NWS13 - Revision history

Alexcrosby: /* Step 2: Apply intermediate input grid to mesh */ add language about fixed basin overlay

2020-06-15T19:52:25Z

Step 2: Apply intermediate input grid to mesh: add language about fixed basin overlay

← Older revision		Revision as of 15:52, 15 June 2020
Line 195:		Line 195:
	===Step 2: Apply intermediate input grid to mesh===		===Step 2: Apply intermediate input grid to mesh===

	The gridded intermediate timesteps are interpolated in space on to the ADCIRC mesh, appropriately overlaying multiple input grids by rank as supplied. At this point the inputs are represented on the mesh at '''WTIMINC''' timesteps.		The gridded intermediate timesteps are interpolated in space on to the ADCIRC mesh, appropriately overlaying multiple input grids by rank as supplied. At this point the inputs are represented on the mesh at '''WTIMINC''' timesteps.

			'''NB:''' For the lowest ranked overlay, the grid-to-mesh interpolation weights are pre-computed, which means that this grid cannot be a storm-following moving grid. For all additional overlays, the weights are dynamically updated.

	===Step 3: Time interpolation to model timestep===		===Step 3: Time interpolation to model timestep===

Alexcrosby: /* Motivation */ enlarge domain thumbnail

2020-06-15T19:49:41Z

Motivation: enlarge domain thumbnail

← Older revision		Revision as of 15:49, 15 June 2020
Line 7:		Line 7:
	A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields. In order to allow ADCIRC access to high quality and efficiently stored inputs the following features were added to the existing [[NWS12]] functionality for the newer NetCDF version ([[NWS13]]).		A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields. In order to allow ADCIRC access to high quality and efficiently stored inputs the following features were added to the existing [[NWS12]] functionality for the newer NetCDF version ([[NWS13]]).

	[[File:overlay_domains.png\|~~200px~~\|thumb\|right\|Example of grid domains for multiple independent wind/pressure overlays, including storm-following.]]		[[File:overlay_domains.png\|400px\|thumb\|right\|Example of grid domains for multiple independent wind/pressure overlays, including storm-following.]]

	* Moving storm-centered grids		* Moving storm-centered grids

Alexcrosby: /* Interpolation Details */ tweak steps formatting

2020-06-15T19:45:18Z

Interpolation Details: tweak steps formatting

← Older revision		Revision as of 15:45, 15 June 2020
Line 186:		Line 186:

	The process occurs as follows:		The process occurs as follows:


	===Step 1: Time/space on input grid(s)===		===Step 1: Time/space on input grid(s)===

	Input fields are interpolated on the input grid (in time, and space if necessary) to the intermediate timesteps specified in the ''fort.15'' parameter '''WTIMINC'''.		Input fields are interpolated on the input grid (in time, and space if necessary) to the intermediate timesteps specified in the ''fort.15'' parameter '''WTIMINC'''.


	===Step 2: Apply intermediate input grid to mesh===		===Step 2: Apply intermediate input grid to mesh===
Line 196:		Line 198:


	===Step 3: ===		===Step 3: Time interpolation to model timestep===

	At each model computational timestep ('''DTDP''') the inputs at intermediate timesteps as represented on the mesh are interpolated in time to the appropriate model timestep node-by-node.		At each model computational timestep ('''DTDP''') the inputs at intermediate timesteps as represented on the mesh are interpolated in time to the appropriate model timestep node-by-node.

Alexcrosby: /* Motivation */ fix bad edit

2020-06-15T19:43:00Z

Motivation: fix bad edit

← Older revision		Revision as of 15:43, 15 June 2020
Line 5:		Line 5:
	==Motivation==		==Motivation==

	A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields. In order to allow ADCIRC access to high quality and efficiently stored inputs the following features were added to the existing [[NWS12]] functionality for newer ~~the~~ NetCDF version ([[NWS13]]).		A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields. In order to allow ADCIRC access to high quality and efficiently stored inputs the following features were added to the existing [[NWS12]] functionality for the newer NetCDF version ([[NWS13]]).

