Data Search seems to be off by one day

[woodbri]

I just noticed that while searching for files in a subscription that the filenames returned are one day behind the requested dates. This seems to hold true for all queries using the web api also.

For example, when I request files for 2024-07-10, I get files for 2024-07-09. Am I doing something wrong? Is this a timezone issue? I'm confused.

Code: Select all

wget -q --post-data="results_as_file=1&sdate=2024-07-10 00:00:00&edate=2024-07-10 23:59:59&subID=2462&subType=1" -O - https://oceandata.sci.gsfc.nasa.gov/api/file_search
JPSS1_VIIRS.20240709T163600.GEO.nc
JPSS1_VIIRS.20240709T163600.L1A.nc
JPSS1_VIIRS.20240709T181200.GEO.nc
JPSS1_VIIRS.20240709T181200.L1A.nc
JPSS1_VIIRS.20240709T181800.GEO.nc
JPSS1_VIIRS.20240709T181800.L1A.nc
JPSS1_VIIRS.20240709T195400.GEO.nc
JPSS1_VIIRS.20240709T195400.L1A.nc
JPSS1_VIIRS.20240709T200000.GEO.nc
JPSS1_VIIRS.20240709T200000.L1A.nc
JPSS1_VIIRS.20240710T161800.GEO.nc
JPSS1_VIIRS.20240710T161800.L1A.nc
JPSS1_VIIRS.20240710T175400.GEO.nc
JPSS1_VIIRS.20240710T175400.L1A.nc

[OB.DAACx - amscott]

Hi Woodbri,

I was able to reproduce your results. The issue has been submitted as a bug. We will be in touch with an update as this gets resolved.

[OB Web - jdawson19]

Hey there, sorry for the long delay since the last comment here. We now understand what's causing this--essentially, the start date provided is actually used to compare against the "staged date" for subscription files, rather than the more usual "start date".

We're now debating whether to swap over to the usual method--that is, so that sdate always indicates start date, regardless of whether the user is querying for subscription files or not--or to preserve the old functionality. However, this is not a bug, it's just a feature that is not explained well by the page--we will either add some text or get rid of it in the future.

[oo_processing]

Greeting to everyone,

As a big user of subscriptions (200+), I prefer the current way, no change. I don't need to manually adjust the backward days (for the start date) every time after a pause of data staging (in the current way, I can see a huge inflow of data right after the restore of data staging, the I know I don't miss any data). In this way, I can set short and fixed backward days for getting a reasonable amount of search results to sort. And I will have a predictable running time of my search & download program. I personally like the current way.

Thanks
Yuyuan