Hyperspectral Imaging Lidar

UCL-MSSL with Lidar Technologies Ltd

Fast Track Project

The main scientific driver for the development of an Imaging LIDAR system is to improve knowledge of the Earth’s carbon cycle through better measurement of biomass from space. In particular, measurement of canopy height and canopy cover/fraction above sloping ground (strongly related to biomass) was chosen as the main scientific application. This driver resulted in a requirement for a 1m ranging accuracy from an orbit around 350-400km with a 30m LIDAR footprint and the ability to identify individual tree crowns so that measurements or extrapolation could be performed across a wide area.

The approach used to design a putative space-borne or airborne LIDAR consisted of a combined simulation system based on a pre-existing LabView application developed at LTL, interfaced to a scene simulation system based on Monte Carlo ray-tracing developed at UCL.

An initial assessment was made of the potential of new Geiger mode Avalanche Photo-Diode (APD) detector imaging array technology. However, employing the full LIDAR simulator, it was found that, with a low mass (<150kg), low power laser at the required orbital altitude, insufficient photons would be returned, so that the Geiger mode APD array could not be used. A revised concept was produced, consisting of a profiling LIDAR coupled in the same optical path as a stereoscopic imager. The stereoscopic imager would allow the canopy top height to be determined, with the profiler providing 2D height slices running down the middle of the image. The two together enable an extrapolation of tree height and tree cover across the scene. The stereoscopic imager allows a more detailed Digital Elevation Model to be built over vast areas of forest cover. To meet the requirement to measure tree height over sloping terrain (with typical slopes of 30º) with a 30m footprint LIDAR, a 2-wavelength system (either 425/850nm or 532/1064nm) was selected. The two wavelengths were chosen so that the lower wavelength was sensitive to soil and the upper wavelength to vegetation. Simulation studies were performed which showed that a fairly reliable and robust retrieval should be feasible dependent on tree spacing and crown closure/density. Hyperspectral sampling using tunable filters was envisaged for measuring both reflectance and fluorescence from a multispectral laser’s interaction with a vegetation canopy. Two technologies were tested at UCL-MSSL: Liquid Crystal (LCTF) and Acoustic-Optical (AOTF). These technologies were demonstrated as being “fit for purpose” for sampling with spectral bands as narrow as a few nm in milliseconds. However, only AOTF is suitable for spaceborne applications, as LCTFs are extremely temperature sensitive and insufficiently robust.Simulations were also performed of sensitivity to tree crown density (fractional vegetation cover), from which it appears that LIDAR can estimate biomass over a much larger range than P-band SAR but this is still to be quantified. In fact the LIDAR was shown to be sensitive to canopy-top height up to fractional covers of 65% and could obtain reasonable results for certain Sitka Spruce densities up to 98%.The final outline design for an imaging LIDAR system consisted of a 1m telescope, with a LIDAR weighing < 15kg, a 1m mirror system around 100kg with all the electronics components attached to the back of the mirror system. In the same optical path, the stereoscopic imager provides detailed information the tree crowns, their spacing and density.

The work is led by Prof Jan-Peter Muller, UCL-MSSL with Lidar Technologies Ltd