End-to-End Simulation for a Forest-Dedicated Full-Waveform Lidar Onboard a Satellite Initialized from Airborne Ultraviolet Lidar Experiments

Abstract

Abstract:In order to study forests at the global scale, a detailed link budget for a lidar system onboard satellite is presented. It is based on an original approach coupling airborne lidar observations and an end-to-end simulator. The simulator is initialized by airborne lidar measurements performed over temperate and tropical forests on the French territory, representing a wide range of forests ecosystems. Considering two complementary wavelengths of 355 and 1064 nm, the end-to-end simulator computes the performance of spaceborne lidar systems for different orbits. The analysis is based on forest structural (tree top height, quadratic mean canopy height) and optical (forest optical thickness) parameters. Although an ultraviolet lidar appears to be a good candidate for airborne measurements, our results show that the limited energy is not favorable for spaceborne missions with such a wavelength. A near infrared wavelength at 1064 nm is preferable, requiring ~100 mJ laser emitted energy, which is in agreement with current and future spaceborne missions involving a lidar. We find that the signal-to-noise ratio at the ground level to extract both the structural and optical parameters of forests must be larger than 10. Hence, considering the presence of clouds and aerosols in the atmosphere and assuming a stationary forest, a good detection probability of 99% can be reached when 4 or 5 satellite revisits are considered for a lidar system onboard the ISS or ICESat, respectively. This concerns ~90% of forest covers observed from the lidar, which have an optical thickness less than 3. OPEN ACCESSRemote Sens. 2015, 7 5223

Tropical Forests of Réunion Island Classified from Airborne Full-Waveform LiDAR Measurements

“…The standard deviation of LiDAR-derived CH was assessed to be~1.5 m when only considering measurement noise and signal processing errors. Shang and Chazette [26] assessed that LiDAR signal distortion due to the surface slope can lead to a relative CH uncertainty of~5% for a slope of 30˝(see Table 1) and a 10-m footprint, as is the case in our present study. As a result, the standard deviation of our LiDAR-derived CH should be of the order of~2 m. We do not consider geolocation errors in this statistical study.…”

“…All of this mathematical development can be used whatever the wavelength, but for visible or infrared wavelengths, it may be necessary to consider the multiple scattering effect due to leaves and branches [26]. The multiple scattering enhances the backscatter LiDAR signal and makes the LiDAR signal distorted.…”

“…As recommended by several authors [20,[26][27][28][29], a large LiDAR footprint is preferred so that the laser can better penetrate the dense tropical forests. Shang and Chazette [26] estimated an optimal laser footprint diameter around 20 m for dense forests, whereas Riaño et al [30] found that the Leaf Area Index (LAI) was better estimated using laser footprints between 7.5 and 12.5 m. The ULICE system was thus modified to obtain a large and controllable Remote Sens.2016, 8, 43 4 between 7.5 and 12.5 m. The ULICE system was thus modified to obtain a large and controllable sounding area on the ground (approximately a 1-m to 10-m footprint diameter for a flight altitude of ~350 m above the ground level (agl)). The effect of LiDAR footprint size will be studied in Section 2.4.…”

“…The backscatter to extinction ratio BER is a classical parameter used in LiDAR analyses [26,36], which characterizes the probability that an intercepted photon would be backscattered by a scattering layer; it is assumed to be constant for all canopy levels in this study. The canopy extinction coefficient α canopy (h) is defined as the sum of the absorption and scattering coefficients in the canopy.…”

“…Nevertheless, the reflectance ratio can be estimated using only LiDAR measurements as follows. Assuming FOT(0) >> 1 (i.e., ε « 1), which is realistic for thick tropical forests, as highlighted by Shang and Chazette [26], an initial FOT estimator can be evaluated from Equations (3) to (6), and is given for h > 0 by:…”