Mostly white images when downloading large areas


#1

Hi, I’m a new sentinelhub user and I’m running into some problems. What I would like to achieve:

In my Python script, given 4 latitude coordinates (longitude and latitude of top left and bottom right corner), I want to download images within this area. However, this may be a large area such as the whole USA.

I followed the guide for Large Area Utilities, and my approach is to create a Polygon object with the 4 given coordinates, then use:

bbox_splitter = BBoxSplitter([my_polygon], CRS.WGS84, (10,10), reduce_bbox_sizes=True)

To split the area into BBox’s, before iterating over each bbox and downloading each image. With the time argument being set as “latest” and a max cloud coverage of 0.5.

However, this results in most images being white, regardless of the max cloud coverage I use. From the FAQ it says that whenever an image has a cutoff with whitespace it means that Sentinel does not have data for the area. There’s no way this is the case as stitching the images together shows over 90% is blank.

What am I doing wrong? Is it the way I configured my BBox splitter?

Any help is greatly appreciated.


#2

There are a couple of possibilities of what could go wrong:

  • Note that each orbit is only acquired once every five days. If you ask for data over large area for only one day, 4/5 of the area will probably not be covered. Try expanding the time period, e.g. to 5 days.
  • You are using trial account and exceeding the limit of 10.000 requests per day so you are getting 429 errors.

I suggest you start with something smaller and get it working there and only later expand to larger area.
If you have some further issues, you will have to provide some more details of what is going wrong.


#3

Hi,

Thanks for the reply. I always use the “latest” argument for the images, shouldn’t this automatically find the latest time that the data is available?

Also, I will test out the theory that the limit is causing the problem, but I haven’t seen 429 errors yet when running the script.


#4

hi @gmilcinski

I think I isolated the root of the problem. It seems to be the grid parameters I pass in to the bbox splitter. If i make the grid more refined eg (40,40) instead of (10,10) for a large area, then the images contain much less “white space” and are all generally almost complete.

However, since the area I’m interested is large, this would result in a LOT of images. Is there a way I could download images from a higher distance to overcome this? Or a way to change the Bbox Splitter?