	[[File:overlay_domains.png\|200px\|thumb\|right\|Example of grid domains for multiple independent wind/pressure overlays, including storm-following.]]		[[File:overlay_domains.png\|200px\|thumb\|right\|Example of grid domains for multiple independent wind/pressure overlays, including storm-following.]]

Alexcrosby: /* Motivation */ add the/typo

2020-06-15T19:42:31Z

Motivation: add the/typo

← Older revision		Revision as of 15:42, 15 June 2020
Line 5:		Line 5:
	==Motivation==		==Motivation==

	A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields. In order to allow ADCIRC access to high quality and efficiently stored inputs the following features were added to the existing [[NWS12]] functionality for newer NetCDF version ([[NWS13]]).		A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields. In order to allow ADCIRC access to high quality and efficiently stored inputs the following features were added to the existing [[NWS12]] functionality for newer the NetCDF version ([[NWS13]]).

	[[File:overlay_domains.png\|200px\|thumb\|right\|Example of grid domains for multiple independent wind/pressure overlays, including storm-following.]]		[[File:overlay_domains.png\|200px\|thumb\|right\|Example of grid domains for multiple independent wind/pressure overlays, including storm-following.]]

Alexcrosby: /* Interpolation Details */ Flesh out the interpolation steps

2020-06-15T19:41:28Z

Interpolation Details: Flesh out the interpolation steps

← Older revision		Revision as of 15:41, 15 June 2020
Line 179:		Line 179:
	==Interpolation Details==		==Interpolation Details==

	[[File:Ch ~~movegrid~~ interp ~~spatial~~.png]]		[[File:Ch maxmin interp timeseries.png\|300px\|thumb\|right\|Aliasing of peak storm conditions in time.]]

	[[File:Ch ~~maxmin~~ interp ~~timeseries~~.png]]		[[File:Ch movegrid interp spatial.png\|300px\|thumb\|right\|Spatial aliasing of peak wind speeds.]]

	[[~~File:nws13_interp1.png~~]]		The [[NWS13]] met module implements a multi-step interpolation procedure, primarily to minimize aliasing of the wind and pressure fields in time and space.

	[[File:nws13_interp2.png]]		The process occurs as follows:

			===Step 1: Time/space on input grid(s)===

			Input fields are interpolated on the input grid (in time, and space if necessary) to the intermediate timesteps specified in the ''fort.15'' parameter '''WTIMINC'''.

			===Step 2: Apply intermediate input grid to mesh===

			The gridded intermediate timesteps are interpolated in space on to the ADCIRC mesh, appropriately overlaying multiple input grids by rank as supplied. At this point the inputs are represented on the mesh at '''WTIMINC''' timesteps.


			===Step 3: ===

			At each model computational timestep ('''DTDP''') the inputs at intermediate timesteps as represented on the mesh are interpolated in time to the appropriate model timestep node-by-node.


			[[File:nws13_interp1.png\|800px]]

			[[File:nws13_interp2.png\|800px]]

Alexcrosby: Add some images, interpolation section

2020-06-15T19:06:04Z

Add some images, interpolation section

← Older revision		Revision as of 15:06, 15 June 2020
Line 176:		Line 176:
	* Best case: the lowest rank grid should just be a less resolved version of exact same higher resolution fields that will be overlaid		* Best case: the lowest rank grid should just be a less resolved version of exact same higher resolution fields that will be overlaid
	* Lowest rank grid should probably cover the entire domain/mesh, and the entire model run time period		* Lowest rank grid should probably cover the entire domain/mesh, and the entire model run time period

			==Interpolation Details==

			[[File:Ch movegrid interp spatial.png]]

			[[File:Ch maxmin interp timeseries.png]]

			[[File:nws13_interp1.png]]

			[[File:nws13_interp2.png]]

Alexcrosby: /* Motivation */ Add overlay domain image

2020-06-15T18:37:11Z

Motivation: Add overlay domain image

← Older revision		Revision as of 14:37, 15 June 2020
Line 5:		Line 5:
	==Motivation==		==Motivation==

	A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields.		A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields. In order to allow ADCIRC access to high quality and efficiently stored inputs the following features were added to the existing [[NWS12]] functionality for newer NetCDF version ([[NWS13]]).

