analysis of bathymetrical derived features on the belgian
TRANSCRIPT
44
!! 4
444
((
((
4!
((
((
44
4
4
4
4
44
4
4
4
(
444
4
4
4
4
444
4
4
4
4
!!
((
((
((
(
(( (
(
(
((
((
(
((
((
((
(
((
(
(
(
((
(
((
(
(
((
(
4
(
((
((
(
(
(
((
(((
((
(((
((
((
((
((
(
((
(
((
(
((
((
((
((
((
((
(
((
((
((
((
((
((
(
(
(( (
((
((
((
((
((
(
( (
((
((
((
((
((
((
((
4
((
4
((
((
((
((
((
((
((
((
((
((
((
((
(
(
((
((
(
((
(
((
(
(
(
((
((
((
((
((
((
(
(
((
(
((
(((
((
(
4
444
!!
!!
!!
466000 468000 470000
5704
000
5706
000
bedforms (Deleu et al., 2004)medium to large dunes
( ( very-large dunes with lee slope to the NE4 4 very-large dunes with lee slope to the SW
! ! very-large symmetric duneBPI classes
small scale crestscrestsdepressionsflatsslopes
0 0.5 10.25km
UTM31N - WGS84 coordinates
Easting (m)
Nor
thin
g (m
)
Analysis of bathymetrical derived features on the Belgian continental shelf as a support for marine habitat mapping
Verfaillie Els1, Foster-Smith, Bob2,Van Lancker Vera1
1Ghent University, Renard Centre of Marine Geology (RCMG), Krijgslaan 281, S8, 9000 Gent, Belgium ([email protected], Tel. +32 9 264 45 73, Fax +32 9 264 49 67)2 Envision, University of Newcastle, Newcastle upon Tyne, United Kingdom
INTRODUCTIONThe study area is a kink area in the middle part of the Westhinder offshore tidal sandbank, where a multibeam dataset is available with a resolution of 2 m (Figure 1). Deleu et al. (2004) studied the morphology of this area thorougly. Both on a large and a medium scale the occurrence of features (from sandbank level until medium dune level using the classification of Ashley (1990)) was described and interpreted. Bedforms were manually drawn based on the bathymetry of the map, indicating their strike and asymmetry.The aim of this study is to facilitate and automate the geomorphological analysis, comparing different bathymetrical derived features. Furthermore the geomorphology is analysed in the context of marine habitat mapping, as topographic features are assumed to be important possible habitats for marine organisms.
!(
!(
!(
!(
!(
!(
!(!(
!(
!(!(
!(
!(
!(
!(
!(
!(
!(
!(
!(
!(
!(
Smalbank
Gootebank
Oostdijck
Trapegeer
Den Oever
Fairy Bank
Bligh Bank
Kwintebank Ravelingen
Broersbank
BalandbankStroombank
Akkaertbank
Thorntonbank
Buiten Ratel
OostendebankWenduinebank
Westhinderbank
Oosthinderbank
Nieuwpoortbank
Noordhinderbank
Middelkerkebank
440000 460000 480000 500000 520000
5660
000
5680
000
5700
000
5720
000
5740
000
0 5 102.5km
UTM31N - WGS84 coordinates
Easting (m)
Nor
thin
g (m
)
Bathymetry data compiled from the Ministry of the Flemish Community AWZ - WWK
added with data from the Hydrographic Officeof the Netherlands and the UKHO
!( Hinderbanken
!( Kustbanken
!( Vlaamse Banken
!( ZeelandbankenBathymetry (m)
460
study area Westhinder
study area Hinderbanken
MATERIALS AND METHODS1. Bathymetric position index or BPI: • A measure of where a location with a defined elevation is relative to the overall landscape (Weiss (2001), Iampietro & Kvitek (2002)
and Lundblad et al. (in press));• Evaluates bathymetry differences between a focal point and the mean elevation of the surrounding cells within a user-defined
rectangle, annulus or circle;• Aims to distinguish slopes, valleys, ridges and flat areas; generally calculated on a broad-scale (B-BPI) and a fine-scale (F-BPI);• Applies a scalefactor (= resolution of bathymetry multiplied by outer radius of annulus) of 24 and 250 (because the medium and large
dunes in this area have an average wavelength of 20 to 25 m, the very-large dunes of 250 m);• Calculated with ArcGIS Benthic Terrain Modeler or BTM (Wright et al., 2005).
44
!! 4
444
((
((
4!
((
((
44
4
4
4
4
44
4
4
4
(
444
4
4
4
4444
4
4
4
4
!!
