Special session: Towards universal mapping models - enhancing the spatial and temporal transferability of empirical and physical models
| Chair: | Sebastian van der Linden, Humboldt-Universität zu Berlin, Germany, sebastian.linden@hu-berlin.de |
| Co-Chair: | Jochem Verrelst, Universitat de Valencia, Spain, jochem.verrelst@uv.es; |
| Akpona Okujeni, Humboldt-Universität zu Berlin, Germany, akpona.okujeni@geo.hu-berlin.de |
Both empirical models based on statistical learning and physical models relying on radiative transfer theory are used to derive quantitative information on land cover or biophysical variables from imaging spectroscopy data. Based on the abundance and quality of the expert knowledge that is used to parameterize the models - e.g. field measurements or spectral libraries - high accuracies can be achieved. However, applications of such models are mostly restricted to local to regional scales or to single ecosystems/biomes. The expert knowledge underlying the model parameterization is often only locally valid. In addition, readily parameterized models are rarely transferred to other regions and concepts for modeling approaches do not strive for universal applicability.
With forthcoming global imaging spectroscopy missions the need for more universally applicable models and for portability at all scales becomes strictly necessary. This special session addresses the challenges for achieving an increased spatial and temporal transfer of expert knowledge, models and concepts. We invite contributions covering empirical and physical approaches that focus on an application oriented development of universal mapping models. Contributions may, for example, cover the harmonization or optimization of expert knowledge/spectral libraries/data sets, the derivation of spatially and temporal invariant features, model portability or scalability, etc.
Keywords: Imaging spectroscopy, empirical and physical modeling, model transferability, universal mapping
Back to special session overview
