retrieve_images() keep on hanging

[dewisa29]

First of all, thank you very much for developing CoastSat, it has been extremely helpful for my coastal research.

I am currently setting up CoastSat (v3.0 and v3.2) on Windows for a shoreline change study, and I am encountering an issue during the image retrieval stage. I was hoping to ask for your advice on proper troubleshooting steps.

Specifically, the function SDS_download.retrieve_images() appears to hang for a long time without returning an error, while check_images_available() runs successfully. I initially suspected issues related to satellite selection, polygon size, internet connection, new environment, or CoastSat version, but the problem persisted.

My questions are:

1. Is this a known behavior when Earth Engine project permissions are misconfigured?

2. Are there recommended troubleshooting steps you suggest when retrieve_images() hangs without error?

3. Would you recommend always explicitly passing the project ID to ee.Initialize() when using CoastSat?

I would greatly appreciate any guidance you may have. Thank you again for making CoastSat openly available and for your time.

[kvos]

3. Yes explicitely pass a project ID to ee.Initialize().

Also you can get the prints of the errors by going into SDS_download.py and looking at the infinite loops inside while True:.
Let me know if you need further help, but I think the point above should solve the problem

[TobyATucker]

@kvos I have come across this issue too. For me it's in the get_image_info() function. When downloading S2 imagery for a long time period (e.g. 40 years), it hits a limit "EEException: Collection query aborted after accumulating over 5000 elements."

[kvos]

that's strange as Sentinel-2 only starts in 2015. Can you check that your polygon was provided in the right coordinates? if you have a very large area, maybe this happens, but I have never seen it before.

[TobyATucker]

You're right - I had the input set for a larger date range as I was using landsat too but this wasn't the issue. Could it potentially be the duplicated images in different UTM projections you mentioned in the code making the querry to large?

I was downloading a 1km x 1km box (i changed it a bit to run InletTracker). I tried adjusting this down and got to 50m x 50m and still had the issue. When I adjusted the dates to be just 7 years (instead of 10) the code worked. For 7 years of data, len(im_list) = 4519 so below the 5000 limit.

This is the polygon where I am having the issues (south of Spain):
[[[-5.6342668283496815, 36.01689351678295],
[-5.6342668283496815, 36.03832420609659],
[-5.60985449225926, 36.03832420609659],
[-5.60985449225926, 36.01689351678295],
[-5.6342668283496815, 36.01689351678295]]]

[kvos]

hey Toby, sorry for the late answer. I could reproduce the error.
I would say you can reduce the number of items by fixing the S2tile. You're probably across one of the regions with maximum overlap between two UTM zones. So you can get the layer of S2tiles and then in QGIS find the one that you want to use and input it in CoastSat, for example inputs['S2tile'] = '56HLH'. This should reduce the number of items. Otherwise you need to break it down in smaller time windows and then merge the outputs.