((
((
(((
(( (
(
(
((
((
(
((
((
((
(
((
(
(
(
((
(
((
(
(
((
(
4
(
((
((
(
(
(
((
(((
((
(((
((
((
((
((
(
((
(
((
(
((
((
((
((
((
((
(
((
((
((
((
((
((
(
(
(( (
((
((
((
((
((
(
( (
((
((
((
((
((
((
((
4
((
4
((
((
((
((
((
((
((
((
((
((
((
((
(
(
((
((
(
((
(
((
(
(
(
((
((
((
((
((
((
(
(
((
(
((
(((
((
(
4
44
4
!!
!!
!!
466000 468000 470000
5704
000
5706
000
bedforms (Deleu et al., 2004)medium to large dunes
( ( very-large dunes with lee slope to the NE4 4 very-large dunes with lee slope to the SW
! ! very-large symmetric duneIsocluster classes
1
2
3
4
5
6
7
8
9
10
0 0.5 10.25km
UTM31N - WGS84 coordinates
Easting (m)
Nor
thin
g (m
)
RESULTS1. Bathymetric position index or BPI: • Classification decision tree (Figure 2), based on B-BPI (4 classes: depressions, crests, flats, slopes) and F-BPI classification (1 class:
small-scale crests) results in 5 classes: Figure 3;• ‘Small-scale crests’ class corresponds to a large extent to the crestlines of the very-large dunes and, partly, to the crestlines of the
medium to large dunes, defined in Deleu et al. (2004) (Figure 3);• Combined with an aspect map (orientation of the slopes: Figure 4) the asymmetry of bedforms can be derived;• Figure 6: SW-NE profile through a sandbank, very-large dunes, medium to large dunes and 9 sediment samples shows that most of
morphological features are extracted using the BPI classification;• Most of the tops/slopes of the very-large dunes (central part of profile between 1000 and 4000 m on Figure 6) are clearly classified
as small-scale crests;• Depressions/crests are only derived for larger features. For medium to large dunes (at right side of 4000 m on
profile) only the largest crestlines, depressions and steepest slopes are derived based on the classification;• Application on larger area in Hinderbanken (Figure 7): clear distinction between bedforms on sandbanks and swales.
2. Topographic feature recognition – Isocluster classification:• Figure 5: similar features could be extracted, although sometimes difficult interpretation of classes;• Better extraction of crestlines of medium to large dunes, especially in the NW part of study area;• Class 1, 3 = noise; class 2 ≅ BPI-flats; class 4, 7, 10 ≅ BPI-depressions; class5, 9 ≅ BPI-small-scale crests;
class 6 ≅ BPI-slopes; class 8 ≅ BPI-crests.
Figure 1: Position of Westhinder study area and larger Hinderbanken study area
FURTHER RESEARCH• Application of bathymetric derived features on other areas• Correlation of bathymetric features with other physical data layers such as multibeam backscatter
and sedimentological samples• Correlation with marine species/communities to test if they are important possible habitats• Other bathymetric derived features will be tested as input parameters for models (e.g. rugosity = topographic
roughness with a surface area to planar area ratio, as input for sand transport or biological models)
2. Topographic feature recognition – Isocluster classification:• Series of different filters with varying window sizes, reveals different scale
features of the bedform morphology;• Subtraction of filtered images from other images, with smaller cell sizes, creates
residual filtered images;• Filtered images, with varying search radii, reveal bedforms ranging from 10 to
100 m and with heights between 0.4 and 3 m;• Key scales on different scales (ranging from fine-scale until broad-scale
features) are selected from the filtered images to come to a classified image;• Calculated with Idrisi software Isocluster unsupervised classification (based on
maximum likelihood procedure) of the residual images, creating 10 clusters ofbedform features.
