, but was similar to your semi-final run. SA 3 took less strokes i.e. lower total stroke count, less forward strokes and draws but more draw-forward strokes (Figure 8.4 and Figure 8.5).

Figure 8.4: Stroke count by gate – compares the number of strokes each athlete took between each gate on the course.

150

Figure 8.5: Stroke Count – The total number of each type of stroke taken by each athlete during the run. Each run / Athlete is represented by a colour.

8.2.3 Stroke duration The stroke duration indicates the average time taken to complete a stroke. The bars above and below the diamond on the graph indicate the standard deviation to display how much your stroke duration time varies in a run. These graphs (Figure 8.6) show your average stroke duration times for your semi-final and final run and that of SA 2 and SA 3. Only a few differences were noted here, such as SA 2 having longer stroke duration for his draw stroke; your semi-final run having a longer draw-draw stroke time; and your draw-forward stroke duration being shorter in your semi-final run compared to the other runs (Figure 8.6).

151

Figure 8.6: Stroke Duration for each athlete, y-axis = time in the water, x-axis = stroke type, separate plot for each athlete.

152

8.2.4 Race Profile The race profile (Figure 8.9, Figure 8.10 and Figure 8.11) is displayed for the entire run and then displayed in sections to allow more detail. These profiles are very useful for identifying: which boat or run reached a gate earlier; the difference in the number of strokes around a gate; sections of the course which require strokes on one side of the boat; and the pattern of strokes you used negotiated upstream gates.

The assessment of sections of the course which require the paddle to be on a specific side of the boat is most observable through a comparison of a right-handed C1M paddler and a left- handed C1M paddler performing the same course (Figure 8.7) because C1M paddlers use their onside and offside differently depending on their favoured side. Thus, comparison of two opposing paddlers reveals the specific sections of the course which required the paddler to perform strokes on a specific side of the boat regardless of which is their preferred side.

Figure 8.7: Comparison between a right handed C1M paddler and a left handed C1M paddlers for the entire race. Highlights the specific sections of the course which favour / require the right hand, those that favour the left hand and those which either could be used.

Upstream gates can be broken down into four lines and sections. The first yellow line before the upstream gate indicates the time when the middle of the boat crosses a parallel line directed out medially from the gates i.e. Number 1 on Figure 8.8: when the boat crosses this line (yellow) prior to entering the gate. The second line is defined as the time when the middle of the boat crosses a perpendicular line directed from the medial gate i.e. Number 2 on Figure 8.8: when the boat crosses this line (yellow) immediately prior to entering the gate. The line immediately following the gate is the time the middle of the boat crosses a perpendicular line directed from the medial gate following exit of the gate i.e. Number 3 on figure below. The fourth line indicates the time following the exit of the gate when the

153

middle of the boat crosses a parallel line directed out medially from the gates again i.e. Number 4 on figure below.

4 1 2

Figure 8.8: The split times taken either side of upstream gates. The positions one to four labelled in the figure divide the gate into four sections.

Breaking each upstream gate into these sections provides an indication of the type of upstream gate as well as the tactics of the paddler. The 1st and 2nd lines give information on the entry and the 2nd line to the gate indicates the time taken below the gate. The 3rd and 4th lines illustrate the type of exit and also display the time taken above the gate. It can also be used to see if more time was spent below or above the gate, if the time differs to that of the other runs or paddlers and if they appear to execute the upstream gate turns differently.

154

Figure 8.9: Full Race Profile – Time line of all the events coded for the entire race.

155

Figure 8.10: Section 1 and Section 2 Race Profiles – zoomed in section of full race profile for first two sections of the race. Start of each section is reset to the same gate so all athletes start equal.

156

Figure 8.11: Section 3 and Section 4 Race Profiles – zoomed in section of full race profile for first two sections of the race. Start of each section is reset to the same gate so all athletes start equal.

157

Using the race profiles it can be seen that you are quicker to the 1st gate but SA 2 is quicker to the 2nd gate. At gate 4 you are head to head with SA 2. Between gate 4 to 5 SA 2 is faster. You have similar times between gates 5 to 6, and SA 2 is quicker from gates 6 to 7. SA 2 gains the most time and gets ahead of you between gates 4 to 5 and 6 to 7. For the next section, gates 7 to 12, you were faster, as well as between gates 12 to 17. In the last section, gate 17 to finish, you and SA 2 had a similar time. This indicates that you were fastest over the middle two sections, but SA 2 was faster on the first section, in particular between gates 1-2, 4 to 5 and 6 to 7, gaining almost a 1.5 second lead from this section.

8.3 Conclusions This type of report provide a large amount of detailed information which, without the knowledge of the course layout, gate sequencing and manoeuvring requirements of the course can be difficult to interpret. However, the coach and athlete have an in-depth appreciation of these aspects of the course, plus the addition of video footage to augment the interpretation of this report. The report does not draw conclusions from the data as specific face to face feedback sessions with sports scientists, coaches and athletes were organised for this purpose. Athletes drew on their experiences from the course as well as the knowledge and expertise of the coaches and sport scientists to discover coaching points for use in their training and future racing.

The report highlights many aspects of a paddler‟s performance from stroke counts to complex race profiling and is useful in identifying deficiencies relative to other paddlers. For example, the split times may indicate that an athlete was slower for a specific section of the course, and from a simple count of strokes for the section, they can learn that a different strategy with respect to strokes used may have been more efficient. The additional information related to stroke sequencing (race timeline plots) and the tagging of strokes provides quantitative understanding on how the paddler was outperformed and strategies for training to improve performance in future competitions can be developed. The objectivity of the data allows coaches and athletes to identify problems which limit performance and to adopt new strategies through modification of training.

158

Analysis of C1 paddlers demonstrated where a paddler must have the paddle on a specific side of the boat and where it is not critical. This knowledge may assist inexperienced paddlers to plan a more efficient course and assist course designers in developing more evenly challenging courses.

Feedback from coaches (Table 8.1) and athletes suggested an appreciation of this kind of report and feedback sessions with the biomechanical staff on how to utilise the volume and depth of knowledge provided in the data sets (“I think the athletes are now much more aware of what you are doing and how it can have a positive impact on their training. The athletes are much more aware of how to read and interpret the information.”). Much still needs to be done to ensure that coaches and athletes get the maximum benefit from the data that biomechanical science provides to their coaches. In addition Mike Druce, the National Coach, suggested that “it would be great if we could develop a simple analysis for the C2 which incorporated the stroke count, stroke timing and gate timing.” This type of simplification of the analysis process would allow the possibly of in situ analysis and in competition feedback.

Finally, video and time-motion analysis cannot describe paddler effort. Chapter 6 demonstrated that combining paddle force data with time-motion analysis provides valuable descriptions of forces generated during particular strokes as well as repeatability information about a paddler. In addition, the trajectory data generated in Chapter 5 complement this broader picture of performance, but for this to be truly beneficial for canoe slalom, it is necessary to analyse entire runs so that it is possible to discriminate between the paths that different paddlers take.

159

Table 8.1: Email of support from Australian Institute of Sport (AIS) Head Coach for Canoe Slalom, Mike Druce after presentation and feedback session with athletes. The session aimed at educating them on how to interpret the information so it could be used in their training. Mice Druce [[email protected]] Sent: Mon 11/12/2006 1:02 PM To: Adam Hunter Cc: Myriam Fox

Hi Adam,

Thanks for coming up to do the presentation on Saturday. I think the athletes are now much more aware of what you are doing and how it can have a positive impact on their training. The athletes are much more aware of how to read and interpret the information.

Could you please send through a copy of your presentation to give to Myriam and the athletes who were unable to attend.

I think it would be great if we could develop a simple analysis for the C2 which incorporated the stroke count, stroke timing and gate timing.

Thanks again for the huge effort you and your team have put in.

All the best

Mike -- AIS Head Coach Canoe Slalom

160

CHAPTER NINE

9 CONCLUSIONS

The published scientific literature investigating athlete driven performance factors under competition conditions for canoe slalom is limited and does not address the topics discussed in this body of research. Time-motion and paddle force analysis methods specific to canoe slalom had to be developed (Chapter 2 Section 2.2) to analyse canoe slalom competition in a more complete manner. Existing motion analysis methods were also adapted for application under competition conditions (Chapter 2 Section 2.3.3). A set of slalom paddle stroke definitions as well as quantitative markers for determining split times on a slalom course were developed and intra-observer and inter-observer reliability tested (Chapter 2 Section 2.2.1).

The research undertaken in this thesis, through the application of these methods, highlight several key performance factors in canoe slalom competition (Chapter 3to 8). Firstly, minimising the distance between the paddler‟s head and the inside pole is an important consideration when negotiating an upstream gate. Secondly, the number of strokes taken on a course is linked to the time taken to complete the course with faster times associated with less strokes. Thirdly, this research identified to both coaches and athletes that much can be learnt by observing paddlers in other canoe slalom categories, especially if the strengths and weaknesses of other categories can be quantified, as done here. Fourthly, the greatest variability in performance occurs on technical sections of the course that are characterised by difficult water conditions and complex gate configurations, thus large reductions in total run time can be made by improving these skills. Finally, canoe slalom courses are designed as complex problem solving tasks; challenging paddlers to find the optimal solution for the fastest run. Furthermore, the changing environmental conditions (flow rate, obstacle locations, water quality and wind) typical of different slalom courses and events together with changing gate locations for each competition suggests that coaches and athletes must ensure that these coaching points are applied across a variety of real world slalom scenarios. For example, concentrating on the distance between the paddler‟s head and the inside pole during an upstream gate should be practised and coached across a number of different water conditions and gate configurations both right and left handed. Therefore the greater the variety of scenarios in which a paddler is coached, the greater their ability to adapt to each

161

different course and challenge. The main outcome of this research is a condensed list of key performance factors that play a major role in faster paddling times down a competition slalom course, thereby directing coaching and training programs. Feedback from Australian Institute of Sport slalom coaches suggests that this list is timely and useful in their coaching programs. Personal communication with younger paddlers who aspire to the national team suggest that they not only see this research as a vehicle towards improved performances, but are actively seeking ways to apply the outcomes to their own paddling and equipment setup.

A model of variables critical to performance at a canoe slalom competition was developed in Chapter 1 and is reproduced here with those variables addressed in this body of research clearly identified (Figure 9.1). The following sections present specific outcomes from this research as related to each of these variables:

9.1 Run Time, Path Taken and Mean Velocity In world championships and Olympic Games between 2004–2008, K1M was the fastest canoe slalom category and had the least variability in run times, followed by C1M, K1W and C2M. Over the past 50 years, C1M have demonstrated the greatest improvement in race time and are now achieving run times within 3% of K1M despite limitations associated with using a one bladed paddle (Chapter 3). Data from the 2005 World Championships, held in Penrith, Australia, showed that paddlers employ different strategies to negotiate the same course and the strategies used not only had an immediate effect but also influenced how subsequent gates were negotiated. For example, paddlers who performed a spin at gate 13 were able to set up the turn at the next gate earlier and without detriment to speed and therefore had a faster line into gate 15. Interestingly, the left handed C1M paddlers performed this manoeuvre, probably because it suited their favoured side (Chapter 4).

Different percentages of time were spent in each section (quarter) of an upstream gate (Chapter 4). Furthermore, upstream gates have different locations relative to the flow and no two gates resulted in the same combination of percentages. In addition, each canoe slalom category had a unique way of approaching the upstream gates. Data from the 2005 World Championships suggest that the majority of paddlers in the final run of each category spent 16%, 36%, 21% and 27% in each quarter of the turn. This implies that speeding up through the second (entrance to the gate) and final quarter of the turn (exit / reacceleration from the gate) would improve overall performance. Furthermore, the time taken to negotiate an

162

upstream gate was influenced by the path chosen as predicted by the average distance between the paddler‟s head and the inside pole (Chapter 5). Slower paddlers with longer path lines were pushed further downstream and laterally across the gate during the second quarter of the turn, which resulted in a slower gate completion times.

9.2 Strokes Used Analysis of the number and type of strokes used by the top ten paddlers in each category at the 2005 World Championships revealed that 67–71% were forward strokes (Chapter 4). Paddlers who took fewer strokes over the course were faster. Furthermore, their paddle was in the water for a longer time. C1M paddlers took a similar number of strokes on their onside as did K1M and K1W on their favoured side, but when considered across the course as a whole, C1M paddlers took significantly fewer strokes, in part because they take fewer strokes on their offside. Further analysis showed C1M paddlers per stroke spent a significantly longer time with the blade in the water on their offside compared with their onside. Both the reduction in strokes and the increased time for each stroke is most likely due to the awkward nature of paddling on their offside. Therefore, to achieve a similar time to a K1M paddler, C1M paddlers use their onside strokes more effectively as identified by Fabien Lefèvre (bronze K1M medallist 2004 Olympic Games) who said, “I studied the C1s a lot and saw they turned faster than we did. So I asked myself why and I tried to copy them.” (Endicott, 2006, pp. 1, Section VI).

C1M paddlers recorded a significantly greater number of forward and supporting strokes on their onside whereas both K1M and K1W paddlers preferred to brace on their right hand side (Chapter 4). This bilateral difference in C1M stroke execution provides a unique opportunity to investigate links between course design and strategies. Data from the 2005 World Championships showed that two C1M paddlers with opposing favoured sides utilised different strategies to negotiate some portions of the course and some gates, but there were gates and portions of the course where similar strategies were employed (Chapter 8). Whilst important for other C1M paddlers, this information identifies critical sections of the course where K1 paddlers can adopt the onside strategies and therefore achieve greater efficiency. This information can also be used by course designers to develop courses that challenge C1M paddlers regardless of their favoured side.

163

9.3 Technique and Errors Integrating CAP and paddle force data allows coaches and paddlers to quantify important components of the complex interaction between the path, strokes taken, strategies and technique used by paddlers to negotiate the course (Chapter 6). Analysis of a C1M paddler over a sequence of six gates showed that paddle effectiveness improved performance when the same path was taken to negotiate the course. Paddle effectiveness was quantified as both measured force on the blade and how the stroke was executed. In the slower run, the paddle was removed from the water whereas in the faster run, the paddler used a multistroke to maintain pressure on the blade for longer.

9.4 Experience and Variability The current investigation deals only with the elite population because this was the target area and the criterion on which all participants were selected. For sub elite and novice paddlers many differences may occur thus the findings from the current investigation cannot be extrapolated outside of the population without potential for error. Even within the elite population substantial variation in skill level was observed, but paddler variability (4-8%), as expressed as a percentile of total run time, was similar among elite paddlers. The upstream gate analysis showed that these trends also existed for a single manoeuvre. Analysis of the 2005 World Championships revealed that the most complex section of gates produced the greatest variation in split times between paddlers because they used numerous different strategies to negotiate the gates.

9.5 Paddle Setup In canoe slalom the ability to accelerate rapidly allows for greater manoeuvrability when negotiating gates and setting up a path through the turbulent water. A paddle shaft length was tested and the shorter paddle resulted in greater acceleration, greater mean impulse and faster sprint times even though the stroke time (time with the blade in the water) was reduced (Chapter 7).

164

Canoe Slalom Competition Result 

Competition

Result from Heat One Result from Result from Semi Final Final  Result from Heat Two

Paddler dependent variables and attributes Performance outcomes analysed in this Thesis analysed in this Thesis

Each Run Run Time  Penalties

Judging Path Taken  Mean Velocity  Errors  Errors

Course Design Paddler Attributes Wind Appeal

Flow Rate Gates Obstacles Strokes Used  Technique  Judge Experience Variability  Experience Downstream Upstream Gates Gates Physiology Psychology

Figure 9.1: The model of the critical variables in a canoe slalom competition as presented in Chapter 1, highlighting the variables which were the focus of this research and the topics presented for discussion in this conclusion.

9.6 Improving the analysis process The CAP for canoe slalom was functional as an analysis tool, however, it suffers from a number of limitations with respect to graphical display and video manipulation. The software would benefit from the following modifications: Improved video control to allow new compression methods (updates from Windows Media Player) and the ability to read in fields (50 Hz) rather than frames (25 Hz) to improve the accuracy of strokes entry and exit. In addition an ability to capture footage directly into the program would allow real-time hand analysis. However, real-

165

time hand analysis requires a simplified definition set for effective live coding and immediate feedback. This improvement should be discussed in conjunction with coaches and athletes to ensure that the simpler definition set captures key aspects of the stroke and run. Improved synchronisation methods to allow multiple videos to be viewed at the same time. It would be useful if the operator could simply drag the location of one video relative to the other in a timeline or step one video while keeping the other stationary. This would allow the operator to more easily ascertain whether the videos are synchronised by stepping or playing through both videos at the same time. Improved operator control to create their own coding buttons and save these in a file which can be altered if the definition set changes. This would increase the flexibility and longevity of the system as it becomes adaptable for different sporting applications. This would also reduce the time requirement of the designer when changes are made to the definition set. Improved feedback section of the program so that all the outputs are built in and the report is generated rapidly in one step as images.

