Re: FW: I mean regridding the data to a rectiliniear lat / lon grid

From: Dennis Shea <shea_at_nyahnyahspammersnyahnyah>
Date: Fri Oct 19 2012 - 07:20:27 MDT

You want to regrid.

The Lambert Conformal
regrid to a 'rectilinear grid'.
the 'ESMF Regrid' page

This archive was generated : Tue Oct 23 2012 - 11:10:04 MDT is 'curvilinear grid' and you want to There are numerous examples on r.edu/Applications/">http://www.ncl.ucar.edu/Applications/ then "ESMF Regridding" serve as a basis for your needs PM, Ping Yang wrote: I am just following on a previous posted topic related to Lambert Conformal Conic grid, I plan to re-project (or regrid) a grid data which has the following information: short lambert_conformal_conic ; lambert_conformal_conic:grid_mapping_name = "lambert_conformal_conic" ; lambert_conformal_conic:longitude_of_central_meridian = -100. ; lambert_conformal_conic:latitude_of_projection_origin = 42.5 ; lambert_conformal_conic:false_easting = 0. ; lambert_conformal_conic:false_northing = 0. ; lambert_conformal_conic:standard_parallel = 25., 60. ; double x(x) ; x:units = "m" ; x:long_name = "x coordinate of projection" ; x:standard_name = "projection_x_coordinate" ; double y(y) ; y:units = "m" ; y:long_name = "y coordinate of projection" ; y:standard_name = "projection_y_coordinate" ; double time(time) ; time:long_name = "time" ; time:calendar = "standard" ; time:units = "days since 1980-01-01 00:00:00 UTC" ; time:bounds = "time_bnds" ; double lat(y, x) ; lat:units = "degrees_north" ; lat:long_name = "latitude coordinate" ; lat:standard_name = "latitude" ; double lon(y, x) ; lon:units = "degrees_east" ; lon:long_name = "longitude coordinate" ; lon:standard_name = "longitude" ; short yearday(time) ; yearday:long_name = "yearday" ; yearday:valid_range = 1s, 365s ; float prcp(time, y, x) ; prcp:long_name = "daily total precipitation" ; prcp:units = "mm/day" ; prcp:missing_value = -9999.f ; prcp:_FillValue = -9999.f ; prcp:valid_range = 0.f, 200.f ; prcp:coordinates = "lat lon" ; prcp:grid_mapping = "lambert_conformal_conic" ; prcp:cell_methods = "area: sum time: sum" ; double time_bnds(time, nv) ; I need to do the following: 1) change the projection into lat/lon 2)regrid the data from higher resolution into lower resolution(aggregating) I am asking which will be the first to do and how can I do it(lambert_conformal_conic to lon/lat). Looking forward to hearing from you. On Fri, Aug 10, 2012 at 4:00 PM, David Brown <dbrown@ucar.edu <mailto:dbrown@ucar.edu>> wrote: Hi Gualberto, I am enclosing a modified version of the script that uses NCL to generate the 2D lat/lon coordinate arrays and write them to a file. To demonstrate their correctness, it uses them to transform the data into an orthographic projection. You can use the lat/lon arrays to perform the re-gridding, step 4 of the process outlined by Dave Allured. For that I recommend studying the ESMF regridding examples at http://www.ncl.ucar.edu/Applications/ESMF.shtml. You will need NCL 6.1.0-beta if you don't have it already. I also am attaching the orthographic plot. On Aug 9, 2012, at 11:41 PM, Gualberto Hernández Juárez wrote: Dave, I offer youexcuseforlatereply, So I understandwith the file yousent me"lcnative.ncl"I can nowrun theregridding? You will havesomevery basicexamplethat can serve asreferenceto do it?I supposeI have to useESMF. Thanks, Gualberto ------------------------------------------------------------------------ Subject: Re: [ncl-talk] FW: I mean regridding the data to a rectiliniear lat / lon grid From:dbrown@ucar.edu <mailto:dbrown@ucar.edu> Date: Thu, 9 Aug 2012 17:02:01 -0600 CC:ncl-talk@ucar.edu <mailto:ncl-talk@ucar.edu> To:betoic@hotmail.com <mailto:betoic@hotmail.com> Hi Gualberto, I have figured out what I think are the correct (or at least close enough) parameters to get the right results from the PROJ4 tool. I think the main issue I was having is that the Lambert Conformal Conic projection has a singularity over the equator, because it ceases to be conic and becomes a cylinder. At least NCL does not work with a single parallel set to 0.0. However, I can make it work using 2 standard parallels very close to each other (0.0 and 0.01). The PROJ4 tool does not give up, but the numbers were wrong if you do not do something similar. Anyway it would now be possible to generate the 2d lat/lon arrays that you will need for re-gridding, either just using PROJ4 or by sampling the projected space in NCL. I am attaching a plot of temperature that I think shows the map boundaries aligning very well with the data. I also include the version of the script that created it. -dave On Aug 8, 2012, at 1:47 PM, Gualberto Hernández Juárez wrote: I have not done, mainly becauseArcGisalsolimitedinNetCDFformat ------------------------------------------------------------------------ From:dbrown@ucar.edu <mailto:dbrown@ucar.edu> Subject: Re: [ncl-talk] FW: I mean regridding the data to a rectiliniear lat / lon grid Date: Wed, 8 Aug 2012 12:46:30 -0600 To:betoic@hotmail.com <mailto:betoic@hotmail.com> One question: Can you find out the projection parameters and/or the corner lat/lon values using ArcGIS? -dave On Aug 8, 2012, at 12:35 PM, Gualberto Hernández Juárez wrote: Dave, This is howI see itinArcGIS,correspondsto the same variableMSLP.As yousaidthe picture yousent meisa little shaky. Gualberto. ------------------------------------------------------------------------ From:dbrown@ucar.edu <mailto:dbrown@ucar.edu> Subject: Re: [ncl-talk] FW: I mean regridding the data to a rectiliniear lat / lon grid Date: Wed, 8 Aug 2012 11:17:02 -0600 To:betoic@hotmail.com <mailto:betoic@hotmail.com> I was hoping for an image with the continental boundaries correctly placed. To show you what I mean here is the best I have been able to do using NCL. The trick is to figure out the correct projection parameters and the lat/lon locations of the lower left and upper right corner points. I did this using the PROJ4 projection tool (similar in capabilities to the tool Dave Allured recommended -- MPS Geotrans). But I am not sure I got the right values because it looks to me like the data does not quite fit the map outlines. I would not trust any re-gridding operation if I could not first get the the map outlines correctly placed. It looks to me (based on plotting a number of the pressure and temperature time steps) that the lower part of Mexico and Central America is skewed to the right and that not so much of South America should be visible. Anyway so you know what I am talking about I am attaching the plot I came up with in NCL. Also the script as it exists now. Perhaps you have access to more information about the projection parameters? -dave On Aug 7, 2012, at 8:20 PM, Gualberto Hern�ndez Ju�rez wrote: Dave, I send the image ------------------------------------------------------------------------ From:dbrown@ucar.edu <mailto:dbrown@ucar.edu> Subject: Re: [ncl-talk] FW: I mean regridding the data to a rectiliniear lat / lon grid Date: Tue, 7 Aug 2012 18:16:24 -0600 To:betoic@hotmail.com <mailto:betoic@hotmail.com> Sorry maybe I misunderstand. I have the NetCDF file MMOUT_DOMAIN1.nc. But I thought you said you uploaded an image (picture ??) of the plot you made in ArcMap. I would like to see the picture if you can send it. -dave On Aug 7, 2012, at 5:23 PM, Gualberto Hern�ndez Ju�rez wrote: This is the fileMMOUT_DOMAIN1.nc The fileyou put inplace it ftp://ftp.cgd.ucar.edu/incoming/ ------------------------------------------------------------------------ From:dbrown@ucar.edu <mailto:dbrown@ucar.edu> Subject: Re: [ncl-talk] FW: I mean regridding the data to a rectiliniear lat / lon grid Date: Tue, 7 Aug 2012 17:19:16 -0600 To:betoic@hotmail.com <mailto:betoic@hotmail.com> (offline) I need to know what you called the file in order to retrieve it. -dave On Aug 7, 2012, at 5:10 PM, Gualberto Hern�ndez Ju�rez wrote: Thanks for your replyDave, Iplotthe fileonly inArcMap ofArcGIS, place an imageinthe directory: ftp://ftp.cgd.ucar.edu/incoming/ I just have toput the mapof the Mexican republic. ------------------------------------------------------------------------ Subject: Re: [ncl-talk] FW: I mean regridding the data to a rectiliniear lat / lon grid From:dbrown@ucar.edu <mailto:dbrown@ucar.edu> Date: Tue, 7 Aug 2012 16:41:41 -0600 CC:ncl-talk@ucar.edu <mailto:ncl-talk@ucar.edu> To:betoic@hotmail.com <mailto:betoic@hotmail.com> Hi Gaulberto, Here is a way to write the text file as X/Y pairs as Dave Allured suggested: f = addfile ("MMOUT_DOMAIN1.nc", "r") x = f->lon y = f->lat nx = dimsizes(x) ny = dimsizes(y) x2d = conform_dims((/ny,nx/),x,1) y2d = conform_dims((/ny,nx/),y,0) outstr = sprintf("%9.3f",x2d) + " " + sprintf("%9.3f",y2d) asciiwrite("xy.txt",outstr) I wonder if you have a plot of this grid in its native (Lambert Conformal) configuration. I have tried to plot it based on the LC parameters given in the data file, but so far I have been unsuccessful. A correct plot of the domain would help. If you can get NCL to plot the data correctly in its native projection, then it is possible to compute the lat/lon coordinates using NCL by itself. -dave On Aug 7, 2012, at 2:06 PM, Gualberto Hern�ndez Ju�rez wrote: Dave, How to make the text file, which you mention in the first point. Thanks Gualberto > Date: Tue, 7 Aug 2012 11:44:09 -0600 > Subject: Re: [ncl-talk] I mean regridding the data to a rectiliniear lat / lon grid > From:dave.allured@noaa.gov <mailto:dave.allured@noaa.gov> > To:betoic@hotmail.com <mailto:betoic@hotmail.com> > CC:ncl-talk@ucar.edu <mailto:ncl-talk@ucar.edu> > > Here is one way. I am sure there are others. > > 1. Use NCL to make a text file containing a 1-dimensional ordered > list of all X/Y coordinate pairs. > > 2. Use MPS Geotrans in batch mode to translate the list of > coordinates from Lambert Conformal to terrestrial lat/lon. It's free > software, and it supports a huge variety of known terrestrial > coordinate systems. > >http://earth-info.nga.mil/GandG/geotrans/ > > 3. Use NCL to read in the translated coordinates, and reshape them > into 2-D lat and lon coordinate arrays. > > 4. Use one of the many NCL regridding functions, plus the 2-D > coordinates from step 3, to regrid the data array to a rectilinear > grid of your choice. > > This is an overview. Steps 1 and 3 are fairly straightforward NCL > coding. Steps 2 and 4 will involve significant study to get them > working correctly. HTH. > > --Dave > > On Tue, Aug 7, 2012 at 10:54 AM, Gualberto Hern�ndez Ju�rez > <betoic@hotmail.com <mailto:betoic@hotmail.com>> wrote: > > I need *regrid* the 2D lat/lon to a rectilinear lat lon. > > > > any comments will be grateful enough support, I'm new to NCL. > > > > Thank you, > > Gualberto > > > > > >> Date: Tue, 7 Aug 2012 10:41:42 -0600 > >> From:shea@ucar.edu <mailto:shea@ucar.edu> > >> To:betoic@hotmail.com <mailto:betoic@hotmail.com> > >> CC:ncl-talk@ucar.edu <mailto:ncl-talk@ucar.edu> > >> Subject: Re: [ncl-talk] I mean regridding the data to a rectiliniear lat / > >> lon grid > >> > >> Technically, this is already a rectilinear grid. > >> > >> double lat ( lat ) > >> _CoordinateAxisType : GeoY > >> axis : Y > >> grid_spacing : 27.0 km > >> long_name : y coordinate of projection > >> standard_name : projection_y_coordinate > >> units : km > >> > >> double lon ( lon ) > >> _CoordinateAxisType : GeoX > >> axis : X > >> grid_spacing : 27.0 km > >> long_name : x coordinate of projection > >> standard_name : projection_x_coordinate > >> units : km > >> > >> float u10 ( time, lat, lon ) > >> long_name : 10 m u wind (m/sec) > >> _FillValue : -9999 > >> grid_mapping : Lambert_Conformal > >> > >> The issue is that the poorly named 'lat' and 'lon' variables > >> should be (say) 'y' and 'x'. > >> > >> Basically, the projection information > >> > >> character Lambert_Conformal ( ncl_scalar ) > >> grid_mapping_name : lambert_conformal_conic > >> long_name : > >> latitude_of_projection_origin : 0 > >> longitude_of_central_meridian : -124.17 > >> false_easting : 0 > >> false_northing : -27 > >> standard_parallel : 0 > >> _CoordinateTransformType : Projection > >> _CoordinateAxisTypes : GeoX GeoY > >> earth_shape : Earth spherical with radius of 6,371,229.0 m > >> earth_radius : 6371229 > >> > >> must be used to generate two-dimensional latitude and longitude > >> coordinates for each grid point. Then the data can be plotted on the > >> lambert map projection. Is this what you want? > >> > >> *** I have no idea how to do this. Maybe someone else does. *** > >> > >> Then, if you really want to *regrid* the 2D lat/lon to a rectilinear > >> lat lon, you would have to use the ESMF regridding > >>http://www.ncl.ucar.edu/Applications/ESMF.shtml > >> > >> > >> On 8/7/12 10:12 AM, Gualberto Hern�ndez Ju�rez wrote: > >> > Hi, > >> > > >> > I mean regridding the data to a rectiliniear lat / lon grid > >> > > >> > locate the file in the directory: > >> > > >> >ftp://ftp.cgd.ucar.edu/incoming/ > >> > > >> > or you can download from this link: > >> > > >> >https://dl.dropbox.com/u/96409134/MMOUT_DOMAIN1.nc > >> > > >> > Thanks I'll be awaiting your response. > >> > Gualberto _______________________________________________ ncl-talk mailing list List instructions, subscriber options, unsubscribe: http://mailman.ucar.edu/mailman/listinfo/ncl-talk _______________________________________________ ncl-talk mailing list List instructions, subscriber options, unsubscribe: http://mailman.ucar.edu/mailman/listinfo/ncl-talk <Huracan Bud.jpg>_______________________________________________ ncl-talk mailing list List instructions, subscriber options, unsubscribe: http://mailman.ucar.edu/mailman/listinfo/ncl-talk <PresionSNivelMar23-05-2012_4-00pm.jpg>_______________________________________________ ncl-talk mailing list List instructions, subscriber options, unsubscribe: http://mailman.ucar.edu/mailman/listinfo/ncl-talk _______________________________________________ ncl-talk mailing list List instructions, subscriber options, unsubscribe: http://mailman.ucar.edu/mailman/listinfo/ncl-talk _______________________________________________ ncl-talk mailing list List instructions, subscriber options, unsubscribe: http://mailman.ucar.edu/mailman/listinfo/ncl-talk Ping Yang, Ph.D. CUNY Environmental Crossroads Initiative Marshak Science Building - Suite 925 The City College of New York - CUNY 160 Convent Avenue, New York NY 10031 Phone: 212-650-5769 Fax: 212-650-7064 <http://mailman.ucar.edu/mailman/listinfo/ncl-talk> _______________________________________________ ncl-talk mailing list List instructions, subscriber options, unsubscribe: http://mailman.ucar.edu/mailman/listinfo/ncl-talk _________________ options, unsubscribe: du/mailman/listinfo/ncl-talk">http://mailman.ucar.edu/mailman/listinfo/ncl-talk on Fri Oct 19 07:20:45 2012 name="navbarfoot" title="Related messages"> [ Message body ] Dennis Shea: "Re: Cross-section plots" Dennis Shea: "Re: Why min/max are different from the actual range?" Ping Yang: "Re: FW: I mean regridding the data to a rectiliniear lat / lon grid" Ping Yang: "Re: FW: I mean regridding the data to a rectiliniear lat / lon grid" href="3129.html" title="Message sent in reply to this message">Ping Yang: "Re: FW: I mean regridding the data to a rectiliniear lat / lon grid" id="options2">Contemporary messages sorted: [ By Date ] [ By Thread ] [ By Subject ] [ By Author ] [ By messages with attachments ] by hypermail 2.1.8