Missing data in chl.nc at 0.25° and 0.1°

[aekiss]

The 0.25° and 0.1° chl.nc chlorophyll input files are missing data at in the far north and south (the 1° data looks OK). This will alter the depth range over which shortwave radiation heats the ocean. Hopefully this isn't upsetting the bottom water formation?

Screenshots below are

/g/data/ik11/inputs/access-om2/input_20201102/mom_1deg/chl.nc
/g/data/ik11/inputs/access-om2/input_20201102/mom_025deg/chl.nc
/g/data/ik11/inputs/access-om2/input_20201102/mom_1deg/chl.nc

Thanks to Yinghuan Xie for alerting me to this issue.

[aekiss]

The origin and history of these files is unclear to me. Can anyone (@russfiedler @PaulSpence) suggest where they came from?

[aekiss]

The metadata doesn't reveal much. They were last modified in 2017, 2011, 2013 (1°, 0.25°, 0.1°, respectively). The 0.25° seems to be derived from seawifs_1998-2006.nc (does anyone recall what this was?), and the NaNs were introduced at 0.25° and then propagated to 0.1° which seems to have been interpolated from it, as the NaN patches have the same shape as at 0.25° are more jagged than the 0.1° grid. However, the 0.1° has an extra band of NaNs at the southern edge, in a latitude range which is not NaN at 0.25°. E.g. here's a zoom into 0.25° at 2:1 scale and 0.1° at 1:1 scale:

I had some guesswork in sec 3.5 of the draft tech report

We use a prescribed monthly surface chlorophyll-a climatology (read_chl=true) from the file chl.nc. This climatology is based on SeaWiFS data from 1998– 2006 TODO: check: Griffies (2015, sec 3.14.2) says 1998–2007 and is presumably TODO: check the same as used in GFDL’s CM2.5 and CM2.6 (Delworth et al., 2012; Griffies et al., 2015) based on the method of Sweeney et al. (2005).

[anton-seaice]

This looks like it was made from a remote sensing product, e.g. SeaWifs as suggested and the gaps could be where the sensor is blocked by clouds / ice / lack of sunlight. Chlorohpyl relies on optical (possibly infrared) remote sensing.

But as the NaNs are reliatively uniform its more likely some sort of reprojection / regrid / mask error?

We could make a new climatology, Ocean Colour V6 might be a good product to choose, (I think they claim 5km resolution) and it should have a longer timeframe for data (1997-2024 or so)

[aekiss]

Martin says

The UM generates its ocean color ancillary file from ~access/umdir/ancil/atmos/master/sea_clim/GlobColour/v2/qrparm.sea.nc
which says the source is http://www.globcolour.info/