Three-dimensional (3D) Light Detection And Ranging (LIDAR) systems designed for foliage penetration can produce
good bare-earth products in medium to medium-heavy obscuration environments, but product creation becomes
increasingly more difficult as the obscuration level increases. A prior knowledge of the obscuration environment over
large areas is hard to obtain. The competing factors of area coverage rate and product quality are difficult to balance.
Ground-based estimates of obscuration levels are labor intensive and only capture a small portion of the area of interest.
Estimates of obscuration levels derived from airborne data require that the area of interest has been collected previously.
Recently, there has been a focus on lacunarity (scale dependent measure of translational invariance) to quantify the gap
structure of canopies. While this approach is useful, it needs to be evaluated relative to the size of the instantaneous
field-of-view (IFOV) of the system under consideration. In this paper, the author reports on initial results to generate not
just average obscuration values from overhead canopy photographs, but to generate obscuration probability density
functions (PDFs) for both gimbaled linear-mode and geiger-mode airborne LIDAR. In general, gimbaled linear-mode
(LM) LIDAR collects data with higher signal-to-noise (SNR), but is limited to smaller areas and cannot collect at higher
altitudes. Conversely, geiger-mode (GM) LIDAR has a much lower SNR, but is capable of higher area rates and
collecting data at higher altitudes. To date, geiger-mode LIDAR obscurant penetration theory has relied on a single
obscuration value, but recent work has extended it to use PDFs1. Whether or not the inclusion of PDFs significantly
changes predicted results and more closely matches actual results awaits the generation of PDFs over specific ground
truth targets and comparison to actual collections of those ground truth targets. Ideally, examination of individual PDFs
for specific collections will provide insight into how collection operations can be optimized in general and whether or
not a generation of representative PDFs of various forest types will be useful for collection planning.
|