Practical experience of bathymetric LIDAR Project U.L.T.R.A DEVELOPING LIDAR-BASED ANALYSIS FOR MARINE MAPPING Johnny Berglund, County Administrative Board of Västerbotten
BACKGROUND TO PROJECT ULTRA Very limited knowledge of underwater nature in combination with rapidly increasing exploitation pressure (marine wind farms) Regional spatial planning and/or environmental impact assessments unreliable without basic seafloor information Existing methods (ship-based echosounding) ineffective in shallow areas (where the highest natural values are found) Airborne LIDAR is very fast, but has so far generated only depth information (bathymetry) The ULTRA-project s main goal = extract additional seafloor information from LIDAR (substrate and vegetation)
WHAT IS LIDAR? LIDAR (Light Detection and Ranging) is an active remote sensing method, based on shooting a laser pulse and measuring how long it takes for the reflection to return. Since the speed of light is a constant the measured time (for the reflection) can then be used to calculate the distance (+/-a few centimeters). Geopositioning is achieved using aeronautical GPS and gyros. Marine LIDAR systems use two different lasers, one red (terrestrial Lidar) and one green (cyan). The red pulse does not penetrate water but is used to calibrate a more exact depth (depth = the difference between the surface and the seafloor).
WHAT IS LIDAR?
TEST AREAS UMEÅ VAASA 10km
TEST AREA IN SWEDEN 3km 7km
Potamogeton perfoliatus
Bladderwreck Fucus radicans SOFT BOTTOMS
BOULDERS - STONES
SAND
BASIC LIDAR BATYMETRY RESULTS LIDAR data (right) has a horisontal resolution of 2x2 meters, resulting in a much more detailed seafloor topography (in shallow areas). 200 m
Resultat, Djupdata
DEPHT DATA IN 3D Reefs Nature 2000 habitat 1170
EXTRACTING NEW DATA FROM LIDAR The seafloor substrate and vegetation generates interference in the reflection of the laser pulse, something which traditionally has been seen as a problem (when trying to measure only depth). The ULTRA project managed to extract geological and biological information contained in this disturbance (the shape of the LIDAR curve) and use the results to classify important habitat groups on the seafloor. SEA SURFACE SEAFLOOR
EXTRACTING DATA (a) (b) Example of a lidar waveform at bottom depth 6.1 m. Parts of the waveform (a), interpolated volume backscatter (b), difference wave of the bottom echo pulse (c), and an example of 50 % and 100 % level of the bottom echo pulse (d).
FIELD WORK
LIDAR SIGNAL DATA Analysis by M Tulldahl, FOI BLACK = hardbottom (Boulders, stones, pebbels) RED = soft bottoms
Bladderwreck Fucus radicans SOFT BOTTOMS
LIDAR SIGNAL DATA Bladderwrack (Fucussp.) is a key ecological species in the Baltic, and it s distribution can be modelled fairly accurately if there is enough input information available (left). There is however a huge difference when the modelled results are compared to the data extracted from LIDAR (which shows the actual real-life distribution). 200 m 200 m
ACCURACY OF THE METHOD
CLASSIFICATION OF LIDAR DATA
LIMITATIONS AND PROBLEMS -Limitations connected to water clarity or colour -Limitations in depth distribution
LIDAR-COVERAGE About 50 % of the water area was lacking LIDAR-data High turbidity or brown water Deeper than 15 m
Turbid or brown water
Problems in real shallow areas
Results in real shallow areas Real shallow areas Waves Land or water (?)
CONCLUSIONS The LIDAR method is very fast: 20 km 2 per hour with 2x2 meter resolution(in optimal conditions) compared to 10 km 2 per summer with 100x100 meter resolution for a 3-person field team (doing underwater video + diving utilizing boats). The new LIDAR inventory yields accurate depths as well as a rough but reliable classification of the natural values on the seafloor, but only down to 2,5-3 x Secchi-depth (12 18 meters in the Baltic). A reliable correlation of the LIDAR signal to seafloor geology and vegetation requires extensive groundtruthing, but the classification can then be extended to very large areas (thousands of km 2 ).
UMEÅ Umeå VAASA 10km
Områdesavgränsning
Resultat, Naturvärdeskartor Höga kärlväxter Kransalger
Hårdbotten -Mjukbotten
CLASSES
How good are the models? Sjökort Digitaliserade djupmätningar Lidar (batymetri) Art/artgrupp Prediktionens kvalitet Prediktorvaria bler 1 Prediktionens kvalitet Prediktorvaria bler 1 Prediktionens kvalitet Prediktorvaria bler 1 Cladophora 25% täckning Intermediär (AUC 0,771) Vågexponering, djup, siktdjup Intermediär (AUC 0,785) Vågexponering, djup, siktdjup God (AUC 0,814) Djup, vågexponering Höga kärlväxter 2 5% täckning God (AUC 0,887) Djup, vågexponering, siktdjup Intermediär (AUC 0,755) Djup, vågexponering, ljusexponering, kurvatur, lutning Intermediär (AUC 0,757) Djup, vågexponering, Chara spp. 10% Intermediär (AUC 0,798) Djup God (AUC 0,839) Djup, ljusexponering, siktdjup Intermediär (AUC 0,778) Djup, kurvatur Potamogeton perfoliatus 5% - - Dålig (AUC 0,680) Djup, vågexponering, ljusexponering, siktdjup, kurvatur Dålig (AUC 0,689) Djup, vågexponering, siktdjup Hårdbotten 25% täckning Dålig (AUC 0,553) Vågexponering, djup God (AUC 0,806) Vågexponering, djup, kurvatur, lutning Intermediär (AUC 0,746) Vågexponering, djup Mjukbotten 25% täckning God (AUC 0,898) Vågexponering, djup God (AUC 0,853) Vågexponering, djup God (AUC 0,842) Vågexponering, djup
Validering av kartor Jämförelse mellan modellerat data från 2009 och externt dataset från 2007