Changes such as in situ coding with a simplified model and rapid reporting would allow the system to be adapted for rapid feedback at competition and training. In addition, feedback on strategies used by other paddlers that resulted in fast split times could be relayed to an athlete prior to their first run. In a sport where no practice runs are allowed, such information would be invaluable for technical and tactical decisions on which path to take and how best to negotiate each gate. The more detailed stroke information could be entered into the software post event.

The original definition set and method was effective in gathering detailed data on canoe slalom competitions. It is possible that a simplification of the definition set and method that provides only a count of different strokes types without the time consuming marking of 200 in and out points would provide sufficient information to coaches and athletes as a feedback tool. In addition, simplification of the stroke classification, such as identifying strokes as: propulsive (forward), turning (draw, c, sweep, reverse sweep, side draw), reverse propulsive (reverse) and neutral (tap, brace) may further reduce analysis time. However, the depth of the analysis depends on the information requested by the coach or athlete. One of the major

166

benefits of this custom developed software is that it is possible to change the information collected and presented based on the individual requirements of one coach or athlete.

Video-based analysis provides one major advantage over other analysis methods for competition in that it can be used for immediate, qualitative athlete feedback. However, to extract quantitative information processes such as competition analysis or three-dimensional analysis must be applied. These processes are time consuming due to their high operator requirement, thus more automated procedures would improve this process. Currently there are modern systems being used by many professional soccer and rugby union teams to automatically track the movements of all players, the ball and officials (Di Salvo et al., 2007; Di Salvo et al., 2006). Eight stable, synchronised cameras and optical sensors are positioned around the top of the stadium in a way that every area of the pitch is covered by at least two cameras for accuracy, occlusion, resolution and resilience (Di Salvo et al., 2007; Di Salvo et al., 2006). The processing and delivery of the data from these systems is completed overnight. This software could be used in canoe slalom to provide accurate positional information, velocities, movement patterns and distances travelled by paddlers for an entire run. The cost and non-portability of these systems at present would be prohibitive for a comprehensive analysis but with the ever reducing size and cost of these new technologies it may be possible to use them during competition in the future. While application of this technology to slalom could be beneficial, the moving water may pose an issue with optical sensors.

The competition analysis program for canoe slalom could be complemented and possibly eventually replaced through the use of a paddle force system. This was highlighted in a case study of an athlete performing repeated runs of a simulated course with a paddle force system (Chapter 6). The findings from Chapter 6 suggest that there is a possibility of future algorithms to identify strokes for canoe slalom. The first step in this process would be to utilise data gathered through CAP and the associated definition set to group force curves into stroke types. The defining features of each stroke type could then be identified in the paddle force curve data and algorithms developed to identify these defining features. However, to replace CAP with the paddle force system for within-competition analysis would depend on the following: 1) the competition rules would need to allow such a device; 2) the athlete would need to be willing to paddle with it during competition. If allowed in competition paddle force data could then be combined with split times which could be collected in situ

167

using current strategies from video or an advanced timing system similar to those used in cycling or triathlon where athletes are registered at each gate based on a unique identification tag which they carry. If such systems are not allowed in competition then they would still provide benefits to coaches and athletes in the training environment.

9.7 Future Research Directions This body of research is mainly based on competition analysis of the 2005 World Championships. The work needs to be extended across multiple competitions and venues to provide detail about how paddlers adapt to new environments. Further information on how C1M paddlers improve their efficiency is needed following the outcomes of this research. In addition, it would be interesting to quantify the effect that the list of key coaching elements identified by this body of research has on sub-elite and novice paddlers. Intervention studies assessing the impact of different paddle setups (paddle length and blade shape) on whitewater performance is a natural extension of the flatwater case study presented in this thesis. Finally, combining paddle force data with time-motion analysis provides valuable descriptions of forces generated during particular strokes as well as repeatability information about a paddler. Thus, development of algorithms to detect and categorise paddle force data into stroke types for canoe slalom would provide substantial feedback for coaches and athletes. In addition, the trajectory data complement this broader picture of performance but for this to be truly beneficial for canoe slalom, it is necessary to analyse entire runs so that it is possible to discriminate between the paths that different paddlers take.

168

LIST OF REFERENCES

Abdel-Aziz, Y. I., & Karara, H. M. (1971). Direct linear transformation from comparator coordinates into object space coordinates in close-range photogrammetry. In Symposium on Close-Range Photogrammetry (pp. 1-18). Falls Church, VA: American Society of Photogrammetry. Aitken, D. A., & Neal, R. J. (1992). An On-Water Analysis System for Quantifying Stroke Force Characteristics During Kayak Events. International Journal of Sport Biomechanics, 8(2). Ariel, G. (2002). Transformation Module Manual: Ariel Dynamics, Inc. Australian Canoeing. (1999). Sprint canoeing factsheet. Retrieved 20/07/2001, from www.ausport.gov.au/canoe/sprint/factsheet.htm Australian Canoeing. (2004). Flatwater racing manual. Retrieved 27/01/2005, from www.canoe.org.au/site/canoeing/ac/downloads/coaching/FW-lvl1/FW-lvl1-ref.pdf Australian Canoeing. (2005). 2005 ICF canoe slalom World Championships. Retrieved 19/7/2008, 2008, from http://www.canoeslalom.com.au/?MenuID=Results%2F13043%2F0%2F Australian Canoeing. (2009). Saftey Guidelines: Minimum standards for the conduct of safe non-competitive canoeing and kayaking activities. Retrieved 27/07/2009. from www.canoe.org.au/site/canoeing/ac/downloads/Safety-Guidelines.pdf. Baker, J. (2008). AIS paddle force protocol. Australian Institute of Sport. Bartlett, R. (1999). Sports biomechanics: reducing injury and improving performance. London ; New York; United Kingdom: E & FN Spon. Bartlett, R. (2001). Performance analysis: can bringing together biomechanics and notational analysis benefit coaches? International Journal of Performance Analysis in Sport, 1(1), 122-126. Bloomfield, J., Jonsson, G. K., Polman, R., Houlahan, K., & O'Donoghue, P. (2005). Temporal pattern analysis and its application in soccer. In L. Anolli, G. Riva, S. D. Jr. & M. S. Magnusson (Eds.), The hidden structure of interaction: From neurons to culture patterns (pp. 237–250). Amsterdam: IOS Press. Canoe Kayak Canada. (2007). Canadian Whitewater News - February4 2007. Ottawa: Canoe Kayak Canada.

169

Cox, R. H. (2006). Sport Psychology: Concepts and Applications (6 ed.). New York: McGraw-Hill. Davids, K., Lees, A., & Burwitz, L. (2000). Understanding and measuring coordination and control in kicking skills in soccer: implications for talent identification and skill acquisition. Journal of Sports Sciences, 18(9), 703-714. Davies, K. (2003). Utilising digital video for game, player, training and opposition analysis. In Communication to the International Conference on the Science and Practice of Rugby (pp. 79). Brisbane, Australia. Deakin, J. M., & Allard, F. (1991). Skilled memory in expert figure skaters. Memory & Cognition, 19(1), 79-86. Di Salvo, V., Baron, R., Tschan, H., Calderon Montero, F. J., Bachl, N., & Pigozzi, F. (2007). Performance characteristics according to playing position in elite soccer. International Journal of Sports Medicine, 28, 222-227. Di Salvo, V., Collins, A., McNeill, B., & Cardinale, M. (2006). Validation of Prozone: A new video-based performance analysis system. International Journal of Performance Analysis in Sport, 6(1), 108-119. Downey, J. (1977). Attempt to construct a theoretical framework for the game in order to provide clear guidelines for the player and the coach. In In, Selected papers presented at the International Coaching Conference, Malmo, Sweden, May 3-7, 12 p. Endicott, W. T. (2006). Slalom E-Book. Retrieved 28/08/2009, from www.canoeicf.com. Fiore, D. C., & Houston, J. D. (2001). Injuries in whitewater kayaking. British Journal of Sports Medicine, 35(4), 235-241. Flynn, R. (1998). Are there statistically crucial tactical moments during squash matchplay? Australian Squash Coach, Spring, 18–21. Franks, I. M., Goodman, D., & Miller, G. (1983). Analysis of performance: qualitative or quantitative. Sports: Science Periodical on Research & Technology in Sport, 1-8. Gulbrandsen, O. (1990). Development of Outrigger Canoes in Sri Lanka. Retrieved. from. Heath, J. D., & Arima, E. Y. (2004). Eastern Arctic Kayaks: History, Design, Technique (pp. 172). Fairbanks: University of Alaska Press. Ho, W. H., Shiang, T. Y., & Chung, P. H. (2000). The principle of a canoe ergometer and biomechanic analysis of the flatwater canoe stroke. In In Hong, Y. (ed.), Proceedings of XVIII International symposium on biomechanics in sports, Hong Kong, Department of Sports Science and Physical Education. The Chinese University of Hong Kong, c2000, p.654-657. Hong Kong.

