FD-3: Optical Remote Sensing
Presented by: Jose F. Moreno
University of Valencia, Spain
Quantitative and spatially resolved information about land surface characteristics is necessary for a large variety of applications. Inversion of surface/atmosphere radiative transfer models against Earth Observation data is considered as an optimum approach for the retrieval of bio/geo-physical parameters because of being physically-based and of general applicability. However, this approach is considered complex and laborious by the broader users community because of its multiple processing steps and expert knowledge is required to realize precise model parameterization and inversion. Moreover, access to the right models is required, such as turbid medium models when interpreting homogenous land covers (e.g. agricultural lands) or 3D models when interpreting more heterogeneous land covers (e.g. forests). On the other side, the usage of advanced model inversion techniques becomes limited by the pre-processing of the data, such as atmospheric corrections. The current tendency to use temporal series of data, with increasing spatial resolution, makes necessary to guarantee temporal consistency in the retrievals, and statistical validation tools to establish uncertainties in all the derived products. This tutorial will review all such aspects of the usage of Optical Remote Sensing data, mostly focused on Land Science and Applications.
The first step for a proper analysis of optical remote sensing data is to have a good understanding of the actual information content of the data, for which forwards models are required. This tutorial review different modelling approaches, and the current status of the understanding of information content is illustrated by the capability to match forward model simulations with actual measured data (top-of-atmosphere radiances of top-of-surface reflectance). Innovative types of information, such as chlorophyll fluorescence techniques, are also reviewed, as they provide new type of information not yet fully exploited.
But before extracting useful information from the data, proper data processing aspects must be taken into account. This implies radiometric calibration, geometric corrections, cloud screening, compensation of atmospheric and topographic effects, and normalization of angular variability due to changes in illumination/observation geometry. Special interest represents the analysis of particularly relevant type of data which demands specific methodologies, such as imaging spectrometer data or optical very high spatial resolution data. Imaging spectrometer data, with many spectral bands and high spectral resolution, are especially useful to map the chemistry of the surface, but special data processing aspects are required for spectral calibration and precise atmospheric corrections. In the case of very high spatial resolution data, some effects that can be neglected in lower resolution data must be considered with special care (like topographic effects, atmospheric adjacency, atmospheric refraction). All these methods will be reviewed and methodologies provided to address each one of the needed pre-processing steps and corrections before data usage in models and applications.
The third part of the tutorial will cover the exploitation of optical remote sensing data for land science (i.e., inputs to land surface processes models, surface-atmosphere interaction models, vegetation dynamics/carbon models) and applications (agriculture, forestry, management of natural resources). Techniques for the retrieval of bio-/geo-physical parameters about the surface are introduced, covering both statistical methods and regression-based models through spectral indices, to model inversion strategies and data assimilation approaches using spatial information and temporal series of data. Error estimation for the resulting products are provided, together with validation approaches based in intercomparison of products and usage of ground measurement networks.
Finally, as a practical approach on how to use such methods with actual data and models, several modelling and information retrieval tools are presented. To support operational inversion-based bio/geo-physical parameter mapping, a new toolbox called ARTMO: "Automated Radiative Transfer Models Operator" has been recently developed, freely available to the scientific community. This toolbox, written in Matlab, brings together several radiative transfer models within one graphical user interface, allowing to choose between various models with varying complexity (from simple 1D turbid medium up to 3D ray-tracing models), spectral band settings of various air- and space-borne sensors, and retrievals schemes running model inversion against optical imagery given several cost options and error estimates. Several applications are illustrated using ARTMO toolbox and data from Landsat, CHRIS/PROBA, MERIS, and other sensors.
Jose F. Moreno obtained the PhD in Theoretical Physics from the University of Valencia, where he is presently teaching at the Faculty of Physics, as Professor of Earth Physics, and head of the Laboratory for Earth Observation of the Image Processing Laboratory-Scientific Park, University of Valencia. He is working on different national and international projects related to remote sensing and space research, with focus on the development of data processing algorithms for modelling and monitoring land surface processes, with special interest in multisource data integration and numerical techniques for model inversion and data assimilation applied to remote sensing data. During 1995-1996 he was a visiting scientist at the NASA/Jet Propulsion Laboratory in Pasadena, California, USA. He has been involved in several European research networks, such as the European Network for the development of Advanced MOdels to interpret Remote Sensing data over terrestrial environments (ENAMORS), and the European Radar-Optical Research Assemblage (ERA-ORA) network, an application projects such as AIMWATER (Analysis, Investigation and Monitoring of Water resources) and DEMETER (DEMonstration of Earth observation TEchnologies in Routine irrigation advisory services).
Author of many publications in the field, including several book chapters, Prof. Moreno has served as Associate Editor for IEEE Transactions of Geoscience and Remote Sensing (1994-2000) and guest editor for the IEEE Journal of Selected Topics in Earth Observations and Remote Sensing, and was a member of the European Space Agency Earth Sciences Advisory Committee (1998-2002), the Space Station Users Panel (SSUP) and other advisory committees for ESA, European Commission, European Science Foundation, Eumetsat and NASA. Prof. Moreno was a member of the writing team for "The Changing Earth: New Scientific Challenges for ESA's Living Planet Programme" and is member of several international societies and scientific committees for international conferences and workshops. He has participated actively in the design and development of several remote sensing experiments in preparation of future missions, and the preparatory activities for the Sentinel missions within the GMES programme (AgriSAR for Sentinel-1, SEN2FLEX for Sentinel-2 and SEN3EXP for Sentinel-3), and actively involved in the exploitation of data from current missions, such as MERIS/ENVISAT and CHRIS/PROBA. He is also involved in the Spanish mission SEOSAT within the ESA GMES programme, as chairman of the SEOSAT/Ingenio Mission Advisory Group. He coordinated the proposal for the Fluorescence Explorer (FLEX), a candidate ESA Earth Explorer Mission, and currently is chairman of the ESA FLEX Mission Advisory Group. He is also a member of the NASA HyspIRI International Science Group.
Prof. Moreno is teaching regularly Masters and PhD courses on Remote Sensing in several universities, and regular courses for students of Physics and Engineering (Numerical and Statistical Methods in Physics, Fluid Mechanics, Experimental Techniques in Physics). He has been a lecturer in several editions of the ESA Advanced Training Course on Land Remote Sensing.