Proceedings Article | 11 October 2018
KEYWORDS: Vegetation, Agriculture, Imaging systems, Data analysis, Satellites, Image resolution, Spectral resolution, Multispectral imaging, Visible radiation, Near infrared
Sentinel-2 is a constellation of satellites, Sentinel-2A and Sentinel-2B, launched on 23 June 2015 and 7 March 2017 respectively. They carry virtually identical decametric resolution multi-spectral imager (MSI) with 13 spectral channels in the visible, near infrared and short wave infrared (Drusch et al., 2012). Together, these imagers provide better than 5-day revisit of the Earth’s land surfaces. The imagers have been designed to record spectral bands directly related to widely used land surface variables, including vegetation biophysical parameters (Martimort, 2007). The main aim of this research study is to validate the vegetation biophysical parameters estimated from Sentinel-2A data over agricultural sites located in Canada. The validation task was performed for leaf area index (LAI), vegetation fractional cover (fcover) and canopy water content (CWC).
A study area located in Manitoba (Canada), between the latitudes 49.3°N – 49.8°N and longitudes 97.7°W - 98.2°W is selected. For this site, ground-based measurements of vegetation parameters were sampled for 50 fields during the SMAP Validation Experiment 2016 (SMAPVEX16, http://smapvex16-mb.espaceweb.usherbrooke.ca/home.php) field campaign conducted over two 2-week periods from 8–20 June and 10–22 July, 2016.
First, the Sen2Cor processor (Mueller-Wilm et al., 2017) was used to convert the available Sentinel-2 L1C product L2A product (top of canopy reflectance). Then, ESA-SL2P algorithm (Weiss and Baret, 2016) was used to derive the vegetation biophysical parameters (LAI, fcover and CWC) over a study area. A 3x3 pixel moving average filter was applied to the derived maps to minimize uncertainties due to product gridding. First validation results showed an important correlation between Sentinel-2 estimates and ground data for LAI (R2 = 0.77) and fcover (R2 = 0.90). However, a lower correlation was obtained for CWC (R2 = 0.49). In terms of RMSE, the error between Sentinel-2 derived data and ground data is about 0.71 for LAI, 0.10 for fcover and 0.42 Kg/m2 for CWC. These results suggest that for the most part the ESA algorithm falls within specification for retrieval of these parameters for crop types evaluated although improvement for dense canopy conditions should be pursued.
References
1. M. Drusch, U. Del Bello, S. Carlier, , O. Colin., V. Fernandez, F. Gascon, B. Hoersch, C. Isola, P. Laberinti, P. Martimort, A. Meygret, F. Spoto, O. Sy, F. Marchese, and P. Bargellinid, “Sentinel-2: ESA's Optical High-Resolution Mission for GMES Operational Services”, Remote Sens. Environ, vol.20, pp.25–36, 20, 2012.
2. P. Martimort, M. Berger, B. Carnicero, U. Del Bello, V. Fernandez, F. Gascon, et al. “Sentinel-2: The optical high-resolution mission for GMES operational services”, ESA Bulletin, vol. 131, pp. 18–23, 2007.
3. U. Mueller-Wilm, O. Devignot, L. Pessiot, “Sen2Cor Configuration and User Manual. S2-PDGS-MPC-L2A-SUM-V2.4”, Issue 1, 2017 [http://step.esa.int/thirdparties/sen2cor/2.4.0/ Sen2Cor_240_Documenation_PDF/S2-PDGS-MPC-L2A-SUM-V2.4.0.pdf].
4. M. Weiss and F. Baret, “S2ToolBox Level 2 products”, Version 1.1, Date Issued: 02.05.2016, [step.esa.int/docs/extra/ATBD_S2ToolBox_L2B_V1.1.pdf].
