We mainly focused on optical remote sensing. We know that optical remote sensing images have been widely used for land cover classification. We noticed that some researchers conducted their study using Sentinel-1 SAR (Synthetic Aperture Radar) images. Even though optical remote sensing data are often contaminated by cloud and thus no many data can be used. However, most land cover types do not changed within short time. Then, when can we choose Sentinel-1 data used for classification topics? In what case, it is helpful and feasible to use it for classification ?
It makes sense to use such a data when you are dealing with risk assessment/disaster management applications. If you need near-real time information from the area of interest. As you mentioned such data is then not dependent on poor weather conditions, clouds, snow, sun illumination conditions, etc. So it can as well be used for land cover monitoring to support rapid mapping of land cover for decision makers.
Thank you. I see. Maybe we will use the data as supplement of optical remote sensing data due to cloud contamination and atmospheric correction error. For land cover mapping, is it recommended to convert the raw data into back scattering coefficients?@Sara Salehi
yes Jiang, I suggest you read these two papers:
Conference Paper LAND COVER MAPPING USING SENTINEL-1 SAR DATA
https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLI-B7/757/2016/isprs-archives-XLI-B7-757-2016.pdf
I think the answer is yes. The main problem is the spackle noise. If you could remove the noise as possible, the outcomes will be satisfacory.
Jingzhe Wang that will affect the spatial boundary between different classes. solution: adding spatial-contextual information. You can use markov random field model and account for such info, also it is possible to extract edge info separately and add that to the model:
Article Unsupervised change detection based on improved Markov rando...
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing