[R-sig-Geo] ESRI Ascii Grid
Roger Bivand
Roger.Bivand at nhh.no
Wed Jan 14 10:14:31 CET 2009
On Tue, 13 Jan 2009, Dylan Beaudette wrote:
> On Tuesday 13 January 2009, Edzer Pebesma wrote:
>> This sounds very familiar; I ran into this about 6 years ago or so.
>
> Could it be related to this ticket:
>
> http://trac.osgeo.org/gdal/ticket/2369
Yes, exactly! I have checked with readGDAL() with GDAL 1.6.0, and the
problem is resolved (in fact it defaults to int32), while the problem
remains in the 1.5.* GDAL series (including 1.5.4 released last week) for
the sample file in the ticket. May I add the sample file to rgdal to give
an indication of this?
The next Windows binary packages will be GDAL 1.5.4, not 1.6.0 - I'm
working on getting 1.6.0 to check through rgdal.
If Brian is working on something other than Windows with the CRAN binary
package, upgrading GDAL will fix the problem, because rgdal will use the
installed GDAL.
Roger
PS. GDAL < 1.6 can be tricked into doing the right thing by using a
NODATA_value in the file that is either less than the minimum int16 or
greater than the maximum int16, so with a text editor this can be a
"solution". The GDAL driver is more flexible in letting you choose
subscenes, etc, so this may be prefered to readAsciiGrid().
>
> ???
>
> should be fixed in a recent GDAL build.
>
> Cheers,
>
> Dylan
>
>
>> The asciigrid driver in gdal reads the first so 100K of the file to
>> figure out whether the data are integer or floating point. It then
>> searches for a dot in any of the numbers, indicating floating point
>> data. If it doesn't find a floating point (g2), it assumes integer and
>> proceeds with that. The funny min and max values equal the min and max
>> value of a signed two-bytes integer, +/- 2^15. The problem is that many
>> images come as integer data, and what do you want gdal to do: read your
>> 13 Mb (zipped!) first to figure out what it is, to start all over again?
>> This would make many other users very unhappy.
>>
>> I remember filing a bug long ago.
>>
>> How about putting a dot after your first value in the grid?
>> --
>> Edzer
>>
>> Brian R Miranda wrote:
>>> Roger,
>>>
>>> Thanks for your help. I believe you are correct that the printed summary
>>> values have been rounded. I couldn't quite figure out how to extract the
>>> range specifically, but when I did a summary for the data.frame from each
>>> grid I got what I was expecting for the one created by readAsciiGrid
>>> (g1). However, the grid created by readGDAL (g2) still shows unexpected
>>> min and
>>>
>>> max values. See results below:
>>>> d1 <- g1 at data
>>>> summary(d1)
>>>
>>> distrail_100m.asc
>>> Min. : 0
>>> 1st Qu.: 2878
>>> Median : 6562
>>> Mean : 7896
>>> 3rd Qu.: 11670
>>> Max. : 37082
>>> NA's :6483714
>>>
>>>> d2 <- g2 at data
>>>> summary(d2)
>>>
>>> band1
>>> Min. : -32768
>>> 1st Qu.: 2844
>>> Median : 6516
>>> Mean : 7718
>>> 3rd Qu.: 11600
>>> Max. : 32767
>>> NA's :6483714
>>>
>>> For my own purposes, I have found that I should continue using
>>> readAsciiGrid to bring in my grids. But I would be happy to share my
>>> dataset to help figure out why readGDAL has these strange results. My
>>> Ascii file is 13Mb zipped, probably too large to email, and I don't have
>>> an FTP site that can be accessed outside of the Forest Service. I can
>>> send the file to you off list using YouSendIt.com if you would like, or
>>> if you have an alternative suggestion for transferring the file, let me
>>> know. Thanks again for your assistance, and if you need any additional
>>> information from me, let me know.