			[[File:overlay_domains.png\|200px\|thumb\|right\|Example of grid domains for multiple independent wind/pressure overlays, including storm-following.]]

	* Moving storm-centered grids		* Moving storm-centered grids

Alexcrosby: NWS13 Creation

2020-06-15T18:29:33Z

NWS13 Creation

New page

=NWS13=

Oceanweather Inc. (OWI) NetCDF Meteorological Input Files

==Motivation==

A major source of error in ADCIRC can come from the accuracy and representation of the wind and pressure fields.

* Moving storm-centered grids
* Grids that can change size
* Curvilinear grids
* Arbitrary # of grid overlays
* New ADCIRC moving interpolation code
* Arbitrary & irregular timesteps

Some future-proofing attempts:

* Multi-resolution representation of tropical cyclone wind fields
* Be able to introduce additional meteorological parameters using the same format/NWS input
** Possibility to include wind stresses?
** Alternative reference heights?
** Ice information?
** Rainfall?

==NetCDF File Schema==

This description assumes some familiarity with NetCDF, HDF or other similar self-describing binary file formats, and in particular features like "groups" available in HDF5 and NetCDF4. Understanding of NetCDF conventions like Climate Forecast (CF) may also be helpful. For more information about NetCDF please visit the links below.

* https://www.unidata.ucar.edu/software/netcdf/
* https://www.hdfgroup.org/
* https://cfconventions.org/
* https://github.com/Unidata/netcdf4-python
* https://github.com/pydata/xarray

===''' '' OWI-NWS13 Convention '' '''===

'''NetCDF4 File''' contents:
* Group(s) – 1 group per grid or overlay
* Variables – U10/V10, PSFC
* Dimensions – time, yi, xi
* Attributes – grid rank/priority

'''Global Attributes''':
* ''group_order'': space separated list of group names reflecting their appropriate rank/order
* ''conventions'': should include “OWI-NWS13“ (Climate Forecast conventions (CF) seemingly don’t support NC groups)

'''Group Attributes''':
* ''rank'': integer representing the order of overlay/precedence in application to nodes

'''Group Dimensions''':
* ''time'' – length of time dimension
* ''yi'' – number of rows in spatial grid description
* ''xi'' – number of columns in spatial grid description

'''Group Variables''':
* ''U10'' (time, yi, xi), U-component of 10m WS (m/s)
* ''V10'' (time, yi, xi), V-component of 10m WS (m/s)
* ''PSFC'' (time, yi, xi), Surface Pressure (mb)
* ''lon'' (yi, xi) or (time, yi, xi), Longitude in Decimal Degrees
* ''lat'' (yi, xi) or (time, yi, xi), Latitude in Decimal Degrees
* ''time'' (time), Datetime-number with units of “minutes from YYYY-mm-dd HH:MM:SS”
* ''clon'' and ''clat'' (time), optional storm center coordinates for Powell drag

===Example===
netcdf fort.22 {

// global attributes:
:group_order = "Main JPM0135" ;
:institution = "Oceanweather Inc. (OWI)" ;
:conventions = "CF-1.6 OWI-NWS13" ;

