Discussion: Use of QGIS and LAStools to process LiDAR data

How do I get QGIS with LAStools to process LiDAR data?

I can get LASground to start but after a few minutes it gives me the following error message - Not Responding.

I have gotten OCAD 12 and Karttapulluatin to process an .las file and QGIS to process a .adf file. But whatever I do with LAStools in QGIS always fails. I am following Martin Isenburg's directions but nothing seems to work. The file is very small 15 mb. The only thing I have been successful with is splitting a LiDAR tile. Maybe I would be better off just using LAStools without the GUI provided by QGIS.

Is there a way to get QGIS to process .las or .laz files without LAStools?

Over the last month I have been able to put together an instruction manual on how to build a basemap using OCAD 12, KP, and USGS maps and clip tiles with LAStools. LAStools is a powerful software package it one knows how to use it.

I've just used LasTools and KP. Is there anything that you're having trouble with using just LasTools command line?

Jim, I am trying to use LAStools to generate contour lines and other data just like KP. I am trying to learn how to use this software because it has a nice GUI interface.

I am not an expert on KP and use OCAD 12. However, there are a few things in the works that hope to be available relatively soon with "how to" do some of these things. If anyone has methods that can be published in a document that can help people deal with the multitude of data available to assist in making maps, let me know by email.

What? Lastools has a GUI? Command line is where you'll find the power. Las2Iso is fairly well documented. Sometimes it is easier to do than to explain.

KP did a great job generating contours for my current map project. I've barely changed anything, just minuscule little tweaks. In fact, I often use the contour detail to help place other objects like boulders. Is there something that KP isn't doing? Or is it just that you prefer GUI? KP with default settings did a pretty good job; editing some of the parameters made a slight improvement for my needs, but since the default was so good, not much difference. I'd be inclined to just chuck stuff into KP and use the output. Use LasTools as a pre-step if tiling or such is needed.

KP has logic that modifies contours to match what orienteers expect to see. The only other ways I've created contours is with with OL-Laser and QGIS, and KP contours are much better. I do like the straight "algorithm" contours at lower interval to help find items or understand the terrain, but slope, relief, and hydrology images are usually more helpful. I still create contours with methods other than KP, but I honestly rarely even look at them anymore.

I use QGIS to make contours when all I have is a DEM, not the actual lidar point cloud files. I wonder if KP can accept a DEM as input?

I was slightly worried when I learned that KP modified the contours to "look right" to an orienteer, and that jagge recommended it only for training maps not a basemap for a fieldchecked map. I wondered whether I should work from the "raw, true, literal, absolute" contours. But having looked at both, I've always been very happy with KP contours. Using the "raw" would have meant a lot of work that I doubt would result any better than the KP contours.

Around here KP produces the best contours for gully spur, but the difference isn't great, other than the noise. But for our detailed gold mining terrain its best to go with absolute contours. KP generalises away much of the detail that an orienteer would expect to be mapped. I guess its tuned for Scandinavian glacial terrain.

I did try LAStools in QGIS. There are several long youtube videos out there with Martin showing examples. There seem to be some issues getting it to work, but you should be able to google a solution. I have not played with it in a couple of years.

Read through the Bathhurst presentation on mapping the Bendigo mining terrain. Towards the bottom, they show comparisons for KP and OL-laser contours.

You can do some adjustments in KP if you want more or less pronounced features. With filtering in QGIS I can get contours very similar to KP.

If you are using KP I would still generate absolute contours down to 0.25 or 0.50 meters with a GIS program(QGIS, Global Mapper, or ArcGIS) or OL-laser. These contours will be very helpful for your mapper. I don't always do this as it can be overwhelming and most of my maps are obnoxiously large.

Yes, I like 1 metre grey contours for detailed mining terrain. But as you say, the files are large. Our gully spur terrain doesn't have much of the sub 5 metre height variations that you find in scandinavian terrain and that KP is designed to deal with. I suspect that our more detailed mining terrain is not mappable with the new ISOM without changing what detail we expect to see on the map. I have wondered if KP is trying to tell us Australians something.

Peter, I have a detailed step by step manual for everything I am doing. I add to it every day when I learn a new step, so I will know how to do it next time.

The March 2017 Issue of The Australian Orienteer had a good overview article on the various software programs available.

I still find OCAD 12 to be the easiest program to use. It will generate the small contour intervals. Just remember the quality of the output depends a lot on the quality of the input (grid spacing in the data).

