DART: Recent Advances in Radiation Budget, Atmospheric Radiative Transfer, Thermal Emission and Error Propagation

New satellite instruments and methodologies for their data processing and Cal/Val activities are continuously evolving. To interpret correctly a broad suite of various Earth observations, one requires increasingly accurate radiometric measurements and models. In this context, sophisticated physical models that simulate the radiative transfer in the Earth-Atmosphere system are essential. More accurate and simultaneously more efficient modelling is becoming increasingly feasible thanks to recent advances in numerical solutions of radiative transfer equations, 3D reconstructions of land surfaces and clouds, and advanced computer technology. Discrete Anisotropic Radiative Transfer (DART, https://dart.omp.eu/#/) is one of the most comprehensive 3D radiative transfer models used by the remote sensing community. Developed at CESBIO since 1992, it simulates remote sensing observations and radiation budget for any 3D natural and urban land surfaces with topography and atmosphere, from visible to thermal infrared domains. DART is used by various space agencies (e.g., CNES, ESA, and NASA) to support the preparation of future satellite missions, such as Sentinel-2 NG, LSTM, and TRISHNA. A number of research institutes use DART for a wide range of remote sensing applications, including leaf traits’ retrievals, surface variables’ inversions, and sensitivity analysis. In this contribution, we present recent DART advances introduced by the DART team (CESBIO, France) and collaborating institutes, with a focus on applications in quantitative remote sensing, specifically:

1. General principles of the DART model, including its physical basis, algorithms, products, and simulations’ accuracy.

2. Recent developments, e.g., Jacobian matrices (error propagation), machine-learning emulation of atmospheric radiative transfer, spherical-shell atmosphere, Monte Carlo radiation budget, mobile LiDAR, night-time imaging, large-scale thermal imaging, etc.

3. Applications of the DART model in quantitative remote sensing, including sensitivity analysis, inversion of land surface variables, training of machine learning models, correction of directional effects, and preparation of future satellite missions.

POSTER