LiDAR’s are just fantastic for mapping forests. Airborne LiDARs are now routinely used for mapping the structural properties of forests – canopy height, number density, gap fraction, understory fuel load, etc. They can do it over large areas and in remote places. And the ICESat satellite has also shown that you can even do this from space.
So, what is the next big thing in forest mapping? We think it is colour LiDAR.
In recent years, there has been growing interest in combing single wavelength LiDAR with hyperspectral imagery – structure information from the LiDAR and spectral information from the imager – a great combination. This has some limitations though: it’s a difficult (but not impossible) process to merge the LiDAR cloud of points with the pixel-based imager; and the spectral information is dominated by the solar reflection at the top of the canopy.
An alternative is to get the spectral information from the LiDAR directly, using different wavelengths of laser light, simultaneously. The idea of using two wavelengths in combination is not new– the traditional bathymetric LiDAR is based on this principle (one NIR and one green LiDAR). But these dual-wavelength LiDARs are not designed to get spectral information (the NIR measures the water surface, the green penetrates through the water).
To get proper spectral information from a LiDAR is more of a challenge. You have to make sure the beams properly overlap and are measured simultaneously. You have to maintain consistent calibration across the different wavelengths. And you have to be able to interpret the signal, a non-trivial exercise in itself.
But if all these technical challenges can be overcome the benefits are enormous. You get spectral information everywhere the LiDAR can measure, not just the top of the canopy. There are no more problems with co-registering the point cloud with the spectral imagery. And perhaps most interesting of all, you can get spectral information of the understory vegetation – no other airborne instrument can do that.
This technique is variously referred to as dual-wavelength, multi-spectral or hyper-spectral LiDAR. I am just calling it colour LiDAR, for now, to avoid semantics.
Here in Edinburgh, we believe that colour LiDAR will be the next big thing in forest mapping ( Woodhouse et al, 2011). We are so confident, in fact, that we spun-out Carbomap as a commercial venture so that when the technology is fully developed we will be able to provide that service to whoever needs it.
Edinburgh is not the only group working on this – there is a group in Finland, one in Japan and NASA Goddard are also active in this subject, but we think Edinburgh and Carbomap are the only group with a combination of academic research, commercial enterprise and a mission concept for a spaceborne version of the same system (the SpeCL mission).
One final note: there is one other type of LiDAR system that has come on the scene in the last few years and is generating lots of excitement, and is considered by some to be the next big thing in LiDAR – photon counting (PC).
Photon counting LiDARs are just amazing. They are so sensitive that they can detect individual photons scattered back to the sensor. That’s right, they measure a return signal that may be composed of only a single photon – quite remarkable and engineers love it. The high level of sensitivity means that you can fly higher and so have larger coverage, so it can be cheaper. It also makes it easier to fly from space (the next ICESat mission from NASA will be a photon counting system and ESA has prioritised the development of PC LiDAR for space).
But from our perspective, PC LiDAR is not the solution for forest mapping. Colour LiDAR relies on looking at the relative contribution in different wavelengths of the same part of the canopy. If you only have a handful, or even a bucketful, of photons being returned at each wavelength you have a problem. For one thing, the photons are likely to come from different parts of the canopy. And for another, if you only have a few photons, you don’t have a measure of spectral intensity – you need lots and lots of photons throughout the full depth of the canopy before you can start looking at vertical structure and spectral reflectivity.
So, what do you think is the next big thing in LiDAR for forestry applications? Do we go down the full waveform, spectral LiDAR, and get the equivalent of an MRI scanner for forests? Or do we go for photon counting, with its limited information but larger coverage?
Let us know what you think.
Pre-publication proof (no account needed): A Multispectral Canopy LiDAR Demonstrator Project
Woodhouse, Iain H.; Nichol, Caroline; Sinclair, Peter; et al. A Multispectral Canopy LiDAR Demonstrator Project IEEE Geoscience and Remote Sensing Letters, 8(5) 839-843, SEP 2011, DOI: 10.1109/LGRS.2011.2113312
Morsdorf, F; Nichol, C; Malthus, T; et al. Assessing forest structural and physiological information content of multi-spectral LiDAR waveforms by radiative transfer modelling Remote Sensing of Environment, 113 (10): 2152-2163, OCT 2009, DOI: 10.1016/j.rse.2009.05.019
Jack, Jim; Rumi, Emal; Henry, David; et al. The design of a Space-borne Multispectral Canopy LiDAR to Estimate Global Carbon Stock and Gross Primary Productivity, Sensors, Systems, and Next-Generation Satellites Xv, Vol 8176 , 2011, DOI: 10.1117/12.898166