Batch processing in KP failed for me with the default settings. I had to use the alternative complete file path to make it work and then it did not process all the .laz files in the folder or produce the training map I was hoping to get from it. I will play with it some more to see if I can make it work.

For all the OCAD 12 users out there, is there an easy way to convert an OCAD 12 file to an OCAD 6 file for some one using the free downloaded version?

My experience with LAStools in QGIS has not been successful even when I follow the directions completely.

> For all the OCAD 12 users out there, is there an easy way to convert an
> OCAD 12 file to an OCAD 6 file for some one using the free downloaded version?
Save it to OCAD10 then use a copy of OCAD10 to save to OCAD6? But I have to report that the one time I tried a backwards compatible save to version six there were problems coming back the other way after the map corrections had been made. Next time the situation arises I will recommend OOM and OCAD8 format.
Yes, I am still on 10.

Is there a version of OL-Laser with English language menus?

yes

Some notes.

Here is comparison of a man made map, KP map, OL laser and las2iso contours.
http://routegadget.net/3d/pukkamaki/ (you can pan/zoom/rotate in 3D)

Location: https://map.routegadget.net/?basemap=0&lat=60.7363...


With proper filtering and smoothing contours would look much more similar, other than areas where KP has tweaked contour levels. You can spot there places where KP contours agree with human mappers, but strict mathematical 2.5 and even 1.25 contours fail to map, any post filtering of vector contours will not make them magically appear.

As JimBaker wrote I recommended KP map as it is only for training maps, not as a basemap for a fieldchecked map. See, the final KP product has 2.5 or 5m contour interval and it has tweaked true elevations to illustrate some land forms. For mapping you better have for example 0.625 m contour interval, to figure out how to draw contours when you have to move them up or down a bit. KP map can be used as initial suggestion, something to start with, and edit it using higher interval template, shade images and forest visits. But it is not single handed that sufficient as it is for a base map in most terrains.

Note that KP always does also raw 0.3 m interval contours without any level tweaking. You can find them from temp folder. Some use those 0.3m contours as a (raster) template and avoid the need to hit any additional contour generation commands. KP 2.5m contours + 0.3 m contours template is what you get out of the box.

Personally I always run a tiny batch process for small base maps I sometimes make, consisting KP vectors including 0.625 contours and some lastols commands for raster images. Here is example
https://www.facebook.com/groups/485564718218028/pe...

Jagge, the KP seems to show some marsh? How does it do that (or what is it)?

The KP looks closest to the final map, with small differences. I'm wondering how much the mapper took from the other contours sources?

Marsh, roads and such vectors are from national map database (open data). Just rendered on top of the lidar based output. Same stuff you see in mapant.fi.

The mapper did not use KP here at all. The basemap and part of the field checking was made before KP existed. I guess OL Laser was used. So everything was taken from "other contour sources", but "taken" may not be the correct word, because it all is drawn by hand in forest with tablet computer, no contours were taken directly really.

By coincidence I used the next hill SW (still unmapped) as one of my dev test areas. The terrain type being similar (and same scanning) may play little part here, and also the fact the mapper usually actually tries note to overmap but generalizes instead. I am sure there is mappers who might claim the area can't be mapped without stacked form lines and normal symbol sizes.

The point is, better have for example 1m contours for mapping. KP contours are fine as first suggestion, something to start with, possibly slightly better than war 2.5m contours, but that's not that big deal really for field checking when pretty much everything gets hand drawn anyway.

KP's power is elsewhere. I believe to get about the same features mapped with iso contours compared KP contours you need to have about 1 ... 1.5 m contour interval, depending on terrain. Just look at the comparison and figure yourself. But you'll easily get a lot of unwanted stuff too, like lots of useless contours for the interval, and you will have to print it out at larger scale to be able to navigate with it. And if you do so it is just bad for your o technique really - you build habit to use wrong interval and wrong scale. That's why I personally find mapant.fi pretty cool. Sure one could make 1 m iso contours of the whole country and that would be cool too, but cool in different way and usable in different way. Or at least that's how i see it.

How many samples per square meter does the lidar data have?

Finland national dataset point density is about is about 0.7. It is intended for making 2m DEM. Some latest scannings are about 1.0, older data like the one used here is about 0.7 points per m2. Makes it impossible to get vegetation mapped accurately and also hard to get small cliffs mapped. For example if distance between two ground point is 2m or over you can't really detect any lower than 2m cliffs.

