On the Calibration Strategy for Biomass Mission Retrieval Algorithms
Villard, Ludovic; Le Toan, Thuy

In the frame of Biomass mission (cf. [1]), the calibration of the developed retrieval algorithm depends on reliable ground data in a range of environmental conditions, and most critically in tropical rainforests (cf. [2]). The inversion algorithm to retrieve forest Above Ground Biomass (AGB) is based on the combined used of P-band SAR intensities and Pol-InSAR height. The combination of these two independent measurement sources minimizes measurements or inversion model errors. For both SAR intensities and Pol-InSAR height, the inversion models are based on log regressions with AGB for which calibration plots are required. To account for the discrepancy between model predictions and reference values on test plots, a specific Bayesian formulation is used (cf. [3]). Such strategy raises the question of the number and the accuracy required for these reference plots, both criteria being often antagonist. To optimize the trade-off between the number of calibration plots and the accuracy of their AGB estimation, the error propagation from the test plots to the retrieved biomass is analysed in this paper. Based on in-situ data at the Paracou test site (located in French Guiana) overflown in 2009 during the TropiSAR campaign (cf. [4]), retrieval results have been obtained from LIDAR, Tomographic SAR, Pol-InSAR and intensities. Starting from the most accurate biomass estimation for the test plots (~15% at 1 ha, obtained from in-situ measurements), these estimates are then degraded to various degrees to perform a sensitivity analysis regarding the Bayesian retrieval performance from SAR intensities and Pol-InSAR height. For example, we consider the possibility to use tomographic acquisitions planned for a preliminary phase of Biomass to have a larger coverage of reference AGB data but with larger errors compared to in-situ plots. Likewise, airborne Lidar could also be used to extend the AGB coverage.
Within the scope of the Biomass mission, the results will contribute to define the calibration/validation strategy.

[1] Report for Mission Selection: Biomass. ESA Communication Production Office, May 2012, vol. ESA SP-1324/1 (3 volume series).
[2] T. Le Toan, S. Quegan, M. Davidson, H. Balzter, P. Paillou, K. Papathanassiou, S. Plummer, F. Rocca, S. Saatchi, H. Shugart, and L. Ulander, "The biomass mission : Mapping global forest biomass to better understand the terrestrial carbon cycle," Remote Sensing of Environment, 2011.
[3] Notarnicola, C., and F. Posa, 2004 : Bayesian algorithm for the estimation of the dielectric constant from active and passive remotely sensed data. IEEE Trans. Geoscience and Remote Sensing , 1, 179-183.
[4] P. Dubois-Fernandez, T. Le Toan, S. Daniel, H. Oriot, J. Chave, J. Blanc, L. Villard, M. Davidson, and M. Petit, "The tropisar airborne campaign in french guiana: Objectives, description and temporal behavior of the back-scatter signal," IEEE Transactions on Geoscience and Remote Sensing, 2011.