group: Main {
dimensions:
yi = 211 ;
xi = 221 ;
time = 133 ;
variables:
float lon(yi, xi) ;
lon:_FillValue = NaNf ;
lon:units = "degrees_east" ;
lon:standard_name = "longitude" ;
lon:axis = "X" ;
lon:coordinates = "time lat lon" ;
float lat(yi, xi) ;
lat:_FillValue = NaNf ;
lat:units = "degrees_north" ;
lat:standard_name = "latitude" ;
lat:axis = "Y" ;
lat:coordinates = "time lat lon" ;
int64 time(time) ;
time:units = "minutes since 1990-01-01T01:00:00" ;
time:calendar = "proleptic_gregorian" ;
float U10(time, yi, xi) ;
U10:_FillValue = NaNf ;
U10:units = "m s-1" ;
U10:coordinates = "time lat lon" ;
float V10(time, yi, xi) ;
V10:_FillValue = NaNf ;
V10:units = "m s-1" ;
V10:coordinates = "time lat lon" ;
float PSFC(time, yi, xi) ;
PSFC:_FillValue = NaNf ;
PSFC:units = "mb" ;
PSFC:coordinates = "time lat lon" ;
// group attributes:
:rank = 1 ;
} // group Main

group: JPM0135 {
dimensions:
time = 133 ;
yi = 501 ;
xi = 501 ;
variables:
int64 time(time) ;
time:units = "minutes since 1990-01-01T01:00:00" ;
time:calendar = "proleptic_gregorian" ;
float lat(time, yi, xi) ;
lat:_FillValue = NaNf ;
lat:units = "degrees_north" ;
lat:standard_name = "latitude" ;
lat:axis = "Y" ;
lat:coordinates = "time lat lon" ;
float lon(time, yi, xi) ;
lon:_FillValue = NaNf ;
lon:units = "degrees_east" ;
lon:standard_name = "longitude" ;
lon:axis = "X" ;
lon:coordinates = "time lat lon" ;
float U10(time, yi, xi) ;
U10:_FillValue = NaNf ;
U10:units = "m s-1" ;
U10:coordinates = "time lat lon" ;
float V10(time, yi, xi) ;
V10:_FillValue = NaNf ;
V10:units = "m s-1" ;
V10:coordinates = "time lat lon" ;
float PSFC(time, yi, xi) ;
PSFC:_FillValue = NaNf ;
PSFC:units = "mb" ;
PSFC:coordinates = "time lat lon" ;
// group attributes:
:rank = 2 ;
} // group JPM0135
}

===Notes===

* Decoupling the yi/xi dimensions from lat/lon allows lat and lon to be 2-d arrays by depending on both dimensions
* Regular grids and Curvilinear grids
* Non-evenly spaced grids as long as they can be expressed in a 2-d "mesh-grid"
* Grids that change spatial resolution or position in time (but have consistent yi/xi array size)
* Each group/sub-grid can define the timesteps independently, including start and stop times
* Fill Value (and NetCDF packing/compression), and ieee nan floats supported

==fort.15 Configuration==

'''WTIMINC''': configurable grid-to-mesh interpolation timestep in seconds, separate from the input data timesteps. If the file contains storm-following grids, there may be some utility in setting this finer than the input timestep(s).

''&owiWindNetcdf'' fort.15 namelist with all inputs quoted as strings
* ''NWS13ColdStartString'': required cold start time of simulation formatted as 'YYYYMMDD.HHMMSS'
* ''NWS13WindMultiplier'': optional wind speed multiplier ('''DWM''' from [[NWS12]])
* ''NWS13File'': optional file name for netCDF file ( ''fort.22.nc'' default )
* ''NWS13GroupForPowell'': optional group # to use for Powell drag
''example'':
&owiWindNetcdf NWS13File='fort.22.nc' NWS13ColdStartString='20000706.000000' /

==Suggestions==

* Keep the spatial and temporal resolutions to only what is required
* Add sub-grids, more complex time dimensions, and complex lat/lon grid definitions only as necessary by sub-grid/group
* In most cases 1-hourly specified--if you have a moving-grid overlay and can rely on the built-in moving-aware interpolation
* Avoid discontinuities between overlays/grids
* Best case: the lowest rank grid should just be a less resolved version of exact same higher resolution fields that will be overlaid
* Lowest rank grid should probably cover the entire domain/mesh, and the entire model run time period