Fiducial Reference Measurements for validation of Sentinel-2 and Proba-V surface reflectance products

Abstract

Many derived Earth Observation products share surface reflectance as a common step in their processing chains. This makes the maintenance and improvement of surface reflectance product quality of fundamental importance to ensure information derived from these downstream products can be trusted. Despite this, the literature is relatively light on the implementation of validation methodologies designed for surface reflectance. In response to the need to improve general EO validation methodologies, the concept of fiducial reference measurements (FRM) was created that would produce validation data that is fully characterised and independent, with associated uncertainties, and traceable to SI.This paper describes a field campaign designed to produce surface reflectance validation data, in line with the FRM ideology, for validating the Sentinel-2 L2A (S2) and Proba-V S1-TOC (Proba) surface reflectance products. The key methodological procedures are outlined in detail to facilitate uptake of FRM methods in future validation studies. These include the calibration and characterisation of field instruments, the field measurement protocol, and uncertainty propagation.Comparison of S2 with field measurements showed agreement within the stated uncertainties present for all bands and locations at the pixel and area scales. The Proba comparison results demonstrated a general disagreement within the stated uncertainties. However, there is a lack of uncertainty information provided by the product as well as publicly available product uncertainty requirements which mean that it is difficult to assess this fully. The aim of this practical demonstration of FRM-based surface reflectance validation, that utilises metrological practices for uncertainty characterisation, is to encourage the adoption of these procedures in future land surface reflectance validation studies and operational activities.