The performance of the dark spectrum fitting (DSF) atmospheric correction algorithm is evaluated using matchups between metre- and decametre-scale satellite imagery as processed with ACOLITE and measurements from autonomous PANTHYR hyperspectral radiometer systems deployed in the Adriatic and North Sea. Imagery from the operational land imager (OLI) on Landsat 8, the multispectral instrument (MSI) on Sentinel-2 A and B, and the PlanetScope CubeSat constellation was processed for both sites using a fixed atmospheric path reflectance in a small region of interest around the system&\#x2019;s deployment location, using a number of processing settings, including a new sky reflectance correction. The mean absolute relative differences (MARD) between in situ and satellite measured reflectances reach <20&\#x0025; in the Blue and 11&\#x0025; in the Green bands around 490 and 560 nm for the best performing configuration for MSI and OLI. Higher relative errors are found for the shortest Blue bands around 440 nm (30&\#x2013;100&\#x0025; MARD), and in the Red-Edge and near-infrared bands (35&\#x2013;100&\#x0025; MARD), largely influenced by the lower absolute data range in the observations. Root mean squared differences (RMSD) increase from 0.005 in the NIR to about 0.015&\#x2013;0.020 in the Blue band, consistent with increasing atmospheric path reflectance. Validation of the Red-Edge and NIR bands on Sentinel-2 is presented, as well as for the first time, the Panchromatic band (17&\#x2013;26&\#x0025; MARD) on Landsat 8, and the derived Orange contra-band (8&\#x2013;33&\#x0025; MARD for waters in the algorithm domain, and around 40&\#x2013;80&\#x0025; MARD overall). For Sentinel-2, excluding the SWIR bands from the DSF gave better performances, likely due to calibration issues of MSI at longer wavelengths. Excluding the SWIR on Landsat 8 gave good performance as well, indicating robustness of the DSF to the available band set. The DSF performance was found to be rather insensitive to (1) the wavelength spacing in the lookup tables used for the atmospheric correction, (2) the use of default or ancillary information on gas concentration and atmospheric pressure, and (3) the size of the ROI over which the path reflectance is estimated. The performance of the PlanetScope constellation is found to be similar to previously published results, with the standard DSF giving the best results in the visible bands in terms of MARD (24&\#x2013;40&\#x0025; overall, and 18&\#x2013;29&\#x0025; for the turbid site). The new sky reflectance correction gave mixed results, although it reduced the mean biases for certain configurations and improved results for the processing excluding the SWIR bands, giving lower RMSD and MARD especially at longer wavelengths (>600 nm). The results presented in this article should serve as guidelines for general use of ACOLITE and the DSF.
Located in
Library
/
RBINS Staff Publications 2020
Remote sensing of Land Surface Temperature (LST) generally requires atmospheric parameters and the emissivity (∊) of the target to be estimated. The atmospheric up- and downwelling radiances and transmittance can be accurately modelled using radiative transfer models and profiles of relative humidity and temperature, either measured by radiosonde probes or retrieved from assimilating weather models. The estimation of ∊ is a large source of uncertainty in the resulting LST product, and there are various approaches using multi-angle observations, multispectral optical or multispectral thermal infrared imagery. In this paper, the estimation of LST from the Thermal InfraRed Sensor (TIRS) on board Landsat 8 is evaluated using more than 6 years of in situ temperature measurements from a network of 14 Autonomous Weather Stations (AWS) in Belgium. ∊ is estimated from concomitant atmospherically corrected imagery from the Operational Land Imager (OLI) using two new neural network approaches trained on ECOSTRESS spectra, and an established NDVI based method. Results are compared to using ∊=1 and the ASTER Global Emissivity Dataset. LST retrievals from L8/TIRS perform well for all emissivity data sources for 500 matchups with AWS subsoil temperature measurements: Mean Differences 0.8–3.7 K and unbiased Root Mean Squared Differences of 2.9–3.5 K for both B10 and B11. The use of unity emissivity gives the best results in terms of MD (0.8 K) and unb-RMSD (3 K). Similar ranges of unb-RMSD are found for 500 matchups with broadband radiometer temperatures (2.6–3.1 K), that have lower absolute MD values (−2.2–0.6 K). For the radiometer temperatures, both the neural net approaches gave lowest MD, in the best case ±0.1 K. The present investigation can hence recommend the neural nets to derive ∊ for the retrieval of LST over the AWS in Belgium. Using published matchup results from other authors however, no single source of ∊ data performed better than ∊=1, but this could be due to their low number of matchups. Further efforts for estimating representative pixel average emissivities are needed, and establishing a denser in situ measurement network over varied land use, with rather homogeneous land cover within a TIRS pixel, may aid further validation of a per pixel and per scene ∊ estimates from multispectral imagery. AWS data seems valuable for evaluation of satellite LST, with the advantage of a much lower cost and higher potential matchup density compared to conventional radiometers.
Located in
Library
/
RBINS Staff Publications 2020
