[ecco-support] About raw uncertainty fields of v4r4

[Wang, Ou (US 329B)]

Hi Yongsu,

One way to obtain raw uncertainties for both data constraints and controls is to remove the grid scaling factor (gamma) from the uncertainties actually used in V4r4’s optimization. Gamma is the square root of fractional grid area defined as
gamma(I,j) = sqrt(A(I,j))/max(A(i,j)), where i,j are grid x- and y-index; A is grid cell area (as in RAC.data from model outpu); sqrt is square-root; max is maximum. Raw uncertainties (e_raw) then can be derived from uncertainties (e) that are actually used in optimization (such as those specified in data.ecco) by using the following equation,
e_raw = e*gamma.
The actual uncertainties (e) are distributed as part of V4r4’s ancillary data https://podaac.jpl.nasa.gov/dataset/ECCO_L4_ANCILLARY_DATA_V4R4.

The raw uncertainties were originally produced by different from a few institutions. That information can be found in V4r4’s synopsis document (https://ecco-group.org/docs/v4r4_synopsis.pdf). Since some of the raw uncertainties for V4r4 are the same as those for V4r3, additional information can be found in V4r3’s synopsis document (https://ecco.jpl.nasa.gov/drive/files/Version4/Release3/doc/v4r3_summary.pdf) and data document (https://ecco.jpl.nasa.gov/drive/files/Version4/Release3/doc/v4r3_data.pdf). In particular, V4r3’s synopsis describes the rationale behind splitting costs to time-mean & time-varying parts and applying area-scaling factor, as well as how to derive actual uncertainties used in optimization from raw uncertainties.

For next release (V4r5), we will include instructions about how to generate uncertainties used in optimization from raw uncertainties.

Best,
Ou Wang

From: ecco-support <ecco-support-bounces at mit.edu> on behalf of NA Yongsu <ynaab at connect.ust.hk>
Date: Tuesday, November 15, 2022 at 12:28 PM
To: ecco-support at mit.edu <ecco-support at mit.edu>
Cc: atnguyen at oden.utexas.edu <atnguyen at oden.utexas.edu>, jclmak <jclmak at ust.hk>
Subject: [EXTERNAL] [ecco-support] About raw uncertainty fields of v4r4
Dear ECCO team,

I'm currently setting up a regional state estimate based on ECCOv4r4. I wonder if there is a way to download raw uncertainties (in lat/lon) of target observations (e.g. satellite) and controls of both oceanic initial conditions and atmospheric forcings (i.e. from Chaudhuri et al. 2013).

Also, it would be great if there is a direction on how to process (e.g. split into time-mean/time-varying components, apply area-weighted scaling while mapping, etc.) the raw files to make the files that have been distributed in v4r4. Thanks in advance!

Best wishes,
Yongsu Na

-----
PhD student
Department of Ocean Science
Hong Kong University of Science and Technology
-----
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mit.edu/pipermail/ecco-support/attachments/20221116/5bbe7b47/attachment-0001.htm>