I'd imagine its in the region of +/-10 cm - with no vegetation you might get very good results. Also, it depends what make of lidar system you use. I am dealing with terrestrial lidar, and while Leica system provides about +/-1 cm (though they claim to give sub-centimeter accuracy), Riegl provides acccuracy of about +/-2 cm - when scanning flat surface it is always a "fuzz" in the cloud data.

Thanks, your answer agrees with that of some other experts.

Is also important to consider stuff like the variation of height, if there is any tilt, etc. Measurements require further correction. The correction though would vary on the flight specifications. That is way is so important having a well trained pilot to do such studies.

Gabriela, thanks, I agree with you completely.

In an unpublished study (with a company called AAMHAtch; and before engaging in the purchase of 1 million hectares of LiDAR data) we found ALS (ie LiDAR) is the most cost effective, and reliable method to derive DTMs based upon a trial project we ran in one of our wetlands. The level of ground detail generated from the ALS surface is superior in defining subtle drainage patterns and although reduced, provides more accurate surface definition within the heavily vegetated areas when compared to the photogrammetric DTMs We generated auto-correlated DTMs at 1, 5 and 10m grid intervals, all three DTMs showed distinct “channels” within the surface model. The overall terrain shape and various thick vegetation within project area is not conjucive to a softcopy auto–correlated DTM solution. Airborne Laser Scanning (ALS) is adopted to achieve 0.15m vertical accuracy based upon the results from the trial area. This is all based on dry ground. Anything within-channel or within a wetland where there was water, the problems with light dispersal made the technology unusable. When we enagaged in the purchase of LiDAR data tilt was an issue. The provider (Fugro) had software to correct for the tilt affect during data acquisition.

The companies promotional material after the trial is attached

Dr. Qassim A. Abdullah, PLS, CP covers this topic in exquisite detail monthly in his "Mapping Matters" column in Photogrammetric Engineering and Remote Sensing journal. Such material is available for free to the public at www.ASPRS.org. I suggest you join the American Society for Photogrammetry and Remote Sensing, as refereed papers appear all the time in the monthly journal on the topic of LIDAR. The Society also publishes bound books on the topic, and has also published LIDAR Standards and Specifications.

Nice to see this discussion, e.g. i'm not the only one who scratches his head. I've been been involved with Airborne LiDAR for mapping snow depth in mountainous terrain. In my in press publication, I had to try to back-calculate these errors (http://onlinelibrary.wiley.com/doi/10.1002/hyp.9618/abstract) because none were provided with the data. I found errors typcially less than .22m for individual scans. These include vegetated areas however and the results agreed with Tinkham et al (http://www.mdpi.com/2072-4292/3/3/638) who do a more careful analysis by cover type though not getting into slope.

There are on-going efforts, namely the Airborne Snow Observatory, where TLS and airborne are being used in conjunction to try to understand how airborne errors depend on terrain and scan geometries... and of course, vegetation. The first season of observations is about halfway through, results should be forth coming in the next year.

You can read this paper about Lidar vertical error modelling.

If you need additional information, please tell me.

http://www.ofertacientifica.ual.es/descargas/investigacion/documentos/RNM368_101.pdf

I have added our committee report on this subject at

https://www.researchgate.net/publication/236741945_Developing_a_Digital_Terrain_Model_for_modelling_floods_in_the_Macquarie_Marshes

We also got high accuracy, but the effect of vegetation in a low relief environment is very significant