Presentation is loading. Please wait.

Presentation is loading. Please wait.

Analysis of the accuracy of Argus in identifying shoreline position: comparison of two procedures with in-situ measurements in Lido di Dante Silvia Medri.

Similar presentations


Presentation on theme: "Analysis of the accuracy of Argus in identifying shoreline position: comparison of two procedures with in-situ measurements in Lido di Dante Silvia Medri."— Presentation transcript:

1 Analysis of the accuracy of Argus in identifying shoreline position: comparison of two procedures with in-situ measurements in Lido di Dante Silvia Medri University of Bologna CoastView Meeting Lisbon 18 – 19 Sept 2004

2 Contents 1. Comparison of two procedures to find waterline  Description of the two procedures  Justification of the use of the two procedures  Results presentation  Comments 2. Comparison of the two procedures in the study of shoreline evolution  Results presentation  Comments 3. Comparison of the two procedures with in-situ measurements  Fieldwork description: GPS Survey, 15-22 July 2004  GPS survey data  GPS-Argus Comparison  Conclusions

3 1. Comparison of the two procedures  Description: first procedure  First merge images with AMT  Then find WL with IBM

4 1. Comparison of the two procedures  Description: second procedure  First find WL with IBM on single oblique images  Then “merge” single cameras WLs in a unique WL with Matlab.

5 1. Comparison of the two procedures  Justification of the use of the two procedures  Discrepancies at a qualitative inspection between obtained WLs  Tidal levels assignment  Operative differences 1st procedure: First merge then use IBM 2nd procedure: First use IBM then merge Advantages: - accurate in the ROI overlap - accurate in identifying obstacles in the far field - at the present level of implementation fast Advantages: - proper tidal level for each image - accurate in identifying details, due to the greater pixel resolution Disadvantages: - approximated tidal levels - decrease of visualization of small objects (persons, umbrellas, etc..) - less image definition Disadvantages: - not accurate in the region of cameras overlap - decrease of the possibility of identifying oblique obstacles -pixel resolution loss in the far field - at the present level of implementation slow

6 1. Comparison of the two procedures  Presentation of results Shoreline at the same hour obtained with the two procedures. [20 July 2004 06 GMT]

7 1. Comparison of the two procedures  Reasons for a reduced area of calculations  Decreasing pixel resolution  worse representation of far field pixel intensities in dry/wet separation procedure  increase of the difficulty for the operator to visually distinguish the shoreline itself  WL position differences are calculated on the X axis of Argus reference system  overestimate these differences where the shoreline is not parallel to Y axis  Systematic tidal data delay Whole area Reduced area

8 1. Comparison of the two procedures  Statistics of differences WLs at the same hour obtained with the two procedures for one day of the fieldwork. [20 July 2004 from 06 to 14:00 GMT]

9 1. Comparison of the two procedures  Statistics of differences WLs at the hour obtained with the two procedures for the whole period of the fieldwork. [ From 15 July to 22 July 2004] The Mean value of the Median of the absolute values of the differences between shorelines obtained with the two procedures indicates the two procedures disagree of about 0.58 m in both the whole and the reduced areas.

10 1. Comparison of the two procedures  Comments  The quantified disagreement between the two procedures (of 0.6 m) is not so important if compared with shoreline dynamics itself.  Against the expectation there aren’t significant improvements/variations in reducing the area of work. ? Probably the two systematic errors in timing the merged images and tide hide the differences.  There is a recurrent presence of outliers especially at the end of the area (due to unavoidable decreasing pixel resolution in the far field) and near the central groin (due to a significant density in that area of tourists and big umbrellas).

11

12 2. Comparison of the two procedures in the study of shoreline evolution  Presentation of results Shoreline evolution during 1 day of the fieldwork with the two procedures. [20 July 2004 from 06 to 15 GMT]

13 2. Comparison of the two procedures  Statistics of shoreline evolution Shoreline evolution during 1 day of the fieldwork with the two procedures calculated for the whole area and for the reduced area [20 July 2004 from 06 to 15 GMT]

14 2. Comparison of the two procedures  Analysis of WL displacements quality  1 st criterion: to verify if the WL variation is coherent in sign with the corresponding tidal level variation, indicating a clear trend of WL evolution (minimum requirement).  2nd criterion: to verify, in the unavoidable presence of small discrepancies (only few points in against trend), if however: |Median - Median abs| < 0.1 m indicating a low percentage and a small entity of values in against trend. In these cases the signs of shoreline displacements are concordant among them, reproducing a mean intertidal beach profile with a constant sign slope (further requirement).

15 2. Comparison of the two procedures Estimated mean intertidal beach profiles The profiles show a qualitative good agreement with the realistic situation, in particular it is shown a simple morphology with lack of significant bars and throughs (Lido di Dante protected area)

16 WLs evolution for every day of the fieldwork with the two procedures calculated for both the whole and the reduced area 2. Comparison of the two procedures  Statistics of shoreline evolution

17 2. Comparison of the two procedures  Comments on shoreline evolution  In the whole area:  The 1 st procedure (First merge then use IBM) gives results in very good agreement with the tidal excursion in about 60% of the instances, and fairly good results in about 30% of the instances in whom it was used.  The 2 nd procedure (First use IBM then merge) gives results in very good agreement with the tidal excursion in about 79% of the instances and fairly good results in about 15% of the instances on whom it was used.  In the reduced area:  The 1 st procedure (First merge then use IBM) gives results in very good agreement with the tidal excursion in about 64% of the instances in whom it was used, with a small improvement with respect to the whole area.  The 2 nd procedure (First use IBM then merge) gives results in very good agreement with the tidal excursion in about 88% of the instances in whom it was used, with a small improvement with respect to the whole area.