170

Hughes, M. D. (1985). A comparison of the patterns of play of squash. In I. D. Brown, R. Goldsmith, K. Coombes & M. A. Sinclair (Eds.), International Ergonomics '85 (pp. 139-141). London: Taylor & Francis. Hughes, M. D. (1995). Computerised notation of racket sports. In T. Reilly, et al. (Ed.), Science and racket sports (pp. 249-256). United Kingdom: London, E & FN Spon. Hughes, M. D., & Bartlett, R. (2002). The use of performance indicators in performance analysis. Journal of Sports Sciences, 20, 739–754. Hughes, M. D., & Franks, I. M. (1997). Notational Analysis of Sport. Melbourne: Taylor & Francis. Hughes, M. D., Hughes, M. T., & Behan, H. (2007). The Evolution of Computerised Notational Analysis Through the Example of Racket Sports. Sports Science and Engineering, 1(1), 3-28. Hunter, A. (2009). Canoe Slalom Boat Trajectory while Negotiating an Upstream Gate. Sports Biomechanics, 8(2), 105-113. Hunter, A., Cochrane, J., & Sachlikidis, A. (2007). Canoe slalom competition – reliability analysis. Sports Biomechanics, 6(2), 153-167. Hunter, A., Cochrane, J., & Sachlikidis, A. (2008). Canoe Slalom Competition Analysis. Sports Biomechanics, 7(1), 24-37. Ille, A., & Cadopi, M. (1999). Memory for movement sequences in gymnastics: Effects of age and skill level. Journal of Motor Behavior, 31(3), 290-300. International Canoe Federation. (2006, 22/8/2006). Slalom racing history. Retrieved 1/4/2008, 2008, from http://www.canoeicf.com/default.asp?Page=1714&MenuID=Slalom%2Fc1021%2F16 17%2CCommittee%2F1051%2F0 International Canoe Federation. (2009a). Slalom racing competition rules. Retrieved 01/03/2009, from http://live.canoeicf.com/dms/icf/documents/Disciplines/CSL/Rules- for-Canoe-Slalom-2009/Rules%20for%20Canoe%20Slalom%202009.pdf International Canoe Federation. (2009b). Slalom results. Retrieved 01/03/2009, 2009, from www.canoeicf.com/default.asp?MenuID=Results/1107/0,Slalom/1284/0 Jelen, K., & Jandová, S. (1999). Results of biomechanical analysis in snowbiking. Gymnica, 29, 43-47. Jelen, K., & Pribramsky, M. (1987). Biomechanics of slalom turns. Teorie a Praxe Telesne Vychovy, 35(10), 629-632.

171

Kamber, R. (Ed.). (2008). International Canoe Federation - Medal Winners. Dornach Switzerland: nauticMEDIA. Krosshaug, T., Andersen, T. E., Olsen, O. E. O., Myklebust, G., & Bahr, R. (2005). Research approaches to describe the mechanisms of injuries in sport: Limitations and possibilities. British Journal of Sports Medicine, 39, 330–339. Liebermann, D. G., Katz, L., Hughes, M. D., Bartlett, R. M., McClements, J., & Franks, I. M. (2002). Advances in the application of information technology to sport performance. Journal of Sports Sciences, 20(10), 755-769. Lyons, K. (1997). Lloyd Howell Messersmith: pioneer of notational analysis of performance in sport. In M. D. Hughes (Ed.), Notational Analysis of Sport I & II (pp. 49-57). Cardiff, UK: University of Wales Institute. Lyons, K. (2002). Performance analysis for coaches: An introduction. Sports Coach, 25(3), 6- 8. Males, J. R., & Kerr, J. H. (1996). Stress, emotion, and performance in elite slalom canoeists. / Stress, emotion et performance pour l'elite des canoeistes de slalom. Sport Psychologist, 10(1), 17-36. Males, J. R., Kerr, J. H., & Gerkovich, M. M. (1998). Metamotivational states during canoe slalom competition: a qualitative analysis using reversal theory. Journal of Applied Sport Psychology, 10(2), 185-200. Marhold, G., & Herrmann, H. (1981). Some possibilities and problems in collecting data for the biomechanical analysis of sport techniques in flatwater canoe events. Paper presented at the Biomechanics VIII-A&B: Procedings of the 8th International Congress of Biomechanics, Nagoya, Japan, Champaign, III. McKenna, M. J., Patrick, J. D., Sandstrom, E. R., & Chennells, M. H. D. (1988). Computer- video analysis of activity patterns in Australian Rules Football. In T. Reilly, A. Lees, K. Davies & W. Murphy (Eds.), Science and football (pp. 274-281). London: E & FN Spon. Messersmith, L. L., & Corey, S. M. (1931). Distance traversed by a Basketball player. Research Quarterly, 2, 57-60. Miao, W., & Bi, K. Q. (2001). Umpire system of canoe slalom wild water games over PC control. Journal of Wuhan Institute of Physical Education, 35(5), 25-27. Murray, S., Maylor, D., & Hughes, M. (1997). The effect of computerised analysis as feedback on the performance of elite squash players. In A. Lees, M. Hughes, T. Reilly

172

& I. Maynard (Eds.), Science and Racket Sports II (pp. 235-240). London: E. & F. N. Spon. Norton, K., & Olds, T. (1996). Anthropometrica: a textbook of body measurement for sports and health courses. Sydney; Australia: University of NSW Press. Oka, H., Okamoto, T., Nakagawa, H., Yamashita, N., Kumamoto, M., Yoshizawa, M., et al. (1983). The dynamic features of the Canadian canoe paddle. Paper presented at the Eight International Congress of Biomechanics, Nagoya, Japan, Champaign. Ong, K. B., Ackland, T. R., Hume, P. A., Ridge, B., Broad, E., & Kerr, D. A. (2005). Equipment Set-up among Olympic Sprint and Slalom Kayak Paddlers. Sports Biomechanics, 4(1), 47-58. Pearce, A. (2005). Video analysis: Where do I start? Sports Coach, 28(2), 34-36. Pendergast, D., Mollendorf, J., Zampero, P., Termin, A., Bushnell, D., & Paschke, D. (2005). The influence of drag on human locomotion in water. Undersea and Hyperbaric Medicine Journal, 32(1), 45-57. Peters, J. R. (1987). Design and construction of an artificial canoe slalom course at Holme Pierrepont, Nottingham. Municipal Engineer, 4(6), 317-330. Rademaker, S. (1977). Flatwater kayaking: Paddle principles. Canoe, 5, 32-39. Randers, M. B., Mujika, I., Hewitt, A., Santisteban, J., Bischoff, R., Solano, R., et al. (2010). Application of four different football match analysis systems: A comparative study. Journal of Sports Sciences, 28(2), 171-182. Ridge, B. R., Broad, E., Kerr, D. A., & Ackland, T. R. (2007). Morphological characteristics of Olympic slalom canoe and kayak paddlers. European Journal of Sport Science, 7(2), 107-113. Ritter, A. (1975). Point tables for eliminary rounds and scoring rounds in canoe slalom. Leistungssport, 5(4), 308-312. Salmon, S. (1998). How reliable is your GPS? Crusing World, 24(2), 59. Sanderson, F. H., & Way, K. I. M. (1979). The development of objective methods of game analysis in squash rackets. British Journal of Sports Medicine, 11, 188. Schmidt, K. H. (1993). The development of a computer-aided video analysis for the objective organization of the competition distance in canoe slalom. Leipziger Sportwissenschaftliche Beitraege, 34(2), 102-114. Schneider, E., Angst, F., & Brand, J. D. (1978). Biomechanics of Rowing. Paper presented at the Sixth Internation Symposium on Biomechanics.

