[R-sig-Geo] Question on hierarchical data format 5 .h5 file format

[Michael Sumner]

see if gdal understands the coordinate arrays already, you can just churn
it through the warper, otherwise construct a vrt that specifies them, which
is a bit obscure how you do it admittedly - but this is a gdal not an R
topic.

Best



On Mon, 15 Nov 2021, 00:22 Edzer Pebesma, <edzer.pebesma using uni-muenster.de>
wrote:

>
> On 14/11/2021 13:59, Edzer Pebesma wrote:
> > Gabriel shared the data with me (off-list, 1Gb) and it contains two hdf5
> > datasets, one containing the raster of the longitude and latitude, the
> > other rasters with the attributres (LST) but no coordinates. I could get
> > a (rough: factor 50 downsampling) plot with stars using:
> >
> > library(stars)
> > lat =
> >
> read_stars("HDF5:ECOSTRESS_L1B_GEO_11712_012_20200730T061029_0601_01.h5://Geolocation/latitude")
>
> >
> > lon =
> >
> read_stars("HDF5:ECOSTRESS_L1B_GEO_11712_012_20200730T061029_0601_01.h5://Geolocation/longitude")
>
> >
> >
> > var =
> > "HDF5:ECOSTRESS_L2_LSTE_11712_012_20200730T061029_0601_01.h5://SDS/LST"
> > lst = read_stars(var, curvilinear = list(x = lon, y = lat))
> > plot(lst, downsample = 50, axes = TRUE, reset = FALSE)
> >
> > library(rnaturalearth)
> > ne = ne_countries(returnclass = "sf")
> > plot(st_geometry(ne), add = TRUE, border = 'yellow')
> > # california border seems to make sense
> >
> >
> > but I'm not sure how to make a regular raster (in some CRS) out of this
> > for exporting to GeoTIFF.
>
> This might be useful: https://github.com/r-spatial/stars/issues/386
> but converting to polygons will be incredibly slow and memory hungry.
>
>
> >
> > On 14/11/2021 11:00, Edzer Pebesma wrote:
> >> This might be because the grid is not regular but possibly
> >> curvilinear, having two raster layers with the lon & lat values for
> >> each pixel. Can you share a sample data set?
> >>
> >> On 14/11/2021 10:47, Gabriel Cotlier wrote:
> >>> Dear Michael,
> >>>
> >>> Following your advice I have tired :
> >>>
> >>> library(terra)
> >>> library(raster)
> >>> library(dplyr)
> >>>
> >>> r = terra::rast(file.choose())
> >>> plot(r$LST)
> >>> r_lst = r$LST
> >>> r_lst %>% raster()
> >>> plot(r_lst)
> >>>
> >>> Then I effectively get a raster layer format from the raster package
> >>> as I
> >>> wanted, but looking at the plot and the raster layer information--see
> >>> bellow-- the scene appears to need both latitude and longitude for
> >>> correct
> >>> geolocation, how can I do this procedure?
> >>>
> >>>> r_lst %>% raster()
> >>> class      : RasterLayer
> >>> dimensions : 5632, 5400, 30412800  (nrow, ncol, ncell)
> >>> resolution : 1, 1  (x, y)
> >>> extent     : 0, 5400, 0, 5632  (xmin, xmax, ymin, ymax)
> >>> crs        : NA
> >>> source     : LST
> >>> names      : LST
> >>> values     : 0, 65535  (min, max)
> >>>
> >>> The extent and the resolution are not the correct ones.
> >>> Maybe in the same .h5  is all the data for georeferencing it although I
> >>> could not see how to get to it and apply it to the layer.
> >>> There is also another file with complementary data related to
> >>> geolocation
> >>> but how can I query it, and get the data for geolocation and apply it
> to
> >>> the raster layer?.
> >>> The objective is to save it as a GeoTIFF file to be opened in QGIS with
> >>> correct geolocation.
> >>>
> >>> What do you suggest for converting the .h5 file to a georeferenced
> >>> Raster
> >>> Layer in GeoTIFF format ?
> >>>
> >>> Thanks a lot.
> >>> Kind regards,
> >>> Gabriel
> >>>
> >>> On Sat, Nov 13, 2021 at 12:22 PM Gabriel Cotlier <gabiklm01 using gmail.com>
> >>> wrote:
> >>>
> >>>> Dear Michael Summer,
> >>>> Thanks a lot for your swift reply.
> >>>> I will give a try to the procedure you mentioned.
> >>>> Kind regards,
> >>>> Gabriel
> >>>>
> >>>>
> >>>> On Sat, Nov 13, 2021 at 12:14 PM Michael Sumner <mdsumner using gmail.com>
> >>>> wrote:
> >>>>
> >>>>> try terra::rast() or stars::read_stars() on the file, both use gdal
> to
> >>>>> interrogate and read. What you get depends on the structure of the
> >>>>> files,
> >>>>> variables interpreted as subdatasets. stars is more general but sees
> >>>>> variables as 2D arrays with bands, terra calls these bands layers.
> >>>>>
> >>>>> Alternatively try raster::raster() or RNetCDF or tidync which might
> be
> >>>>> more straightforward than rhdf5 but all are quite different to each
> >>>>> other
> >>>>> and to the gdal view.
> >>>>>
> >>>>> hth
> >>>>>
> >>>>> On Sat, 13 Nov 2021, 18:56 Gabriel Cotlier, <gabiklm01 using gmail.com>
> >>>>> wrote:
> >>>>>
> >>>>>> Hello,
> >>>>>>
> >>>>>> I would like to be able to visualize / plot a specific data layer
> >>>>>> within
> >>>>>> the structure a hierarchical data file format .h5  in R. I could
> >>>>>> observe
> >>>>>> the structure as follows :
> >>>>>>
> >>>>>> library(rhdf5)
> >>>>>> r= h5ls(file.choose())
> >>>>>> r
> >>>>>>
> >>>>>> and also in this way identify the Group and Name of the data to be
> >>>>>> plotted,
> >>>>>> however I've tried I could not access the data itself, nor plot it
> >>>>>> as a
> >>>>>> raster layer with its corresponding geographic coordinates.
> >>>>>> I would like to be able to select such data, plot it as a raster
> >>>>>> layer
> >>>>>> and
> >>>>>> if possible save it as GeoTIFF file format to be further processed
> >>>>>> with
> >>>>>> the
> >>>>>> raster package as a common raster layer.
> >>>>>>
> >>>>>> If any small guidance example, reprex, of this procedure is possible
> >>>>>> it would be appreciated.
> >>>>>>
> >>>>>> Thanks a lot in advance
> >>>>>> Kind regards,
> >>>>>> Gabriel Cotlier
> >>>>>> --
> >>>>>> Gabriel Cotlier, PhD
> >>>>>> Haifa Research Center for Theoretical Physics and Astrophysics
> >>>>>> (HCTPA)
> >>>>>> University of Haifa
> >>>>>> Israel
> >>>>>>
> >>>>>>          [[alternative HTML version deleted]]
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> R-sig-Geo mailing list
> >>>>>> R-sig-Geo using r-project.org
> >>>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
> >>>>>>
> >>>>>
> >>>
> >>>     [[alternative HTML version deleted]]
> >>>
> >>> _______________________________________________
> >>> R-sig-Geo mailing list
> >>> R-sig-Geo using r-project.org
> >>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
> >>>
> >>
> >
>
> --
> Edzer Pebesma
> Institute for Geoinformatics
> Heisenbergstrasse 2, 48151 Muenster, Germany
> Phone: +49 251 8333081
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

	[[alternative HTML version deleted]]