18 2. Comparison of the two procedures  Conclusions on shoreline evolution  Considering only ‘very good’ results, the 2 nd procedure (SINGLE) is better then the 1 st one (MERGE).  Considering the whole of ‘very good’ + ‘fairly good’ results, both the procedures are equivalently of high quality in the great part (about 90%) of the instances.  Both the procedures improve their performance in the reduced area.  Robustness test: the 2 nd procedure is the most robust because it gives better results even in the whole area where there are the great part of outliers.

19

20 3. Comparison with in-situ measurements Objectives  To validate video derived data  To provide demonstration CSIs for the Lido di Dante site.  To check what is the best way to use Argus IBM. Strategy  To compare a significant set of surveyed shorelines with the corresponding Argus waterlines, in order to evaluate the capability of Argus in identifying WLs for both the analysed procedures  To measure a significant set of surveyed cross-shore transects in order to have a more realistic estimate of intertidal beach slope.

21 3. Comparison with in-situ measurements Area of interest 40 shorelines  15 July (GMT 09:00, 11:00, 12:00 = 3 WL)  19July (from GMT 09:00 to GMT 16:00 = 7 WL)  20 July (from GMT 06:00 to GMT 15:00 = 10 WL)  21 July (from GMT 07:00 to GMT 16:00 = 10 WL)  22 July (from GMT 06:00 to GMT 15 :00 = 10 WL) 60 transects  15 July ( 2 transects)  19July (11 transects)  20 July (10 transects)  21 July (12 transects)  22 July (25 transects) Available data set 15-22 July 2004 GPS Survey

22 3. Comparison with in-situ measurements GPS Technical Specifications: LEICA SR510 Receiver Satellite Reception: single-frequency Receiver Channels: 12 L1 continuous tracking L1 channels: Carrier phase, C/A narrow L1 Carrier Tracking: reconstructed carrier phase via C/A code Satellites tracked: up to 12 simultaneously Kinematic Mode : after a static or known point initialisation, kinematic surveys can be carried out and with dynamic procedures in post processing yields 10 to 20mm + 2ppm rms Antenna The AT501 is a small precision centered L1 antenna, L1 microstrip with Built-in groundplane. CORS: Continously Operating Reference Station Manager: Prof. Susanna Zerbini, Dept. of Geophysic, University of Bologna. Position: Marina di Ravenna (8 Km far from Lido di Dante site) Data collection rate: 30 sec 15-22 July 2004 GPS Survey

23 3. Comparison with in-situ measurements Shoreline surveyed hourly basis, simultaneously with the recording of time-averaged video images by the 4-cameras of Argus video station in LdD. 15-22 July 2004 GPS Survey

24 3. Comparison with in-situ measurements Offshore Wave Heights range: (0.07 ÷ 0.5) meters 15-22 July 2004 GPS Survey

25 3. Comparison with in-situ measurements Wind Direction (°N) (°N) Temperature (°C) (°C) Wind Direction (°N) (°N) Sea Water Level (m) (m) Atmospheric Pressure (hPa) (hPa) Water temperature (°C) (°C) Wind Intensity (m/s) (m/s) Tidal level range: -0.5 ÷ +0.3 meters 15-22 July 2004 GPS Survey

26 Presentation of results

27 3. Comparison with in-situ measurements  1 st procedure (First merge then use IBM) – GPS Comparison

28 3. Comparison with in-situ measurements  2 nd procedure (First use IBM then merge) – GPS Comparison

29 3. Comparison with in-situ measurements  GPS - Argus Comparison  Results  The two procedures are equivalent on a set of normal (without spurious results) data.  If there are significant outliers the 1 st procedure seems better than the 2 nd one.

30 3. Comparison with in-situ measurements  Comments  Result interpretation: the combined effect of two systematic errors in timing the merged images and tide results in hiding the great part of differences on merged images (1 st procedure), while frustrating the precision in the synchronism with tidal data of oblique images (2 nd procedure).  Result judgement: there is a good agreement of both the used Argus procedures to find shoreline location with ground truth data of GPS survey, considered the shoreline dynamics itself and the time shift in tidal data. (With correct tide data the 1 st procedure is expected to give the better results (with a dispersion <0.5 m )).

31 3. Comparison with in-situ measurements  Future work  To estimate the systematic error in our tide data by comparison with local measured data (past intensive fieldwork).  To estimate intertidal beach slope through the analysis of GPS intertidal transects.  To compare the two Argus procedures with GPS data even in the reduced area.  To carry out another GPS survey in October to have “cleaner” shorelines.

32

33 2. Shoreline Variation due to a storm Topographical Survey - O8 Oct 2003 Topographers Nourishment works

34 3. Comparison with in-situ measurements July 2004 GPS Survey

35 3. Comparison with in-situ measurements 15-22 July 2004 GPS Survey Available data set 40 shoreline :  15 July (GMT 09:00, 11:00, 12:00 = 3 WL)  19July (from GMT 09:00 to GMT 16:00 = 7 WL)  20 July (from GMT 06:00 to GMT 15:00 = 10 WL)  21 July (from GMT 07:00 to GMT 16:00 = 10 WL)  22 July (from GMT 06:00 to GMT 15:00 = 10 WL) 60 transects  15 July ( 2 transects)  19July (11 transects)  20 July (10 transects)  21 July (12 transects)  22 July (25 transects)

36


Download ppt "Analysis of the accuracy of Argus in identifying shoreline position: comparison of two procedures with in-situ measurements in Lido di Dante Silvia Medri."

Similar presentations


Ads by Google