173

Schokman, P., Le Rossignol, P. F., & Sparrow, W. A. (2002). Validity and reliability of a voice-recognition game analysis system for field sports. Journal of Science and Medicine in Sport, 5, 362-371. Shapiro, R. (1978). Direct linear transformation method for three-dimensional cinematography. Research Quarterly, 49(2), 197-205. Sidney, K., & Shephard, R. J. (1973). Physiological characteristics and performance of the white-water paddler. European Journal of Applied Physiology & Occupational Physiology, 32(1), 55-70. Smith, D., & Bar-Eli, M. (2007). Essential readings in sport and exercise psychology. Champaign: Human Kinetics. Spencer, M., Lawrence, S., Rechichi, C., Bishop, D., Dawson, B., & Goodman, C. (2004). Time-motion analysis of elite field-hockey, with special reference to repeated-sprint activity. Journal of Sports Sciences, 22, 843–850. Sperlich, J. (1995a). Biomechanical considerations on possible performance improvements through boat and paddle in canoe sport. Unpublished Translation: Australian Institute of Sport, 1-10. Sperlich, J. (1995b). Information for trainers and athletes on open water measurement in canoe sports. Unpublished Translation: Australian Institute of Sport, 1-4. Sperlich, J., & Klauck, J. (1992). Biomechanics of canoe slalom: measuring techniques and diagnostic possibilities. In R. Rodano (Ed.), Proceedings of the 10th Symposium of the International Society of Biomechanics in Sports (pp. 82-84). Milan – Italy: Edi-Ermes. Starkes, J. L., Caicco, M., Boutilier, C., & Sevsek, B. (1990). Motor Recall of Experts for Structured and Unstructured Sequences in Creative Modern Dance. Journal of Sport & Exercise Psychology, 12(3), 317-321. Stothart, J. P., Reardon, F. D., & Thoden, J. S. (1987). A system for the evaluation of on- water stroke force development during canoe and kayak events. In In, Terauds, J. (ed.) et al, Biomechanics in sports III & IV : proceedings of ISBS, Del Mar, Calif., Academic Publishers, 1987, p. 146-152. United States. Vaccaro, P., Gray, P. R., Clarke, D. H., & Morris, A. F. (1984). Physiological characteristics of world class white-water slalom paddlers. Research Quarterly for Exercise & Sport, 55(2), 206-210. Vaillancourt, H. (1984). Building an Algonquin Birchbark Canoe. Retrieved 20/07/2009, 2009, from www.birchbarkcanoe.net/

174

White, A., & Hardy, L. (1998). An in-depth analysis of the uses of imagery by high-level slalom canoeists and artistic gymnasts. Sport Psychologist, 12(4), 387-403. Wing, A. M., & Woodburn, C. (1995). The coordination and consistency of rowers in a racing eight. Journal of Sports Sciences, 13, 187-197. Witte, T. H., & Wilson, A. M. (2004). Accuracy of non-differential GPS for the determination of speed over ground. Journal of Biomechanics, 37, 1891-1898. Witte, T. H., & Wilson, A. M. (2005). Accuracy of WAAS-enabled GPS for the determination of position and speed over ground. Journal of Biomechanics, 38, 1717- 1722. Zamparo, P., Tomadini, S., Didone, F., Grazzina, F., Rejc, E., & Capelli, C. (2006). Bioenergetics of a Slalom Kayak (K1) Competition. International Journal of Sports Medicine, 27(7), 546-552. Zumwechick, J. (1997). Encyclopedia of sport science. New York: Simon and Schuster- Macmillan.

175

APPENDIX A INFORMED CONSENT – TESTING PROCEDURES FOR CANOE SLALOM SIMULATED COMPETITION

176

PARTICIPANT INFORMATION

Biomechanical and Competition Analysis of Canoe Slalom 1

Researchers Mr Adam Hunter (Hons), Doctor of Philosophy in School of Health Science at the University of Canberra Ms Jodie Cochrane (Hons), Biomechanics Department of Australian Institute of Sport

Supervisors Dr Margi Bohm (PhD), School of Resource, Environmental and Heritage Science Dr Keith Lyons (PhD), Performance Analysis Department, Australian Institute of Sport

Funding / Support This project is funded and supported by the Australian Institute of Sport and the University of Canberra.

Project Aim Currently there is very little published on biomechanical testing in canoe slalom. Therefore the aim of this work will be to obtain biomechanical data of the top athletes in a simulated competition environment and identify biomechanical factors related to good performance. This will enable comparisons for future biomechanical testing in training and simulated competitions.

Benefits of the Project These simulated competition data will be processed and statistically analysed to identify variables and technique factors that are related to performance. These data will form a database of leading athletes to be used for future testing and comparisons. Currently there is very little data available on canoe slalom athletes, and there is only limited data on force testing of Australian and international paddlers. Using the appropriate protocol, a biomechanical profile of elite performers will be developed to determine relationships between biomechanical measures and performance. As a participant you will receive a research report which contains information about your paddle forces, boat orientation and position, accurate splits and average velocity information along with video footage from each trial. As a participant the report will be sent directly to you.

General Outline of the Project Elite slalom paddlers from men’s K1 (single kayak) and Elite men’s C1 (single canoe) classes will be tested during this research project. Each session will involve the two paddlers 1 performing trials alternately over a ¼- /3 length slalom course (approximately 30 seconds). One paddler will return to the start and rest while the other paddler performs the trial. An expected trial turnaround of 3 minutes allows 6 minutes for each paddler to perform their trial and the return to the start. This will be repeated as many times as possible in a 30 min period without causing fatigue on the paddlers (maximum 5-10 trials per paddler). The nature of the trials is similar to the type of activity you would be involved in during training but about ½ the volume.

Postal Address: University of Canberra ACT 2601 Australia Location: Kirinari Street Bruce ACT Telephone: +61 (0)2 6201 5111 Facsimile: +61 (0)2 6201 5999 World Wide Web: http://www.canberra.edu.au

PARTICIPANT INFORMATION

Biomechanical and Competition Analysis of Canoe Slalom

As a participant you will have your paddle instrumented with non-permanent force transducers which are fastened to the shaft using clamps fittings. In addition to the force transducers you will have two boxes placed in your boat, one which records the information from the force transducers on your paddle and one which records information about the position and orientation of your boat. Once these testing devices are fitted to your equipment and calibrated testing can begin.

During each trial your split times will be recorded via an electronic timing system. Your performance will also be recorded on two cameras to provide a record of their overall performance for competition analysis and stroke identification. At two gates your performance will be recorded using 3-4 high speed or 50 Hz cameras each allowing 3D data to be obtained. The information from the force transducers and the boat orientation / position sensors will also be collected during each trial. These data will be correlated to determine if any relationships which exist between the variables which can explain a faster time. Results will be presented in a form where no single paddler will be identifiable

Participant Involvement As a participant you will be asked to perform between 5 and 10 trials over a set white water course during a 30 min period. Each session will involve the two paddlers performing trials 1 alternately over the /3 length course (approximately 30 seconds). One paddler will return to the start and rest while the other paddler performs the trial. Expected trial turn around of 3 min allows 6 min for each paddler to perform their trial and the return to the start. Additional time either side of test will be required for researchers the setup and calibrate testing equipment on your paddle and boat, but as a subject you need not be present during these periods. Estimated time for equipment setup and calibration is expected at two hrs before and 30 min after.

Discomfort / Risks Potential risks to the paddler during the collection of this data are no greater risk than those experienced during a normal training session. Considering the experience of the potential participants the risks associated with this testing are even smaller. In the event of something occurring there are always Lifeguards from the Penrith White Water Stadium on patrol, during any period where the water is running.

Confidentiality / Anonymity To protect the privacy and anonymity of the participants the identity (names) of all paddlers will be removed from all the data and replaced with a number system (Participant 1) which allows only the researchers to identify the paddlers.

Data Storage To maintain security of data and participant identity information regarding the study will be stored in a computer with a password or in a locked filing cabinet. All data will be stored at the AIS under lock and key for a period of five years after completion.

178

PARTICIPANT INFORMATION

Biomechanical and Competition Analysis of Canoe Slalom

Ethics Committee Clearance This project has been approved by the Committee for Ethics in Human Research of the University of Canberra and the Ethics Committee from the Australian Institute of Sport.

Queries and Concerns If you have any further queries and concerns about the research then contact Adam Hunter, Jodie Cochrane or Margi Bohm: Adam Hunter Jodie Cochrane Margi Bohm (02) 6214 7914 (02) 6214 1558 (02) 6201 2058 (02) 6214 1593 (02) 6214 1593 (02) 6201 2328 [email protected] [email protected] [email protected]

If you have any concerns with respect to the conduct of this study, you may contact the Secretary of the AIS Ethics Committee (Mr John Williams) on (02) 6214 1816.

179

INFORMED CONSENT

Biomechanical and Competition Analysis of Canoe Slalom

Principal Researchers: Adam Hunter & Jodie Cochrane

This is to certify that I, ______hereby agree to participate as a volunteer in a scientific investigation as an authorised part of the research program of the Australian Sports Commission under the supervision of Adam Hunter & Jodie Cochrane.

The investigation and my part in the investigation have been defined and fully explained to me by Adam Hunter & Jodie Cochrane and I understand the explanation. A copy of the procedures of this investigation and a description of any risks and discomforts has been provided to me and has been discussed in detail with me.

I have been given an opportunity to ask whatever questions I may have had and all such questions and inquiries have been answered to my satisfaction. I understand that I am free to deny any answers to specific items or questions in interviews or questionnaires. I understand that I am free to withdraw consent and to discontinue participation in the project or activity at any time. I understand that any data or answers to questions will remain confidential with regard to my identity. I certify to the best of my knowledge and belief, I have no physical or mental illness or weakness that would increase the risk to me of participating in this investigation. I am participating in this project of my own free will and I have not been coerced in any way to participate. I understand that I will receive a report containing results from testing. I understand that this data cannot be duplicated. I understand that this data can only be published, reported or presented by the named researchers or supervisors on this project: Adam Hunter, Jodie Cochrane, Margi Bohm and Keith Lyons.

Signature of Subject: ______Date: ___/___/___

A summary of the research report can be forwarded to you when published. If you would like to receive a copy of the report, please include your mailing address below. Name______Address______

I, the undersigned, was present when the study was explained to the subject/s in detail and to the best of my knowledge and belief it was understood.

Signature of Researcher: ______Date: ___/___/___

180

APPENDIX B PRINCIPAL COMPONENT FACTOR ANALYSIS FOR VARIABLES RELATING TO UPSTREAM GATE PERFORMANCE

181

B PRINCIPAL COMPONENT FACTOR ANALYSIS FOR VARIABLES RELATING TO UPSTREAM GATE PERFORMANCE

B.1 Introduction This section presents the methods results and discussion behind the selection of variables used in the upstream gate analysis. To select the variables for analysis in chapter 3 a principal component factor analysis was performed. The data for this section was the data as described in chapter 3 section 3.4. The aim of this analysis was to determine which variable were most related to upstream gate performance.

B.2 Methods A principal component factor analysis with an inclusion factor criteria of Eigenvalues > 1 was performed on the mean data for each participant, including all variables identified as potentially linked to performance. These included the mean, standard deviation (SD) and coefficient of variation for the following variables between points one and four: total time, average velocity, SD velocity, minimum velocity, time to minimum velocity, maximum boat pitch, average boat pitch, SD boat pitch, time spent near minimum velocity, average boat roll, maximum boat roll, average distance between the paddler‟s head and the inside pole, SD distance between the paddler‟s head and the inside pole, maximum turn rate and average turn rate. In addition the mean, SD and coefficient of variation for the following variables at points one, two, three and four were included in the model: velocity, distance between the paddler‟s head and the inside pole, heading, angle and split time. The variable of interest in the overall scheme of the race was total time therefore a correlation matrix was also setup to determine which variables correlated. Selection of variable was to include single point and range variables.

B.3 Results Six components were returned with a Eigenvalues of greater than one explaining 90% of the total variance within the variables (Figure B.0.1 and Table B.1). However there were a large number of components which contributed very little, indicating there was substantial overlap in the variables entered. The last three components added 20% to the model compared to the first three which explained 70.07% of the variability observed in the data from all variables

182

(Table B.1). Thus these were analysed in greater detail to determine what was related to each component. The percentage of the variance which each component was able to explain was: component one 33.30%, component two 25.58% and component three 13.19% (Table B.1). The correlation matrix between each variable and the three components (Table B.2) revealed that component one was most highly correlated with total time (r = 0.94), mean distance between head and inside pole (r = 0.93) and split between point two and point three (r = 0.91). Component two was most highly correlated with the split between points three and four (negative correlation) (r = -0.85) the distance between head and pole at point two (r = 0.80). However, this component appears to have very little impact on total time (r = 0.02). Component three was most closely related to heading at point two (r = 0.76) and maximum turn rate (r = 0.70) but weakly negatively correlated with total time (r = -0.27). Scree Plot

Eigenvalue 4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Component Number

Figure B.0.1: Scree Plot presenting the Eigenvalue for each component from the principal component factor analysis.

183

Table B.1: Total Variance Explained by each component from the principal component factor analysis.

Initial Eigenvalues Extraction Sums of Squared Loadings % of % of Component Total Variance Cumulative % Total Variance Cumulative % 1 11.987 33.296 33.296 11.987 33.296 33.296 2 8.490 23.582 56.879 8.490 23.582 56.879 3 4.750 13.194 70.073 4.750 13.194 70.073 4 3.537 9.826 79.898 3.537 9.826 79.898 5 2.010 5.583 85.482 2.010 5.583 85.482 6 1.637 4.546 90.028 1.637 4.546 90.028 7 .965 2.680 92.708 8 .710 1.972 94.680 9 .500 1.390 96.070 10 .414 1.149 97.219 11 .328 .912 98.131 12 .280 .778 98.910 13 .173 .479 99.389 14 .105 .292 99.681 15 .061 .170 99.851 16 .054 .149 100.000 17 5.18E-016 1.44E-015 100.000 18 4.94E-016 1.37E-015 100.000 19 4.56E-016 1.27E-015 100.000 20 3.81E-016 1.06E-015 100.000 21 2.92E-016 8.10E-016 100.000 22 2.43E-016 6.74E-016 100.000 23 2.11E-016 5.86E-016 100.000 24 1.75E-016 4.87E-016 100.000 25 1.07E-016 2.97E-016 100.000 26 4.44E-017 1.23E-016 100.000 27 1.38E-017 3.84E-017 100.000 28 -6.00E-017 -1.67E-016 100.000 29 -7.47E-017 -2.07E-016 100.000 30 -1.48E-016 -4.12E-016 100.000 31 -1.83E-016 -5.09E-016 100.000 32 -2.03E-016 -5.63E-016 100.000 33 -2.93E-016 -8.14E-016 100.000 34 -3.64E-016 -1.01E-015 100.000 35 -4.82E-016 -1.34E-015 100.000 36 -5.06E-016 -1.40E-015 100.000

184

Table B.2: Component Matrix from Principal Component Factor Analysis. Component 1 2 3 4 5 6 TotalTime .942 .021 -.267 -.142 .065 .054 MeanVelocity .276 .600 -.108 .702 -.040 -.076 SDV .685 .386 .244 .389 -.032 -.292 MinV -.516 .436 -.169 .447 .028 .155 TimeToMinV .220 .414 -.689 -.048 .046 -.153 VatS .686 .441 .310 .372 -.059 -.185 VatD .777 .010 .106 .459 -.357 -.098 VatG .176 .616 -.202 .493 .076 .109 VatE .284 .549 -.475 .349 -.123 -.260 HPatS -.238 .789 -.088 -.151 .409 -.185 HPatD .036 .801 -.290 -.181 .393 -.207 HPatG .796 -.142 -.514 .123 -.037 .100 HPatE .195 -.707 .217 .435 .326 -.151 HatS .622 -.034 .639 .098 .345 -.182 HatD .380 .047 .762 .076 -.266 -.052 HatG .468 .781 .067 -.309 .019 -.021 HatE .483 .712 .018 -.274 -.186 .259 AatS .747 .029 -.088 .057 .391 .251 AatD .823 -.345 .238 .058 -.183 .095 AatG -.369 .684 -.067 -.173 -.551 -.016 AatE .648 .376 -.480 -.158 -.284 .120 SatD -.536 .691 -.126 -.182 .373 -.053 SatG .914 .252 -.190 -.090 .002 .012 SatE .312 -.852 -.077 .022 -.099 .121 MaxPitch .444 .390 .603 -.434 -.094 -.136 MeanPitch .525 -.346 .116 -.460 -.021 -.456 SDPitch .205 .648 .625 -.287 -.108 .172 TimeSpentNearMin .610 -.286 -.251 -.603 .169 .220 SumPoles .741 -.230 -.232 -.453 .240 .102 Trap .943 .107 -.237 -.104 .017 .062 MeanRoll -.339 .300 .426 .175 .279 .638 MaxRoll -.243 .741 .181 .185 .069 .341 MeanHeadToPole .928 .031 -.272 .140 .068 .073 SDHeadToPole .753 -.261 .077 .277 -.076 .443 MaxTurnRate .211 .537 .704 -.347 -.105 .016 MeanTurnRate -.597 .093 -.497 -.255 -.513 .105

