Research Data for "Comparison of detectability of ship wake components between C-Band and X-Band synthetic aperture radar sensors operating under different slant ranges"
Description
This research data repository contains the datasets for publication "Comparison of detectability of ship wake components between C-Band and X-Band synthetic aperture radar sensors operating under different slant ranges" in ISPRS Journal of Photogrammetry and Remote Sensing. For each sensor and each wake component one data file is added to the uploaded .zip-archive. The sensors TerraSAR-X, CosmoSkymed, Sentinel-1 and RADARSAT-2 are abbreviated tsx, csk, s1 and rs2, respectively. The wake components near-hull turbulence, turbulent wake, Kelvin wake arm and V-narrow wake are abbreviated nt, tw, kw and vw, respectively. The result is a collection of 16 main datasets. For uncertainty estimation each of the main datasets is split into five data subsets. The subsets extend the file name of the MAIN dataset by MIX1, MIX2, MIX3, MIX4 and MIX5, respectively. The results are a collection of 80 data subsets. In total 96 dataset are available. The format of the datasets follows the .arff format, which is readable by the Data Science Tool “Weka” (https://www.cs.waikato.ac.nz/ml/weka/index.html) and also by human using any text editor. Each data row defines one candidate wake sample. The columns define the five influencing parameters and the wake component length of the respective dataset by: Wind speed, incidence angle, ship velocity, ship length, 90°-projected relative ship heading, wake component length
Files
Steps to reproduce
1. Intersection of SAR images with data from the Automatic Identification System (AIS) to locate all AIS reported ship positions in the SAR images 2. Identification of candidate wake samples, i.e. all reported positions with minimum reported vessel velocity of 1 m/s 3. Filtering of all positive identifications containing SAR-artifacts or -anomalies not related to ship movements 4. Pre-processing of SAR images, i.e. radiometric calibration into NRCS, filtering of land and maritime objects and rescaling into uniform pixel spacing of 1.5 m per pixel 5. Cutting-out subimages from pre-processed SAR images with dimension of 5100 m in Azimuth and Range (3400 pixels each dimension) 6. Manual tracing of the four wake components (near-hull turbulence, turbulent wake, Kelvin wake arms, V-narrow wakes) in the subimages by mouse cursor to build tracing vectors, which define the shape of each wake component by multiple image coordinates 7. Calculation of wake component lengths from the tracing vectors (zero means respective wake component is not detectable in the SAR image)