add PACE-OBIS notebook#328
Conversation
There was a problem hiding this comment.
Could probably skip all of cell[3] to save some STDOUT printing. That was just my testing of earthaccess.search_datasets(). cell[4] does the query we're interested in.
After reviewing cell[4] I recall this was a single time instance query. Is it possible to collect all time instances and generate an average? Then compare that to what we actually observe in OBIS (ie. the pyobis response). Then we can ask the question if the satellite observations match in-situ observations.
There was a problem hiding this comment.
Increasing the time span of the pace request can be quite computational and "bandwith" and heavy. While I agree that for a fair comparison we would need a larger time Windows, the comparison is already qualitative and I don't think it is worth adding that cost to the request.
With that said, we can try to fetch at least 1 year so we can have all seasons and increase the quality of the comparison. How does that sound?
PS: 1 year is ~23.76 G !
There was a problem hiding this comment.
Holy smokes! 24GB! Not worth it. This is a proof of concept. Let's keep it to that one time instance.
There was a problem hiding this comment.
It is still a lot, but I noticed that we have 1 dreegee and 4 km resolutions in the same response. We can filter for the 1 def only.
The question now is, if I understand it correctly, OBIS response is number of observations while PACE is Chlorophyll concentration. How do we compare those? My suggestion would be to normalize each one so both end up between 0-1 and plot the heatmap. Would that work?
There was a problem hiding this comment.
Yeah, we can normalize them to 0-1. We're just using chl-a as a proxy for surface phytoplankton concentration/abundance.
It is still a lot, but I noticed that we have 1 dreegee and 4 km resolutions in the same response. We can filter for the 1 def only.
Yeah! I haven't fully explored the earthaccess.search_data() response to see what all we can work with. I found the first chl-a dataset and ran with that. If there are other datasets that would be more useful/performant, lets use those.
ocefpaf
force-pushed
the
add_pace_obis_notebook
branch
from
bf0b5d7 to
65687ed
Compare
2 participants
Closes #317