Which image(sensor) and method(technique) can result in more accurate results?
Hi Faisal,
With respect to the best data sources, high-resolution hyperspectral satellite images are indeed useful and valuable; but aerial photographs are also very good in shoreline change detection, as well as LIDAR data.
Regarding the methods, they vary and depend on the available data as well as on the shoreline indicator (dune toe, vegetation line, HWL, etc) you choose as proxy. I guess the automated (detection and extraction is automatic) approach should give more reliable results, especially when you quantitatively evaluate the robustness of your findings, based on statistical analysis (RMSE, Cross validation, etc.).
Here you will find a very good review dealing with shoreline detection and analysis (data sources, detection and extraction methods, etc..).
Cheers
A high or very high-resolution image (< 1m) with an infrared band is ideal to separate water from land. Watershed segmentation seems to be a very effective algorithm to classify shorelines.
PL YOU TAKE FIRST TOPOSHEET MAP (1:50000) OR (1:25000) WITH LATEST HIGH RESOLUTION SATELLITE IMAGERY OF THAT AREA AFTER RAINFALL SEASONS . REGISTERED BOTH IN ARC GIS MAP OR MAPINFO . OVERLAP EACH OTHER YOU WILL GET CLEAR CUT CHANGES AREA OF SHORELINE CHANGES. THIS BEST AND ACCURATE METHOD. FALLOW IT FOR YOUR STUDY.
As Javier said >1m resolution with an IR band, meaning that IKONOS for instance would be a good solution, or some of the SPOT5 imagery... otherwise you can mount an IR and normal camera on a drone to follow shoreline changes... doing this in New Zealand on rivers and lagoons.
The real answer depends upon the estimated amount of shoreline change for the are in question. One to two feet of shoreline change is not detectable within satellite imagery, whereas 30 to 200 feet of change would be. High resolution aerial imagery is best for electing and measuring small mounts of shoreline change. Also, the precision of the spatial geometry is potentially the most important parameter to be aware of. If the imagery spectral resolution is 1 meter or less, but the spatial resolution is 3 meters or more (as most satellite imagery is) then you can not be sure that your amount of shoreline change measured is than the spatial resolution. In fact, the estimated error would be double the spatial resolution, meaning the placement of your shoreline will be somewhere within six meters of its location in the image.
We need also chose with method you will use to extract the coastline and mesurement the changes. Some sector in the coast line change more that other. Your detection depend of spatial and spectral resolution of the imagery, but also of the dimension of your site for analysis. What you nedd to show. Be conscious of the error and measurement in relation the coastal coastal of you study.
If you are wanting to detect change in the shore line (that is the erosion and/or build up of the high water mark) then the most important characteristics of the image data, in my view, will be the spatial resolution (versus the spectral or temporal resolutions). If this is the main objective then a traditional CIR (visible with a nir band) will in most cases be sufficient BUT with AT LEAST 1m spatial resolution. I would suggest spatial resolutions in the range of 0.5m to 0.2m the best.
If the main objective is to detect, map, and monitor changes on land near or right next to the shore line then 0.5m to 1m resolution will be acceptable.
Pat
You use a satellite image
Free access data
ttps://glovis.usgs.gov/
(Landsat, Sentinal 2, Aster, IRS, Hyperian, SRTM etc.)
https://scihub.copernicus.eu/dhus/#/home (Sentinal 1, 2, 3)
https://eoportal.eumetsat.int/userMgmt/protected/welcome.faces (Sentinel all products)
and Best Commercial software for better analysis
IKONOS
Worldview
Cartosat 3
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining