Create a Raster file out of a ASCII grid file?

Question

I'm using a special system to do geomagnetic surveys. The software let's me output the information as image files, ASCII grid or in its own proprietary file format.

Now I want to get its information in my GIS (using QGIS). I can of course import the exported image files but I would love to import the raw data to play around with the visualisation a bit more, not depending on the original software.

The files look like this:

X   Y   Messwert
995.00  1000.00 0.00
995.00  1000.05 0.14
995.00  1000.10 0.28
995.00  1000.15 -0.07
995.00  1000.20 -0.42
995.00  1000.25 -1.26
995.00  1000.30 -2.17

and so on..... (looooong file ;) )

Every 0.05 cm movement on X and Y contains a data value, positive and negative.

I want to convert this data into a raster file, a pixel for every data value. My goal is to tune the visualisation comparable to a DEM TIFF file in my QGIS project, without having the problem to export this through the original software every time.

What would be the best way to do this?

I think GDAL is the program to use I need some help. Perhaps there is even a way to do this in QGIS?

Update:

So I finally imported a fraction of my data, sorting everything to Y. Saving everything as TIFF wasn't a problem as well. Now my next step is to get this spatially correct data (in terms of length) into my project. The coordinates in the file are just a local project oriented coordinate system.

Georeferencing the created TIFF wasn't a big problem but it results in a little annoying problem. after georeferencing my perfect square gets rotated a bit, resulting in big nodata areas.

My data also contains positive as well as negative data and even zero is important.

I couldn’t find a way to get this nodata area to disappear, QGIS georeferencing gives it a value that is contained in the data areas as well. if I set this to transparency my raster files gets some annoying holes.

Answer 1

You can easily open ASCII xyz triplicate data in QGIS under "Add Raster Data" with a "ASCII Gridded XYZ (.xyz)" file type. You can also covert it to a different format under the "Raster > Conversion > Translate (Convert format)" menu. Alternately, you can do this under the "Raster > Conversion > Rasterize" menu with a "Comma Separated Value (.csv)" file type.

Answer 2

According to http://www.gdal.org/frmt_xyz.html:

Cells with same Y coordinates must be placed on consecutive lines

which is not fulfilled by your dataset.

So you can

• resort the tabular data with an external program
• exchange X and Y in the header (you have to mirror your raster later)
• load the data as point data using Delimited text, then rasterize it.

Answer 3

I would hit it with a python script.

import numpy as np
asc = open('yourfilename', 'r')
grid = []
for line in asc:
  line = line.strip()
  if line == 'X   Y   Messwert':
    continue
  grid.append(map(float, line.split(' ')))
asc.close()
grid = np.array(grid)

#naively get number of rows
rows = np.sum(grid[:,0] == grid[0,0])

#naively get number of colums
cols = np.sum(grid[:,1] == grid[0,1])

#get values
vals = grid[:, 2]

#pad some nodata values if the total number of elements does not form
#a nice rectangle with exactly rows*cols number of cells.
if len(vals) != rows * cols:
   vals = np.hstack((vals, -9999 * np.ones(rows * cols - len(vals))))

#turn data into 2-d matrix(column-wise using order='F')
value_grid = np.reshape(vals, (rows, cols), order='F')

#write out dummy ascii raster file:
out = open('raster.asc', 'wb+')
out.write('ncols         %i\n' % cols)
out.write('nrows         %i\n' % rows)
out.write('xllcorner     %i\n' % 1111)#dummy or put your own
out.write('yllcorner     %i\n' % 2222)#dummy or put your own
out.write('cellsize      %i\n' % 0.05)
out.write('NODATA_value  %i\n' % -9999)

np.savetxt(out, value_grid, fmt='%%.%2f')
out.close()

granted, you can load the file with a numpy command which would save a few lines and you can combine some operations but my intent is to give something readable...and hopefully useable. Please excuse any syntax errors, I didn't run this myself.