Opened 4 years ago

Last modified 3 years ago

#258 assigned

APS1 & 2 city model land sea masks

Reported by: Martin Dix Owned by: Martin Dix
Priority: major Component: ACCESS model
Keywords: Cc: Wenming Lu, chs548, Peter Steinle, Yi Xiao, Imtiaz Dharssi, Milton Woods, Gary Dietachmayer

Description

The IGBP land cover data used to generate the land sea masks has position errors that show up in the 4 and 1.5 km models.

This ticket will investigate how best to either correct this or use alternate data sources.

Attachments (13)

CapeWilloughby_APS1_AN.png (31.9 KB) - added by Martin Dix 3 years ago.
APS1 orography and mask around Cape Willougby station
orog_9s_kangaroo_island.png (69.1 KB) - added by Martin Dix 3 years ago.
GA 9s orography for Kangaroo Is.
igbp_kangaroo_island.png (36.3 KB) - added by Martin Dix 3 years ago.
IGBP surface type for Kangaroo Is.
cci_kangaroo_island.png (48.2 KB) - added by Martin Dix 3 years ago.
lakes_entrance_GA.png (60.3 KB) - added by Martin Dix 3 years ago.
lakes_entrance_SRTM.png (135.9 KB) - added by Martin Dix 3 years ago.
sydney_GA.png (78.4 KB) - added by Martin Dix 3 years ago.
sydney_SRTM.png (200.2 KB) - added by Martin Dix 3 years ago.
wollongong_GA.png (58.0 KB) - added by Martin Dix 3 years ago.
wollongong_SRTM.png (146.1 KB) - added by Martin Dix 3 years ago.
lakes_entrance_IGBP.png (31.0 KB) - added by Martin Dix 3 years ago.
sydney_IGBP.png (50.2 KB) - added by Martin Dix 3 years ago.
wollongong_IGBP.png (33.3 KB) - added by Martin Dix 3 years ago.

Download all attachments as: .zip

Change History (21)

comment:1 Changed 4 years ago by Martin Dix

Owner: set to Martin Dix
Status: newassigned

comment:2 Changed 3 years ago by Martin Dix

Tomasz flagged problems with the APS1 AN land sea mask having ocean all around the Cape Willoughby station on Kangaroo Island. Maps shows the model mask and orography with the GA coastline.

APS1 orography and mask around Cape Willougby station

The ancillary program calculates the mask from the IGBP 1 km land surface classification dataset. This has displacement errors (https://code.metoffice.gov.uk/trac/TI/ticket/2) and worse the errors vary with location. Near Kangaroo Is. there seems to be an offset of about 2 km W and 1 km S.

IGBP surface type for Kangaroo Is.

By comparison there's no problem with the GA 250 m orography
GA 9s orography for Kangaroo Is.

Last edited 3 years ago by Martin Dix (previous) (diff)

Changed 3 years ago by Martin Dix

Attachment: CapeWilloughby_APS1_AN.png added

APS1 orography and mask around Cape Willougby station

Changed 3 years ago by Martin Dix

Attachment: orog_9s_kangaroo_island.png added

GA 9s orography for Kangaroo Is.

Changed 3 years ago by Martin Dix

Attachment: igbp_kangaroo_island.png added

IGBP surface type for Kangaroo Is.

comment:3 Changed 3 years ago by Martin Dix

Usual APS city mask generation uses IMETHOD=2 for data aggregation, https://code.metoffice.gov.uk/doc/ancil/ancil9.0/ATDP/ATDP01.html#agregation (most of the non-variable grid Met Office LAM configurations also use this).

With IMETHOD=5 there's options for X and Y offsets of the IGBP data, https://code.metoffice.gov.uk/doc/ancil/ancil9.0/ATDP/ATDP01.html#atdp01-ancilscr-mask-land-sea-masks.

This is designed to work around the offset problems we have here.

https://code.metoffice.gov.uk/trac/ancil/ticket/72 is a open request to add this capability to method 2 but for now we have to use method 5 for the offsets.

The same aggregation method is used for the vegetation properties so the same offsets should be used there.

Last edited 3 years ago by Martin Dix (previous) (diff)

comment:4 Changed 3 years ago by Martin Dix

In the longer term, the Met Office are considering replacing the IGBP data with CCI from ESA, http://www.esa-landcover-cci.org/. This has 300m resolution.

Plotting this with the GA coastline shows no alignment problems.


Color key at http://maps.elie.ucl.ac.be/CCI/viewer/download/CCI-LC_Maps_Legend.pdf

Last edited 3 years ago by Martin Dix (previous) (diff)

Changed 3 years ago by Martin Dix

Attachment: cci_kangaroo_island.png added

comment:5 Changed 3 years ago by Martin Dix

Tools for working with masks now in repository https://code.metoffice.gov.uk/svn/utils/access/branches/dev/martindix/r133_mask_tools/mask_tools

Installed as a module on raijin

% module use ~access/modules
% module load pythonlib/mask_tools

% python ~access/apps/pythonlib/mask_tools/BOM_station_check.py -h     
usage: BOM_station_check.py [-h] [-d DOMAIN] [-p ANCIL_PATH] [--pdf PDF]

Plot BOM station locations on APS city land-sea masks

optional arguments:
  -h, --help     show this help message and exit
  -d DOMAIN      Domain name used on plots
  -p ANCIL_PATH  Path to ancillary files (if not current directory)
  --pdf PDF      Save output to a PDF file (default is to screen)

Example

% python ~access/apps/pythonlib/mask_tools/BOM_station_check.py -p /short/dp9/wml548/ancil/1.5km/SY -d 'APS2 Sydney' --pdf=sydney.pdf

This creates a multi-page PDF with a map of each station that is not over a land point.

Leaving off the –pdf argument gives output to the screen (closing each one moves to the next).

GA coastline data and station locations are in ~access/data/orography.

Last edited 3 years ago by Martin Dix (previous) (diff)

Changed 3 years ago by Martin Dix

Attachment: lakes_entrance_GA.png added

Changed 3 years ago by Martin Dix

Attachment: lakes_entrance_SRTM.png added

Changed 3 years ago by Martin Dix

Attachment: sydney_GA.png added

Changed 3 years ago by Martin Dix

Attachment: sydney_SRTM.png added

Changed 3 years ago by Martin Dix

Attachment: wollongong_GA.png added

Changed 3 years ago by Martin Dix

Attachment: wollongong_SRTM.png added

comment:6 Changed 3 years ago by Martin Dix

It's also possible to generate a mask by regridding either the GA (250m resolution) or SRTM (90m resolution) orography.

This starts with a mask for the target region to define the grid (just copy the standard one)

module load pythonlib/masktools
python ~access/apps/pythonlib/mask_tools/orog_to_mask.py --source=GA qrparm.mask

This can be run interactively on raijin and creates newmask.nc. This can be used to replace the mask field in qrparm.mask by

module load pythonlib/umfile_utils
python ~access/apps/pythonlib/umfile_utils/um_replace_field.py -v 30 -n newmask.nc -V height qrparm.mask


Use --source=SRTM for the SRTM data.

The GA orography seems consistent with the GA coastline. However both orography and coastline exclude some tidal regions that probably should be treated as ocean (and are so with the IGBP data).

Plots here show the SRTM and GA orography with the GA coastline.

Lake Illawarra and the Gippsland Lakes are both missing and Sydney Harbour only extends to the bridge.










Last edited 3 years ago by Martin Dix (previous) (diff)

comment:7 Changed 3 years ago by Robin Bowen

Cc: Wenming Lu chs548 Peter Steinle Yi Xiao Imtiaz Dharssi Milton Woods added

hi Martin

cc to others for their info.

Do you have the relevant parts of this ticket in the wiki docum for ancillaries ?

And question does this need to be documented as specific to city suites ?

cheers

Robin

Changed 3 years ago by Martin Dix

Attachment: lakes_entrance_IGBP.png added

Changed 3 years ago by Martin Dix

Attachment: sydney_IGBP.png added

Changed 3 years ago by Martin Dix

Attachment: wollongong_IGBP.png added

comment:8 Changed 3 years ago by Robin Bowen

Cc: Gary Dietachmayer added

added Gary

Note: See TracTickets for help on using tickets.