Extraction of coastal ocean wave characteristic parameters using optical remote sensing and computer vision technologies

Johnson, Michael Christopher (2018-03)

Thesis (MSc)--Stellenbosch University, 2018.

Thesis

ENGLISH ABSTRACT: Ocean waves are among the most important forces shaping the world’s coastlines. They drive environmental processes and human activity that occurs within the coastal zone as well as on the open ocean. The assessment of wave characteristics, such as wave direction, wavelength, wave period and wave velocity, are critical in understanding coastal processes that serve as a baseline for better coastal management. However, the monitoring and assessment of wave characteristics is challenging, given the high complexity of ocean dynamics and large spatial extent. While globally the development of ocean state forecasting models has been fruitful in recent decades, these capabilities are only just being developed in (South) Africa and require ground truth data for model development and validation. The work presented here assesses whether optical imagery from the RapidEye satellite can be used to extract ocean wave characteristic parameters such as wave direction, wavelength, wave period and wave velocity for the purpose of further developing and validating wave forecasting models. Two techniques were adopted. The first made use of a Fourier transform to extract the directional wave spectrum. The characteristic wave parameters were then calculated from the spectrum. The second approach made use of normalised cross correlation (NCC) to extract the phase velocity field. The techniques identified and developed were tested using RapidEye imagery on four study areas on the South African coast. These included Richards Bay, Durban, East London and Cape Point. Ground truth data in the form of wave measurements captured by wave buoys were used for validating the results. The results from the Fourier transform show a generally high agreement with the directional spectra derived from the wave buoys. In the context of characteristic parameters extracted from these spectra, two of the four study sites produced highly accurate results with all parameters within 10% deviation of the wave buoy data. The results obtained from the NCC revealed that the shoaling of ocean waves has a positive influence on the reliability of measurements near the shoreline. The directions for two of the study areas had differences of less than 10° from the wave buoy directions. It was also observed that when waves are present with vastly different directions the NCC approach fails. These results show a promising advance in the use of optical imagery for the monitoring and assessment of near-shore wave conditions. However, it is recommended that further development and validation of these techniques be undertaken before operationalization.

AFRIKAANSE OPSOMMING: Oseaangolwe is een van die belangrikste kragte wat die wêreld se kuslyne vorm. Hulle bepaal omgewingsprosesse en menslike aktiwiteite wat binne die kussone, asook in die oop see omgewing, plaasvind. Die evaluering van golfeienskappe, soos golfrigting, golflengte, golfperiode en golfsnelheid, is van kritieke belang om kusprosesse te verstaan as 'n basis vir beter kusbestuur. Monitering en assessering van golfkenmerke is egter uitdagend, gegewe die hoë kompleksiteit van die oseane-dinamika asook die groot ruimtelike omvang. Terwyl die ontwikkeling van modelle vir oseaanstaatvooruitsigte in die afgelope dekades wêreldwyd baie suksesvol gebruik word, word hierdie vermoëns nou eers in (Suid) Afrika ontwikkel en daarom word grondwaarheidsdata vir modelontwikkeling en -verifikasie benodig. Die werk wat hier aangebied word, beoordeel of optiese beelde van die RapidEye-satelliet gebruik kan word om golfeienskappe soos golfrigting, -lengte, -periode en -snelheid te onttrek ten einde golfvoorspellingsmodelle verder te ontwikkel en te verifieer. Twee tegnieke is ondersoek. Die eerste het gebruik gemaak van 'n Fourier transform om die golfrigtingsspektrum te onttrek. Die kenmerkende golfparameters is dan uit die spektrum bereken. Die tweede benadering het gebruik gemaak van genormaliseerde kruis korrelasie om die fase snelheids veld te onttrek. Hierdie tegnieke wat geïdentifiseer en ontwikkel is, is getoets met behulp van RapidEye-beelde op vier studieareas aan die Suid-Afrikaanse kus. Dit sluit in Richardsbaai, Durban, Oos-Londen en Kaap Punt. Grondwaarheidsdata van golfmetings wat deur golfboeie vasgevang is, is vir die verifiering van die resultate gebruik. Die resultate van die Fourier transform toon 'n algemeen hoë ooreenkoms met die rigtingspektra wat van die golfboeie verkry is. In die konteks van kenmerkende parameters wat uit hierdie spektra onttrek is, het twee van die vier studiegebiede hoogs akkurate resultate met alle parameters binne 10% afwyking opgelewer. Die resultate wat verkry is uit die genormaliseerde kruis korrelasie het getoon dat wrywing van die oseaangolwe met die seebodem die betroubaarheid van metings naby die kuslyn laat toeneem. Die rigtings vir twee van die studiegebiede het met minder as 10° van die golfboeirigtings verskil. Dit is ook waargeneem dat hierdie benadering meer geneig om te misluk wanneer golwe teen baie verskillende rigtings teenwoordig is. Stellenbosch University https://scholar.sun.ac.za vi Hierdie resultate is belowend vir die gebruik van optiese beelde vir die monitering van golftoestand assessering naby die kus. Daar word egter aanbeveel dat verdere ontwikkeling en verifiering van hierdie tegnieke onderneem word voor operasionalisering onderneem word.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/103604
This item appears in the following collections: