Abstract
Because of the critical role they play in Earth observation workflows, radiative transfer models (RTM) must undergo thorough validation campaigns that will help guarantee that their output is representative of the physical reality. In this study, we test the ability of the Eradiate RTM to simulate the reflectance of a manufactured target given SI-traceable measurements of its shape and optical properties. To address issues identified in similar prior work, we select a material with reflective features that can be accurately modelled using a straightforward data-driven approach. We produce an artificial target design that is easy to manufacture with good precision. The material and artefact are characterized optically using an SI-traceable 3D goniospectrophotometer, and the artefact is characterized geometrically in an SI-traceable facility. The geometry of the artificial target and the optical properties of the material are used to build a numerical experiment with Eradiate that simulates the optical characterization of the artificial target. The simulation includes the propagation of uncertainties on the input parameters. We compare the simulated and measured data and analyse the performance of this method using three different metrics. Our findings demonstrate consistent performance across all considered illumination and sensor pointing configurations. Simulations deviate from measurements by less than 2 %, and more than 80 % of measured and simulated data points agree.
Original language | English |
---|---|
Number of pages | 22 |
Journal | IEEE Transactions on Geoscience and Remote Sensing |
Volume | 63 |
DOIs | |
Publication status | Published - 2025 |
MoE publication type | A1 Journal article-refereed |
Keywords
- Accuracy
- bidirectional reflectance factor (BRF)
- data-driven model
- model validation
- Monte Carlo (MC) methods
- radiative transfer (RT)
- ray tracing
- SI traceability
- uncertainty