Polarimetric Radar Analyses And Monitoring Of Sediments And Habitats On Exposed Intertidal Flats

Intertidal flats are the coastal areas that fall dry once during each tidal cycle.Those are the intermediate areas between the land and sea,and play a critical role in global ecosystem.Therefore,frequent and accurate surveillance of such vulnerable ecosystems is one of the most important approaches in ocean remote sensing.However,the time-window for the intertidal flats observation is only limited within one hour before and after the low tide;existing monitoring techniques vary with different polarization modes;the radar backscattering mechanisms of sediments and habitats on exposed intertidal flats are not yet fully understood.Due to these difficulties and challenges,we conducted with detailed analyses of radar backscattering mechanisms of typical targets on intertidal flats,demonstrated the influence of imaging geometry?radar frequency and incidence angle?and environmental conditions?wind speed,water level,and vegetation period?on the radar return,and proposed a detection and classification chain for sediments and habitats on intertidal flats.The main conclusions are as follows.?1?The conditions of intertidal sediments vary from time to time.Thus,it is difficult to objectively describe their backscattering characteristics when we only use a few SAR images.In this dissertation,we used a large number of SAR images for the following statistical analyses.Our results show that several parameters have to be taken into consideration when SAR images of intertidal flats are interpreted.First,the local incidence angle has a strong influence on the strength of the radar return.Steep incidence angles are recommended in polarization analyses of sediments,while shallow angles make the lower backscattered radar signal being merged into the noise floor.Second,local wind speed has only little impact on the radar backscattering from exposed intertidal flats.With the increasing wind speed,the sediment moisture decreased,leading to a transition of backscattering signals from weaker,mixed?odd-and even-bounce?backscattering to stronger,single-bounce backscattering.Remnant water may be roughened by the wind,and the wind-induced sand waves may increase the surface roughness,which both effect the overall backscattered signals.Third,water level differences of only a few decimeters,or even less,may also results in different soil moisture and a different amount of remnant water on sand and mud flats.As the water level increases,the even-bounce components decreases obviously and their difference with the odd-bounce relatively falls down.?2?The intertidal vegetation,such as sea grass,also have influence on local sediment distribution.Our results indicate that sea grass enhances the radar backscattering,because of increased local surface roughness,even though the grass may lay flat on the ground at low tide.Moderate wind speed has no significant influence on the radar return of sea grass meadows,so does water level and moisture.Our SAR dataset acquired during the vegetation period?from May to October?shows that sea grass impacts the radar return.Its backscattering signals at HH polarization becomes stronger,and the same as the increasing even-bounce scattering.Meanwhile,bivalve beds own spatially complex and rough surface structures.The heterogeneous surface structure of bivalves causes a strong spatial variability of even-and odd-bounce backscattering,which is different from the surroundings.?3?Based on the fully understanding of the backscattered radar return of intertidal sediments and habitats,we proposed FCD-RF classification chain.The Freeman-Durden and Cloude-Pottier components are chosen,further with DERD parameter,followed by the Radom Forest classifier.We demonstrated its higher discrimination accuracy of mud and sand flats based on X,C and L band SAR images.The common used intensity channels own weaker abilities to distinguish mud and sand flats on intertidal flats.?4?In order to get rid of the dependence of quad-polarization SAR images in polarization processing schemes,this dissertation proposed an algorithm based on dual-copolarization?HH and VV?for the detection of bivalve beds on exposed intertidal flats.Our results found that dual-copolarization SAR images can also be used in monitoring of habitats in coastal areas.Among various SAR indicators,the real and imaginary part of the correlation coefficient of both channels have the ability to describe their backscattering attributes of oysters or blue mussels.We introduced the indicators D₃and D₇ for further demonstration,and found their best performance in shorter band.Although the weaker efficiency in longer band,D₃ and D₇ always give better availability for the detection of bivalve beds,which satisfying with the requirements of frequent and accurate monitoring of exposed intertidal flats.The main scientific contributions in this dissertation are summarized as follows.?1?The radar backscattering characteristics of sediments and habitats on intertidal flats are fully understood.Based on a number of polarimetirc SAR images,we consider various imaging and environmental conditions of intertidal flats targets.We then are able to explain the corresponding backscatterring mechanisms,and break through the bandage of the narrow observation time-windows for intertidal observations.?2?The response of different environmental conditions and SAR imaging geometry on the polarimetric radar returns are more specific.We take the advantages of our abundant SAR datasets for detailed analyses.We explain how the radar frequency,radar incidence angle,coastal wind speed,water level and seasonal change,can influence the radar backscattered return.?3?The detection and classification accuracies of intertidal targets are obviously increased.We proposed FCD-RF classification scheme for mud and sand flats as well as the detection indicators for bivalve beds,and demonstrated their availabilities in X,C,and L band.Meanwhile,we extend our research on the developing neutral network techniques into the monitoring of intertidal flats with better results.?4?The dependence of the common used quad-polarization channels is broken through.We used the Kennaugh element framework to deeply exploit the potentials of dual-polarization modes for the monitoring of sediments and habitats on exposed intertidal flats.Results approved their abilities.