>>>
>>> Cheers,
>>> -Brian
>>>
>>> =======================
>>> Brian R. Miranda
>>> USDA Forest Service
>>> Northern Research Station
>>> brmiranda at fs.fed.us
>>>
>>> Institute for Applied Ecosystem Studies
>>> 5985 Highway K
>>> Rhinelander, WI 54501
>>> phone: 715-362-1186
>>> fax: 715-362-1166
>>>
>>>
>>>
>>>
>>>
>>> Roger Bivand <Roger.Bivand at nhh.no>
>>> 01/13/2009 02:49 PM
>>> Please respond to
>>> Roger.Bivand at nhh.no
>>>
>>>
>>> To
>>> Brian R Miranda <brmiranda at fs.fed.us>
>>> cc
>>> r-sig-geo at stat.math.ethz.ch
>>> Subject
>>> Re: [R-sig-Geo] ESRI Ascii Grid
>>>
>>> On Tue, 13 Jan 2009, Brian R Miranda wrote:
>>>> Hello,
>>>>
>>>> I have an Ascii grid that was created using the GridAscii command in
>>>> ArcGrid. The grid represents cell distance to railroads and has values
>>>> ranging from 0 to 37,082. When I bring this grid into R as a
>>>> SpatialGridDataFrame, the maximum and/or the minimum values in the grid
>>>
>>> do
>>>
>>>> not exactly match the original grid. I have tried using both readGDAL
>>>
>>> and
>>>
>>>> readAsciiGrid for this purpose. I have confirmed in a text editor that
>>>> the Ascii file does contain the maximum value of 37,082. Below are the
>>>> results I have seen. Note the differences in the Max. value and Min.
>>>> value (from readGDAL).
>>>
>>> First, beware of the possible difference between the printed summary
>>> value
>>>
>>> and the actual value - printed summaries may round. What does range() of
>>> the variable say? The difference between the two functions is puzzling -
>>> could you make the data available (off list for example, zipped,
>>> preferably for download) to try to see what is going on?
>>>
>>> Roger
>>>
>>>>> g1 <- readAsciiGrid("G:/Oconto/WI_Fire/distrail_100m.asc")
>>>>> summary(g1)
>>>>
>>>> Object of class SpatialGridDataFrame
>>>> Coordinates:
>>>> min max
>>>> coords.x1 -263487.6 149512.4
>>>> coords.x2 358821.1 666821.1
>>>> Is projected: NA
>>>> proj4string : [NA]
>>>> Number of points: 2
>>>> Grid attributes:
>>>> cellcentre.offset cellsize cells.dim
>>>> 1 -263437.6 100 4130
>>>> 2 358871.1 100 3080
>>>> Data attributes:
>>>> Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
>>>> 0 2878 6562 7896 11670 37080 6483714
>>>>
>>>>> g2 <- readGDAL("G:/Oconto/WI_Fire/distrail_100m.asc")
>>>>
>>>> G:/Oconto/WI_Fire/distrail_100m.asc has GDAL driver AAIGrid
>>>> and has 3080 rows and 4130 columns
>>>>
>>>>> summary(g2)
>>>>
>>>> Object of class SpatialGridDataFrame
>>>> Coordinates:
>>>> min max
>>>> x -263487.6 149512.4
>>>> y 358821.1 666821.1
>>>> Is projected: NA
>>>> proj4string : [NA]
>>>> Number of points: 2
>>>> Grid attributes:
>>>> cellcentre.offset cellsize cells.dim
>>>> x -263437.6 100 4130
>>>> y 358871.1 100 3080
>>>> Data attributes:
>>>> Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
>>>> -32770 2844 6516 7718 11600 32770 6483714
>>>>
>>>> The results from readAsciiGrid are pretty close (Max = 37,080) to the
>>>> original grid, but readGDAL results in much different values. I would
>>>> expect both functions to create an object with exactly the same values
>>>
>>> as
>>>
>>>> the Ascii grid. Can anyone explain why this occurs, and if there is a
>>>> different way to get my grid into R (I eventually need an 'im' to work
>>>> with)? I am new to spatial analysis using R, so if there is something
>>>> obvious I am missing I apologize. Thanks in advance for any help!
>>>>
>>>> Cheers,
>>>> -Brian
>>>>
>>>>
>>>> =======================
>>>> Brian R. Miranda
>>>> USDA Forest Service
>>>> Northern Research Station
>>>> brmiranda at fs.fed.us
>>>>
>>>> Institute for Applied Ecosystem Studies
>>>> 5985 Highway K
>>>> Rhinelander, WI 54501
>>>> phone: 715-362-1186
>>>> fax: 715-362-1166
>>>>
>>>>
>>>> [[alternative HTML version deleted]]
>>>>
>>>> _______________________________________________
>>>> R-sig-Geo mailing list
>>>> R-sig-Geo at stat.math.ethz.ch
>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>
>
>
--
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no
More information about the R-sig-Geo
mailing list