Gap-filled NDVI timeseries to xarray

[singhamninder]

Hi,
Thanks for putting together this excellent package. I find it very useful for a lot of my workflows.
I have exported an ImageCollection to assets, consisting of gap-filled smooth NDVI time-series following this tutorial. Then I am using the following code to create a xarray -

chico_ts = wxee.TimeSeries("users/amninder/chico_ic") \
            .sort("system:time_start") \
            .filterDate("2018-01", "2018-03") \
            .select(["smoothed"])

ts_arr = chico_ts.wx.to_xarray(region=polygon)

However, I keep getting nan values for data variables. Any thoughts on how I can fix that?

[aazuspan]

Hi @singhamninder, glad you're finding wxee useful!

Are you sure the NaN pixels you're getting aren't masked in Earth Engine? to_xarray assigns NaNs to masked pixels, so maybe something is going wrong with the gap filling and those pixels have no data.

If you haven't already, I would try checking those images without wxee, either by looking at them with geemap or downloading them manually with ee.batch.Export.image.toDrive. If those pixels do have valid data and it gets lost by to_xarray, I'll definitely look into that!

Also, if you want to post your entire code (and make the assets public) I can take a closer look.

EDIT: Adding the answer from below to the top:

Okay, you just need to set the scale parameter. This should work!

ts_arr = chico_ts.wx.to_xarray(region=polygon, scale=10)

If you don't specify a scale, Earth Engine tries to guess based on the image projection. It definitely got it wrong here, which is why it just gave you 1 pixel with no data.

[aazuspan]

Thanks for the details, and for confirming that the data isn't masked. I'd like to take a closer look and see if I can recreate the issue, but I'm not able to access the asset you linked. Could you make sure it's shared? If you click on the asset in the Code Editor there's a "Share" tab in the top right that you can set to "Anyone can read".

[singhamninder]

Oops sorry about that. Just shared it

[aazuspan]

Okay, you just need to set the scale parameter. This should work!

ts_arr = chico_ts.wx.to_xarray(region=polygon, scale=10)

If you don't specify a scale, Earth Engine tries to guess based on the image projection. It definitely got it wrong here, which is why it just gave you 1 pixel with no data.

[singhamninder]

It works! Thank you so much for your time :)

[aazuspan]

No problem!