Hello all, I am working on this custom script. It uses data fusion request (Sentinel 1 and Sentinel 2 data). This script focuses on Dakar and other African terrotries. Sentinel 1 has data access from 2015 onwards and Sentinel 2 has data access of this specific area from 2016 onwards. I need satellite images or multi temporal data access from year 2000 to onwards using the same cusotm script. Can you please guide me how can I achieve desired result?
Thanks
Hi,
The custom script you are referring to requires SAR (used to extract buildings and water bodies) and optical data (to obtain information about bare soil, sand, and vegetation).
To replace Sentinel-2, you could use Landsat 7, available since April 1999 (although in 2003 it started suffering from banding issues) or maybe MERIS/MODIS at lower resolution.
For the SAR data it’s more tricky. There are a couple of sources that are open, but not currently supported as default options in Sentinel Hub services. Nevertheless, you can easily ingest the data using our BYOC API.
A couple of places to search for data are the ASF website, ESA, the earthdata catalog…
The alternative would be to try and replace the radar computations with optical-satellite derived products.
Hey, thanks for the reply. I will look into the given suggestions.
Thanks again
hey, could you please tell me about the best band combination of Landsat 7 to extract urban area for the Dakar Region.
Thanks
Looking at this custom script, “all […] pixels with B11 values greater than 0.8 OR NDVI values lower than 0.1 [are considered] to be built up”, so transferring the script to Landsat 7, you could calculate:
-
NDVI = (B04 - B03) / (B04 + B03)
-
Use B06 for Landsat instead of B11 for Sentinel-2
-
Play with the thresholds to fit your use-case
You could also consider using NDBI (Normalized Difference Built-up Index), as mentioned here.
This is just one idea, there are surely many other approaches out there that you could apply.