The correlation matrix between each variable (Table B.3) revealed that total time was correlated closest with the split between point two and point three (r = 0.94), followed by mean distance between head and inside pole (r = 0.93). Analysis of single point descriptors revealed that the points that were correlated to total time included the distance between head and inside pole when passing the gate (r = 0.87) and the split between point two and point three (r = 0.94) (Table B.3).

185

Table B.3: Correlation Matrix from the principal component factor analysis.

VatS VatE AatS AatE SatE

VatD HatS HatE AatD SatD

VatG HatD AatG Sat

HatG

HPatS HPatE

HPatD

HPatG

TotalTime

SDTotalTime

MeanHeadToPole SDMeanHeadToPole Correlation TotalTime 1.000 .925 .228 .149 .521 .607 .148 .336 -.133 .158 .869 .057 .406 .128 .496 .503 .771 .671 -.314 .766 -.412 .941 .298 MeanHeadToPole .925 1.000 .265 .014 .571 .729 .338 .440 -.196 .146 .912 .191 .446 .160 .388 .424 .711 .673 -.359 .695 -.437 .873 .309 SDTotalTime .228 .265 1.000 .354 .246 .277 -.167 -.303 -.288 -.339 .105 .485 .538 .511 .028 -.093 .208 .402 -.321 -.201 -.449 .171 .355 SDMeanHeadToPole .149 .014 .354 1.000 -.298 -.150 -.436 -.568 -.176 -.347 .085 .378 .205 .082 -.282 -.307 .101 .141 -.509 -.171 -.247 -.038 .499 VatS .521 .571 .246 -.298 1.000 .781 .413 .471 .144 .242 .334 .067 .628 .524 .582 .522 .449 .495 .006 .421 -.206 .661 -.187 VatD .607 .729 .277 -.150 .781 1.000 .369 .344 -.372 -.199 .593 .252 .471 .487 .206 .288 .390 .761 -.182 .445 -.649 .667 .224 VatG .148 .338 -.167 -.436 .413 .369 1.000 .527 .341 .438 .182 -.276 .002 -.114 .362 .380 .233 -.027 .208 .265 .282 .362 -.585 VatE .336 .440 -.303 -.568 .471 .344 .527 1.000 .296 .541 .480 -.248 -.075 -.144 .520 .346 .199 -.099 .352 .576 .184 .445 -.328 HPatS -.133 -.196 -.288 -.176 .144 -.372 .341 .296 1.000 .871 -.319 -.517 -.094 -.180 .527 .358 -.054 -.585 .443 .090 .864 .005 -.763 HPatD .158 .146 -.339 -.347 .242 -.199 .438 .541 .871 1.000 .012 -.514 -.023 -.279 .677 .516 .114 -.420 .371 .354 .756 .279 -.683 HPatG .869 .912 .105 .085 .334 .593 .182 .480 -.319 .012 1.000 .203 .184 -.033 .215 .283 .661 .570 -.342 .748 -.479 .744 .484 HPatE .057 .191 .485 .378 .067 .252 -.276 -.248 -.517 -.514 .203 1.000 .458 .147 -.563 -.621 .160 .386 -.776 -.396 -.583 -.117 .674 HatS .406 .446 .538 .205 .628 .471 .002 -.075 -.094 -.023 .184 .458 1.000 .702 .290 .163 .512 .582 -.508 -.040 -.285 .398 .151 HatD .128 .160 .511 .082 .524 .487 -.114 -.144 -.180 -.279 -.033 .147 .702 1.000 .222 .226 .125 .497 -.024 -.027 -.328 .153 .131 HatG .496 .388 .028 -.282 .582 .206 .362 .520 .527 .677 .215 -.563 .290 .222 1.000 .834 .387 .077 .431 .608 .332 .661 -.546 HatE .503 .424 -.093 -.307 .522 .288 .380 .346 .358 .516 .283 -.621 .163 .226 .834 1.000 .392 .201 .430 .760 .229 .613 -.377 AatS .771 .711 .208 .101 .449 .390 .233 .199 -.054 .114 .661 .160 .512 .125 .387 .392 1.000 .591 -.503 .483 -.221 .724 .178 AatD .671 .673 .402 .141 .495 .761 -.027 -.099 -.585 -.420 .570 .386 .582 .497 .077 .201 .591 1.000 -.509 .309 -.813 .629 .506 AatG -.314 -.359 -.321 -.509 .006 -.182 .208 .352 .443 .371 -.342 -.776 -.508 -.024 .431 .430 -.503 -.509 1.000 .233 .492 -.134 -.616 AatE .766 .695 -.201 -.171 .421 .445 .265 .576 .090 .354 .748 -.396 -.040 -.027 .608 .760 .483 .309 .233 1.000 -.077 .754 .036 SatD -.412 -.437 -.449 -.247 -.206 -.649 .282 .184 .864 .756 -.479 -.583 -.285 -.328 .332 .229 -.221 -.813 .492 -.077 1.000 -.306 -.758 SatG .941 .873 .171 -.038 .661 .667 .362 .445 .005 .279 .744 -.117 .398 .153 .661 .613 .724 .629 -.134 .754 -.306 1.000 .007 SatE .298 .309 .355 .499 -.187 .224 -.585 -.328 -.763 -.683 .484 .674 .151 .131 -.546 -.377 .178 .506 -.616 .036 -.758 .007 1.000

186

B.4 Discussion Three components were found to explain the majority of the variability observed in the data from all variables. Component one, which described the greatest percentage of the variance, was also the most closely related to the upstream gate performance as measured by total time. Variables which were related to component one were also indicative of total time and upstream gate performance. These included the mean distance between the paddler‟s head and inside pole and split between points two and three. Paddlers wishing to improve their upstream gate performance could achieve the greatest improvements by concentrating on the time they spend between points two and three. In addition, concentrating on the distance between the paddler‟s head and the inside pole appears also to have a strong influence on upstream gate performance. Component two had almost no impact on upstream gate performance as it had no relationship with total time. The types of variables that were related to component two were the inverse of the split time between points three (gate) and four (end) (negative correlation) and the distance between the paddler‟s head and pole at point two. Component three was most closely related to heading at point two (down) and maximum turn rate but weakly negatively correlated with total time. Therefore if a single component was to be used to describe upstream gate performance, it would be component one and the main variable that is not time based is the mean distance between the head and inside pole. However to further reduce the analysis required to extract upstream gate performance, discrete descriptors (measured at one point in time) were investigated. Of these, the best measures of upstream gate performance (total time) were distance between head and inside pole when passing the gate and the split between point two (down) and point three (gate).

Paddlers wishing to improve their upstream gate performance could achieve the greatest improvements by focusing on the time they spent between points two and 3. In addition, concentrating on the distance between the paddler‟s head and the inside pole also appears to have a strong influence on upstream gate performance. However, it would be expected that there would be an optimal distance beyond which any further reduction in the distance would impede technique and performance due to touching the gate becoming an increased risk.

187

APPENDIX C INFORMED CONSENT – TESTING PROCEDURES FOR FLATWATER SPRINT ANALYSIS

188

PARTICIPANT INFORMATION

Performance and Biomechanical Analysis of Flatwater Kayaks

Researcher Mr Adam Hunter (Hons), Doctor of Philosophy in School of Health Science at the University of Canberra

Supervisors Ms Jodie Cochrane (Hons), Biomechanics Department of Australian Institute of Sport Dr Margi Böhm (PhD), School of Resource, Environmental and Heritage Science Dr Keith Lyons (PhD), Performance Analysis Department, Australian Institute of Sport

Funding / Support This project is funded and supported by the Australian Institute of Sport and the University of Canberra.

Project Aim The aim of this project is to obtain biomechanical data on a variety of boat types and athletes to identify biomechanical factors which result in faster boat and paddler performance. There is very little published literature on optimising boat and paddler performance in kayaking. The results of this investigation will have implications for all kayak paddlers as there are currently no scientifically justified methods to setup a kayak for optimal performance. The development and refinement of the methodology used in this investigation will allow paddlers to be assessed quickly and easily to determine their individual optimum setup.

Benefits of the Project The data gathered in the project will be processed and statistically analysed to identify variables that are related to boat and paddler performance. These data will form a database of information about each boat type which can be updated following additional testing and comparisons and which can be used for boat selection. Currently there is a paucity of scientific data on boat setup and paddler performance in kayaking. Using the appropriate protocols, a biomechanical profile of each boat will be developed to determine relationships between biomechanical measures and performance such as the coefficient of drag for various boats. As a participant you will receive a research report which contains information about your performance across the boats you were tested in and if you request you can be provided with video footage from each trial. As a participant the report will be sent directly to you. Please note the assurances of confidentiality listed in this information sheet on page 2.

General Outline of the Project This research involves assessment of a number of different boat types to determine factors which relate to the optimum performance of each boat. As a participant you will be involved in a range of trials varying in speed and distance. You will be given sufficient time to recover between trials. The length of each session will be dependent on your availability. The nature of these trials is similar to the type of activity you would be involved in during a normal training session.

As a participant you will have your boat instrumented with a non permanent sensor box that will be attached to the floor of the kayak using adhesive tape which records information about the position and orientation of your boat. For velocity controlled trials a global positioning

Postal Address: University of Canberra ACT 2601 Australia Location: Kirinari Street Bruce ACT Telephone: +61 (0)2 6201 5111 Facsimile: +61 (0)2 6201 5999 World Wide Web: http://www.canberra.edu.au

PARTICIPANT INFORMATION

Performance and Biomechanical Analysis of Flatwater Kayaks system will be temporarily attached to your kayak so you can view boat velocity. In addition to this boat instrumentation you may have your paddle instrumented with non-permanent force transducers which measure the forces you are applying to the paddle. These are fastened to the shaft using clamps fittings and a second box placed in the boat which records the information from the force transducers on your paddle. To allow measurements to be taken from video some flat non-permanent adhesive markings will also be attached to the exterior of the boat. Once these testing devices are fitted to your equipment and calibrated testing can begin.

During each trial your performance will also be recorded on up to two cameras and an electronic timing system to provide a record of your overall performance as well as some reference information. During paddling trials participants will also be asked to wear a heart rate monitor strap around their chest so that their levels of physical exertion during the trials can be assessed and monitored for recovery purposes. The information from the boat orientation / position sensors will also be collected during each trial. Information from the force transducers will be collected during trials where theses sensors are attached to the paddle. These data will be correlated to determine if any relationships exist and therefore highlight variable which may contribute to an improved performance of the boat. Results will be presented in a form where no single paddler will be identifiable for confidentiality reasons.

Participant Involvement As a participant you will be asked to perform only as many trials as you feel comfortable doing in each session which will vary depending on the actual length of the trials to be completed and the length of time which you have available. Over a number of sessions you will be involved in paddling a variety of boats types. Sessions will involve you performing a variety of different trials ranging from stationary (non active trials) to paddling trials controlled for velocity with a maximum distance of approximately 1500 m. Velocity controlled trials will range in speed from very slow (eg 1 km/hr) up to race speeds (eg 18 km/hr). You will be given sufficient time to recover between trials using a work to rest ratio of 1:10. Additional time either side of test will be required for researchers the setup and calibrate testing equipment on your paddle and boat, but as a subject you need not be present during these periods. Estimated time for equipment setup and calibration will be between 10 min and two hrs before the session, depending on the specific trial, and 5 to 30 min after.

Example Trials: Stationary trials – no paddling, sitting in the boat to determine trim of boat Deceleration runs – paddling the boat up to near top speed in their own time (should take approximately 30 to 60 seconds of paddling to reach top speed) and then stopping paddling and letting the boat coast to a stop without putting the paddle in the water. Trim trials – altering the position of the centre of mass within the boat so that the stationary trim of the boat is altered (measured through stationary trial). Then through a deceleration run, measure the effect on the boats performance. Active trim trials – paddling at a specific speed (ranging from 1km/hr to 18km/hr) for approximately 100 m so that boat’s active trim range and other boat specific variables can be monitored. Paddle and Boat Performance trials – during these trials force on the paddle, boat performance (including acceleration) will be measured while paddling at a specific

190

PARTICIPANT INFORMATION

Performance and Biomechanical Analysis of Flatwater Kayaks speed (see Active trim trials). Trials similar to active trim trials however the acceleration portion of the trial will also be recorded. Rudder trials – altering the size and shape of the rudder of a specific boat to determine the influence of the rudder size and shape on the performance of the boat. Trials performed according to the protocols for Active trim and Paddle and Boat Performance trials . Boat performance comparison – this involves paddling at a constant work rate (based on heart rate) for a 1.5 km course in a variety of boats to allow for a comparison between boats based on time and velocity.

Example Session: (paddler commitment 1 ½ hours) Equipment setup 1 ½ hours (paddler not required to be present) - testing equipment attached to boat and paddle - calibration of testing equipment Warm-up 10-20 min (paddler’s standard warm-up procedure) Testing Session 1 hour (depending on paddler availability) - 10 x deceleration trials - 30 sec paddling followed by 5 min recovery Cool-down 10-20 min (paddler’s standard cool-down procedure) Equipment Removal 30 min (paddler not required to be present)

Discomfort / Risks Potential risks to you during the collection of these data are no greater than those experienced during a normal training session.

Ethics Committee Clearance This project has been approved by the Committee for Ethics in Human Research of the University of Canberra and the Ethics Committee from the Australian Institute of Sport.

Queries and Concerns If you have any further queries and concerns about the research then contact Adam Hunter, Jodie Cochrane or Margi Bohm: Adam Hunter Jodie Cochrane Margi Böhm (02) 6214 7914 (02) 6214 1558 (02) 6201 2058 (02) 6214 1593 (02) 6214 1593 (02) 6201 2328 [email protected] [email protected] [email protected]

If you have any concerns with respect to the conduct of this study, you may contact the Secretary of the AIS Ethics Committee (Dr John Williams) on (02) 6214 1816. 191

INFORMED CONSENT

Performance and Biomechanical Analysis of Flatwater Kayaks

Principal Researchers: Adam Hunter

The investigation and my part in the investigation have been defined and fully explained to me by Adam Hunter and I understand the explanation. A copy of the procedures of this investigation and a description of any risks and discomforts has been provided to me and has been discussed in detail with me.

I have been given an opportunity to ask whatever questions I may have had and all such questions and inquiries have been answered to my satisfaction. I understand that I am free to deny any answers to specific items or questions in interviews or questionnaires. I understand that I am free to withdraw consent and to discontinue participation in the project or activity at any time. I understand that any data or answers to questions will remain confidential with regard to my identity. I certify to the best of my knowledge and belief, I have no physical or mental illness or weakness that would increase the risk to me of participating in this investigation. I am participating in this project of my own free will and I have not been coerced in any way to participate. I understand that I will receive a report containing results from testing. I understand that this data cannot be duplicated. I understand that these data can only be published, reported or presented by the named researchers or supervisors on this project: Adam Hunter, Jodie Cochrane, Margi Bohm and Keith Lyons.

Signature of Subject: ______Date: ___/___/___

A summary of the research report can be forwarded to you when published. If you would like to receive a copy of the report, please include your mailing address below. Name______Address______

I, the undersigned, was present when the study was explained to the subject in detail and to the best of my knowledge and belief it was understood.

Signature of Researcher: ______Date: ___/___/___

Postal Address: University of Canberra ACT 2601 Australia Location: Kirinari Street Bruce ACT Telephone: +61 (0)2 6201 5111 Facsimile: +61 (0)2 6201 5999 World Wide Web: http://www.canberra.edu.au