Some cities has open data with better point dencity. Helsinki dataset is about 15 pt/m2. You can see KP map made out of it here and comapre it agains the KP map (MapAnt) made from thin national dataset.
https://map.routegadget.net/?basemap=6&lat=60.2254...

Do you find that the higher density yields KP maps that are usable as basemaps? I find that the KP map of my current project, made from ten points per square meter, has nearly perfect contours, and nearly all the cliffs, rock faces and yellow, plus much of the small light green (though the latter is not so common in this area). I haven't found a reason that I'd want a different contour set, so I've been puzzled as to why you think that KP isn't right for basemaps.

I've mapped an area that had a two foot (0.6m) contour basemap. These contours were of very slight use when adjusting contours in the field, but largely the interpolated 5m contours and 2.5m form lines, generated manually at home, were all I wanted. If the contours were wrong, I was changing them by eye anyway. Anything that I wanted from the 2ft contours I could pick out at home. Which I'm guessing, from its description, is more or less what KP is doing algorithmically.

I like reducing manual effort through remote sensing. I once made a map almost from white paper, so bad was the government map, including all the contour shapes and much else. Then some years later an improved government map came out, made from aerial photographs via some algorithm if I understood correctly the government info, showing all my hard worked contours. It's sweet when base materials are really good, reducing effort by a lot over the least helpful base materials.

Left is from 0.7 pt/m2 scanning and right from ~15 pt/m2.



Higher point density usually gives cleaner contours and less debris contours for flat/even areas like mashes with dense tree canopy. More points does not make this directly, it is ground classification issue. With more points ground classification ends up better, less points close to ground but not quite ground gets classified as ground. See, if there is no other low points around withing 10 meters no algorithm can say is it ground or tree trunk/boulder or something else 1m above ground. Making 0.5m contours does not help at all because data just isn't there, you just get more false contours.

When mapping intricate areas like those mining areas or most scandi bare rock hilltops with temptation to use 2.5m interval with several form lines in final map those KP 2.5m contours just don't include enough data to locate details correctly, it is very helpfull to have something like 0.625 contours or shade images. But if size of land forms are suitable for KP contours then those might not give much.

Ok, maybe that's it. I'm in rolling Colorado terrain, 5m contours with some form lines, a bit open, high point density.

KP outputs dxf files with severacrtl different layers that can be imported into OCAD. Does KP publish a master list with the names for each layer? If so where would I find it? I have a list from cedarcreek.

When I hunt for information on CRT it seems like a big mystery.

Just create a new map in OCAD, then "File-Import" and select every single dxf file in the KP and KP temp directory, point to the CRT file I created (or any other CRT), and then save the CRT file OCAD creates afterwards (you just have to hit okay, I think, and check it's saving to a good spot). That output CRT will have all the layer names identified.

Here's the CRT I use.

https://drive.google.com/open?id=1ums2kU3VKpIIULJK...

It's literally this text in a file:

115.000 udepression
206.000 uglydotknoll
112.000 dotknoll
313.000 uglyudepression
103.000 formline
101.000 contour
705.000 depression_intermed
0.0 contour_intermed
102.000 contour_index
707.000 depression_index
704.000 depression

If you are looking for stream hydrology, KP doesn't do that.

Dear cedarcreek:
I must be doing something wrong or missing a step. When I do "File>Import" with OCAD 12 it only allows me to select the dxf files to import and then a CRT file to use (open). After I hit ok it imports the dxf files but it does not generate/output a new CRT file with all the layer names or allow me to save the CRT file.

The only place to generate and save a CRT file that I could fine is from "MAP>Convert Imported layers to Symbols..." tool.

It might be a change from OCAD 11 to OCAD 12. Look in the directories for CRT files---perhaps it doesn't ask, but just saves it anyway.

Read the last paragraph here:

http://www.ocad.com/wiki/ocad12/en/?title=Cross_Re...

At the end of the importing process you have to save another CRT-File. This is the same CRT-File as the loaded one but, in addition, the layer names of objects, which the standard symbol set does not provide, are added with a symbol number 0.0. You can add these symbols to the symbol set and enter the number directly in the CRT-File with a normal Text Editor (e.g. Microsoft Notepad). Then use the Convert Imported Layers to Symbols function to symbolize the Unsymbolized Objects.

That's the same language as the same URL with ocad11 instead of ocad12. So it appears it should be asking you to save a CRT file, unless they've made a change and not documented it.