Figure 2: Classification decision tree developed for Westhinder study area of a B-BPI and a F-BPI classification, resulting in 5 classes: depressions, crests, flats, slopes and small-scale crests
460000 465000 470000 475000
5695
000
5700
000
5705
000
BPI classessmall scale crestscrestsdepressionsflatsslopes
0 2 41km
UTM31N - WGS84 coordinates
Easting (m)
Nor
thin
g (m
)
W9413
WH0203.058
WH0203.057 WH0203.052
WH0203.051
WH0203.050
WH0203.049
WH0203.048
WH0203.043
WH0203.042
WH0203.041
WH0203.040
WH0203.036
WH0203.035
WH0203.034WH0203.033
WH0203.028
WH0203.027
WH0203.026
WH0203.025
WH0203.022
WH0203.021
WH0203.020
WH0203.019WH0203.018
WH0203.013
WH0203.012
WH0203.011
WH0203.010
WH0203.007
WH0203.006
468000 470000
5704
000
5706
000
0 0.5 10.25km
UTM31N - WGS84 coordinates
grab samples
Bathymetry (m)
-6.00
-38.00
B
A
Nor
thin
g (m
)
Easting (m)
1
2
3
4
5
6
7
8
9
10
Isocluster classification
A
B
wh0203.058 wh0203.049wh0203.042
wh0203.035wh0203.026
wh0203.021
wh0203.010
wh0203.007W9413
small scale crests
crests
depressions
flats
slopes
vertical exaggeration of profiles: 40
BPI classification
A
B
wh0203.058 wh0203.049 wh0203.042
wh0203.035wh0203.026
wh0203.021
wh0203.010
wh0203.007W9413
AB
0 1000 2000 3000 4000 5000distance (m)
-30
-20
-10
bath
ymet
ry (m
)
wh0203.058 wh0203.049 wh0203.042wh0203.035 wh0203.026
wh0203.021
wh0203.010
W9413wh0203.007
REFERENCES• Ashley, G.M., 1990. Classification of large-scale subaqueous bedforms: a new look at an old problem. Journal of Sedimentary Petrology, 60 (1), 160-172. • Deleu, S., V. Van Lancker, D. Van den Eynde, and G. Moerkerke, 2004. Morphodynamic evolution of the kink of an offshore tidal sandbank: the Westhinder Bank (Southern North Sea). Continental Shelf Research, 24 (15), 1587-1610.• Iampietro, P., and R. Kvitek, 2002. Quantitative seafloor habitat classification using GIS terrain analysis: Effects of data density, resolution, and scale. In Proceedings of the 22nd Annual ESRI User Conference. San Diego, CA, July 8-12.
http://gis.esri.com/library/userconf/proc02• Lundblad, E., D. J. Wright, J. Miller, E. M. Larkin, R. Rinehart, S. M. Anderson, T. Battista, D. F. Naar, and B. T. Donahue, in press, A Benthic Terrain Classification Scheme for American Samoa. Submitted to Marine Geodesy.• Weiss, A. D. 2001. Topographic Positions and Landforms Analysis (Conference Poster). ESRI International User Conference. San Diego, CA, July 9-13.• Wright, D.J., Lundblad, E.R., Larkin, E.M., Rinehart, R.W., Oregon State University Davey Jones Locker Seafloor Mapping/Marine GIS Lab (project background at http://dusk.geo.orst.edu/esri04/p1433_ron.html) Joshua Murphy, Lori Cary-Kothera, Kyle Draganov NOAA Coastal
Services Center
Figure 3: B-BPI and F-BPI classification based on classification decision tree (Figure 2)
Figure 5: Isocluster classification of 4 key scale filtered images resulting in 10 classes
Figure 6: SW-NE profile through the study area; comparison between BPI and Isocluster classification
Figure 7: BPI classification applied on multibeam bathymetric dataset (resolution 2 m) of larger area in Hinderbanken: clear distinction between bedforms on sandbanks and swales. Patchy pattern in swales possible indicator of gravel
B-BPI
<= -1 stdev -1 stdev < B-BPI < 1stdev>= 1 stdev
Depressions Crests Flats & slopes
Slope
<= 4,01
Flats
> 4,01
Slopes
F-BPI
>= 1 stdev
Small scale crests
< 1 stdev
Crests
F-BPI
>= 1 stdev
Small scale crests
< 1 stdev
Depressions
F-BPI
>= 1 stdev
Small scale crests
< 1 stdev
Flats
F-BPI
>= 1 stdev
Small scale crests
< 1 stdev
Slopes
Criteria used in BPI classifications1 stdev = 1 standard deviation, corresponding to 100 grid value units
Raster grid derived from bathymetry
Resulting classes
Figure 4: Aspect map of Westhinder area makes it possible to derive asymmetry of bedforms
44
!
4
444
((
(
4!
((
((
44
4
4
44
44 4
4
4
(
4
44
4
44
4
44
44
4 !(
(
((
(
(
(( (
(
(
((
(
((
(
((
((
(
((
(
(((
((
4
(
((
(
(
(
(
(
(
((
((
((
((
((
((
((
(
((
((
(
(
((
((
((
((
((
(
((
((
((
((
((
(
(
(( (
((
((
((
((
(
( (
((
((
((
((
((
(
4
((4
((
((
((
((
((
((
((
((
((
((
((
((
((
((
((
(
(
((
((
((
((
((
(
(
(
((
((
(
((
4
4
4
!!
!!
!
466000 468000 470000
5704
000
5706
000
bedforms (Deleu et al., 2004)medium to large dunes
( ( very-large dunes with lee slope to the NE4 4 very-large dunes with lee slope to the SW
! ! very-large symmetric duneaspect (orientation of slope)
Flat (-1)North (0-22.5)Northeast (22.5-67.5)East (67.5-112.5)Southeast (112.5-157.5)South (157.5-202.5)Southwest (202.5-247.5)West(247.5-292.5)Northwest (292.5-337.5)North (337.5-360)
0 0.5 10.25km
UTM31N - WGS84 coordinates
Easting (m)
Nor
thin
g (m
)