sst bad result near coast

[redhat007]

Wind stress, surface net heat flux, surface net fresh water flux, net solar radiation flux, are applied force files in my case. Then, SST and dqdsst (base on formulation in Barnier et al 1995) have been added to get better model's SST. BUT now model's SST has problem near the shore line (area in elliptic). It is strange that in that shallower area, model's SST is less than middle area(especially in summer months ). I can't recognize that input files cause this error or it needs to active another option in header files what is your comment for me? using SST from another data base, or using larger time interval for SST nudging ( i have used 3 hourly)?
I have attached June and July model's SST and also some input data to model (such as shflux, sst and dqdsst) for both months.
Many thanks for your attention to help me
P.S
All input data provided form ECMWF_Interim in forecast version with 3 hourly time interval. Also applying its SST in reanalysis version (with 6 hourly time interval) cant solve above problem, too.

[kate]

This looks to me like you are interpolating zero values from the ECMWF land mask into your coastal ocean. You instead need to be extrapolating ocean values into the land areas before you do any interpolation. ROMS can handle the interpolation, but not the extrapolation - this should be done as a preprocessor step in the tool of your choice. There is a sosie package which someone helped me get going with to do the extrapolations for the MERRA fields. I'm sure others do this in Matlab.

[ymamoutos]

Hi Kate

Sorry for writing to this old post but I don't think
there is a point to start a new thread. When you
using sosie to extrapolate MERRA2 dataset are you using
east-west periodicity (ewper flag) for the input grid?
I am searching to MERRA2 site and manual to find information
about MERRA2 grid but I didn't have any luck so far.

Thanks in advance

Giannis

[kate]

Raf Dussin set me up with sosie. Can you tell from this if it's periodic?

Code: Select all

#!/bin/bash

SOSIE=/u1/uaf/kate/bin
# Note - this is my Archive copy now
dirin=/archive/u1/uaf/kate/MERRA/Forcings
dirout=/center/d/kate/MERRA_drowned

year="2016"
ybeg=2016
yend=2017

# build the list of variables to proceed
list_var=$( ls $dirin | grep $year | sed -e "s/MERRA_//" -e "s/_$year.nc//" | awk '{ print $1 }' )

for year in $( seq $ybeg $yend ) ; do

    for var in $list_var ; do

        filein=MERRA_${var}_$year.nc
        timevar=$( ncdump -h $dirin/$filein | grep time | head -1 | awk '{ print $1 }' )
        ncvar=$( ncdump -h $dirin/$filein | grep float | tail -1 | sed -e "s/(/ /" | awk '{ print $2 }' )

        echo Processing variable $ncvar in file $filein

        $SOSIE/mask_drown_field_roms.x -D -i $dirin/$filein -v $ncvar -t $timevar \
                                     -m /center/d/kate/MERRA/mask_MERRA.nc -o $dirout/drowned_$filein



    done

done

[ymamoutos]

Thanks a million Kate. From the above script I understand that you
use the default option - when you don't set explicitly -p (periodicity)
flag - which is east-west periodicity with zero (0) overlapping points.

Giannis

[kate]

It probably should be periodic - can you try it both ways?

[ymamoutos]

Greetings Kate and sorry for my late reply.
Yes it is possible to run it both ways.
Currently I am using the default option. I'll
probably use the non periodic option to check if
there are any significant differences to my results.
I don't know if it's right but my domain is
relatively small and I want to use MERRA2 data
because of it's high temporal resolution.

Thanks again for your help.
Kind Regards

Giannis