[R-sig-Geo] issue when opening raster image in R, the pixel values a rescaled somehow

[Edzer Pebesma]

I see at the end of the output obtained with running gdalinfo on the file:

Band 1 Block=1200x3 Type=UInt16, ColorInterp=Gray
  NoData Value=0
  Unit Type: K
  Offset: 0,   Scale:0.02
  Metadata:
    DESCRIPTION=8-day daytime 1km grid Land-surface Temperature

This means that Kelving values are obtained by computing

0 (offset) + 0.02 (scale) * pixel_value

packages stars and readGDAL do this transformation for you, so you get
actual Kelvin values, and represent them with doubles so there is no
rounding problem. Other, lower level tools may give you INT2 values
where you need to take care of the offset & scale yourself.

On 3/24/20 10:28 AM, Roger Bivand wrote:
> Opening in stars shows that the read data are interpreted as degrees
> Kelvin:
> 
>> library(stars)
>> ss <- read_stars("MOD11A2.A2000049.h17v05.006.2015058135048.tif")
>> print(ss, n=1e8)
> stars object with 2 dimensions and 1 attribute
> attribute(s):
>  MOD11A2.A2000049.h17v05.006.2015058135048.tif [K]
>  Min.   :272.1
>  1st Qu.:293.3
>  Median :296.5
>  Mean   :296.1
>  3rd Qu.:298.8
>  Max.   :309.2
>  NA's   :926190
> dimension(s):
>   from   to   offset    delta  refsys point values
> x    1 1200 -1111951  926.625 unnamed FALSE   NULL [x]
> y    1 1200  4447802 -926.625 unnamed FALSE   NULL [y]
> 
> The summary agrees with
> 
>> library(rgdal)
>> o <- readGDAL("MOD11A2.A2000049.h17v05.006.2015058135048.tif")
>> summary(o)
> Object of class SpatialGridDataFrame
> Coordinates:
>        min     max
> x -1111951       0
> y  3335852 4447802
> Is projected: TRUE
> proj4string :
> [+proj=sinu +lon_0=0 +x_0=0 +y_0=0 +R=6371007.181 +units=m +no_defs]
> Grid attributes:
>   cellcentre.offset cellsize cells.dim
> x          -1111487 926.6254      1200
> y           3336315 926.6254      1200
> Data attributes:
>      band1
>  Min.   :272.1
>  1st Qu.:293.3
>  Median :296.5
>  Mean   :296.1
>  3rd Qu.:298.8
>  Max.   :309.2
>  NA's   :926190
> 
> and
> 
>> library(raster)
>> r <- raster("MOD11A2.A2000049.h17v05.006.2015058135048.tif")
>> summary(values(r))
>    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's
>   272.1   293.3   296.5   296.1   298.8   309.2  926190
> 
> So my feeling is that the representation is correct in degrees Kelvin.
> 
> Others please join in, I'm not familiar with MODIS HDF data. I think
> something gives the metadata in the GeoTIFF the wrong data bounds - they
> do seem to be UInt16. The output of gdalinfo (command line) includes:
> 
> ...
>   units=K
>   valid_range=7500, 65535
> ...
> Band 1 Block=1200x3 Type=UInt16, ColorInterp=Gray
>   NoData Value=0
>   Unit Type: K
> ...
> 
> Roger
> 
> 
> On Tue, 24 Mar 2020, Victor F. Rodriguez Galiano wrote:
> 
>> Hi Roger, the link to he image is given below. This image was generated
>> from a .hdf file using the function gdal_translate.
>>
>> gdal_translate(sds[1], dst_dataset=paste0(substr(filename, 1,
>> nchar(filename)-4) ,".tif", ))
>>
>> https://wetransfer.com/downloads/4569e9a85c02faad5016b83d0b50217820200324084015/f2eafdfafaae74de9ac8e25017fc5cc520200324084015/e46e80
>>
>>
>>
>> Thanks
>>
>> Victor
>>
>> El lun., 23 mar. 2020 a las 21:07, Roger Bivand (<Roger.Bivand using nhh.no>)
>> escribió:
>>
>>> Please make the file in question available for download (url), or the
>>> url
>>> you downloaded it from (no login), on this list. It looks as though the
>>> unsigned 16 bit integer is not right.
>>>
>>> Roger Bivand
>>> Norwegian School of Economics
>>> Bergen, Norway
>>>
>>> ------------------------------
>>> *Fra:* R-sig-Geo <r-sig-geo-bounces using r-project.org> på vegne av Victor F.
>>> Rodriguez Galiano <vrgaliano using gmail.com>
>>> *Sendt:* mandag 23. mars 2020, 21.02
>>> *Til:* r-sig-geo using r-project.org
>>> *Emne:* [R-sig-Geo] issue when opening raster image in R, the pixel
>>> values a rescaled somehow
>>>
>>> I opened a Geotiff image in R using the raster function. However, the
>>> image seems to be rescaled. The minimum and maximum values should be
>>> 13607
>>> and 15461 but are 275 and 305. The Geotiff image when opened in a GIS is
>>> correct but not in R. This is my code: Script: library(raster)
>>> trial<-raster("MOD11A2.A2000049.h17v05.006.2015058135048.tif",
>>> datatype =
>>> "INT2U") trial plot(trial) [[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
>>>
>>
> 
> 
> _______________________________________________
> 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, 48149 Muenster, Germany
Phone: +49 251 8333081

-------------- next part --------------
A non-text attachment was scrubbed...
Name: pEpkey.asc
Type: application/pgp-keys
Size: 3110 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20200324/7b7c1c4b/attachment.bin>