I'm trying to extract data from a raster file from "Snow Data Assimilation System (SNODAS)" and convert it to a MMQGIS (plugin) grid layer (Hexagon) with QGIS. This raster file stores the snow depth of the whole of the United States.

  1. I load my raster (tif)
  2. I extract the data with a terrain analysis (Aspect)
  3. I create my grid layer as a hexagon
  4. I run Zonal statistics from my Aspect to my Grid
  5. At this point my grid has "count", "sum" and "mean" attributes. I use the "sum" attribute.
  6. I load it on TileMill and I style the "sum" attribute.

I would like to know if I have the correct process and the right information about snow depth on my Grid layer? Is "Sum" the right attribute to print?

  • 1
    sorry but I don't understand what exactly is your question.
    – underdark
    Commented Oct 4, 2012 at 18:12
  • Sorry, I was not clear. I edited my question.
    – uto
    Commented Oct 5, 2012 at 0:05
  • What are you trying to display? The calculation you describe gives the sum of values of your original raster within each zone. Normally this makes no sense, because its units of measurement is a length. (For example, would you really want to say the snow "depth" in a zone is, say, 200 meters because the snow is 2 meters deep and the zone happens to be occupied by 100 raster cells?)
    – whuber
    Commented Oct 5, 2012 at 3:14
  • Waw okay i'm very fare of that. My goal is to print on a hexagon grid layer the snow depth of USA. Like this link
    – uto
    Commented Oct 5, 2012 at 6:40
  • I'm a bit puzzled too. I don't know this data set but presumably the pixel values are for snow depth in metres, feet, or some other unit of length? So why Step 2, above? Why not run the zonal statistics plug-in directly on the source raster to obtain a mean value for each hexagonal polygon? Or perhaps 'snow depth' doesn't mean what I think it means. N.
    – nhopton
    Commented Oct 5, 2012 at 13:13

1 Answer 1


I work for NOHRSC, the weather service office that creates the SNODAS rasters. Each pixel in a snow depth raster represents the model's best estimate for the value over the 1km x 1km cell. So, there is an implicit 3000x7000 grid in the data. Depth is measured in millimeters, in 16-bit big-endian integers. Raster data is available at http://www.nohrsc.noaa.gov/archived_data .

The respondent wants to re-aggregate the data to much larger hexagons.
MEAN would be the proper Zonal parameter.

Note: SNODAS models Snow Water Equivalent, the liquid water content of the snowpack. SWE is a much more robust statistic than Depth, which changes the minute the snowflakes hit the ground. A twelve-inch snowfall slumps to eight inches after a day of sunshine and wind as the snowpack compresses and metamorphs, without any change to the SWE value. Depth is computed after that fact, by multiplying SWE against an independent estimate of Snow Density.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.