Demonstration of Proposed QA4EO Geometric Quality Indicators for SAR Products
Schubert, Adrian1; Miranda, Nuno2; Small, David1
1Remote Sensing Laboratories, University of Zurich, SWITZERLAND; 2ESA-ESRIN, ITALY

The Global Earth Observation System of Systems (GEOSS), an initiative by the Group on Earth Observations (GEO), aims to achieve harmonisation and facilitate interoperability between data types and sources. The European Commission has established the Global Monitoring for Environment and Security (GMES) program, whose objective is "to provide, on a sustained basis, reliable and timely services related to environmental and security issues in support of public policy makers' needs" [1].

To help advance the objectives of GEOSS and GMES, the Committee on Earth Observation Satellites (CEOS), the space arm of GEO, defined a quality assurance strategy (QA4EO) to facilitate interoperability of GEO systems [2]. The framework document states that two key elements need to be considered: (1) a communication / data policy, and (2) a means to ascribe Quality Indicators (QIs) to a Knowledge Information product. The QI should be based on a quantifiable metric or statistically based measure.

This document describes and demonstrates several possible QIs (previously outlined in [3]) that may be used to describe the geometric quality of SAR products. Certain QIs are only meaningful for certain product types or geometric projections. Others are intended to be applicable to all slant-range (SR), ground-range (GR), and geocoded ellipsoid-corrected (GEC) products generated from SAR sensors, product types with established track records for a wide range of users.

Presently, no standard geometric quality indicators exist that have been defined in a consistent way for a given product type, across all SAR sensors. The products provided by the major space agencies can be inconsistent with one another in a number of ways:
- Data formats (CEOS mixed-binary including sensor-specific "dialects", XML, mixed ASCII-binary, GeoTIFF, etc.)

- Existence of ground-to-slant-range polynomial (in the case of GR products)
- Annotation of incident angles (if they exist, delivered as polynomial coefficients or a look-up table? Corner coordinates only?)
- Radiometric calibration constants (look-up table? Unambiguous instructions for their calculation?)
- Orbital state vectors (temporal sampling? Attributed accuracy?)
- Raster orientation (quasi-map orientation or slant/ground range geometry?)
- Azimuth timing convention (Doppler centroid? Zero-Doppler azimuth times?)

Geometric QIs, once defined and accepted in the context of QA4EO, would be calculated in standard and publicly documented ways, and annotated in the same way for all SAR products. Their values would be understood in terms of established reference quantities and/or statistical metrics, and would help users determine the suitability of the given product(s) for their particular application.

We selected a small set of several test products from ENVISAT ASAR, ALOS PALSAR and TerraSAR-X that display a range of geometric properties (e.g. different topographical variation, geographic location, etc.). The most important QIs are identified and calculated for the test products. For each QI, a number of characteristics were analysed:

- applicability (e.g. whether it quantifies an instrument- or processor-level feature; whether it applies to all SAR product types or only certain types),
- sensitivity to different quality levels (discrimination ability),
- associated measurement error (reliability), and
- implementation complexity.

A tentative list of the potentially "most useful" QIs is proposed here. The suggested QIs would provide information to a range of users, with information such as the overall expected horizontal deviations, as well as more detailed descriptions of the type(s) and source(s) of the measured errors. We recommend that the QIs applicable to specific products be incorporated in the product annotations themselves in dedicated annotation fields or files. If this is not possible (e.g. if a specific QI calculation takes more time than product processing and delivery, especially in cases where quick delivery is a priority), then an alternative delivery method, such as an online database of QIs for all available products from a given sensor, should be considered.

References:
1. Commission of the European Communities, Global Monitoring for Environment and Security (GMES): From Concept to Reality, Document SEC(2005)1432: 565 final, Brussels, November 10, 2005.
2. Committee on Earth Observation Satellites (CEOS), QA4EO, A Quality Assurance Framework for Earth Observation: Operational Guidelines, v.3.0, July 1, 2009.
3. Schubert A., Small D., Miranda N., "Geometric Quality Indicators for SAR Products," Proc. CEOS SAR Calibration & Validation Workshop, Zurich, Switzerland, 2010.