How to extract impervious areas (like buildings, rails, pavements etc.) from las dataset and receive it like a raster? I tried to do it with intensity, but there are many valids with vegetation.

I am working with ArcGis Pro 10.1. I generated and reclassified a raster of intensity with range 0-180 => value 1 (values without NODATA), then I created next raster, a nDSM without medium and high vegetation => value 1 (medium, high vegetation - NODATA).

After that, I used a raster calculator and summed this two rasters; value 2 responds like my results. I was doing it with raster cell sizes equal to 0.2m, but the nDSM with this cell size was not satisfactory (the las dataset have 12 pts/m²). For example, buildings were not filled completely, and I created this nDSM with cell size 1m. So, at the moment my results are a little bit generalized, which is not satisfactory too.

  • 2
    You need to provide a bit more detail as to what have you already tried, what software are you planning on using, etc... – PolyGeo Apr 5 '16 at 21:05
  • When you say that you are working with "ArcGis Pro 10.1", do you mean ArcGIS 10.1 for Desktop? ArcGIS Pro 1.1 seems less likely. – PolyGeo Apr 18 '16 at 22:54

The only "measurements" associated with lidar that would provide any indication of impervious surfaces would be the intensity and that would be suspect because the reflected laser intensity could be any highly reflective object in the NIR range. You would also need calibrated intensity, which is quite uncommon for lidar vendors to collect/provide and difficult to do yourself.

| improve this answer | |

Not sure whether the 12 pts/m² is a high enough density but you could try filtering the cells based on the standard deviation of points elevation which would normally be larger for vegetation cells. See papers by Brasington, Rychkov et al.

| improve this answer | |

Jeffrey Evans has already pointed out that the intensity field wouldn't be expected to do a good job of separating impervious from non-impervious.

You could try converting the LAS data to a canopy height model and then combine it with spectral data of some sort.

| improve this answer | |
  • Can you elaborate? How would this data be combined to extract impervious areas? What type of spectral data one would use? If this spectral data could distinguish such types of areas, why then using the LiDAR data? – Andre Silva Apr 7 '18 at 0:26

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