Recovering Meteosat First Generation spectral response characteristics

In June, Jon Mittaz outlined the origins of the FIDUCEO project and explained why the operational calibration for many Earth Observation sensors introduces biases and errors. One reason for this is that attaining high radiometric accuracy was not an objective for many of the early operational sensors, which had been designed with weather forecasting in mind. In particular, instruments had often not been characterised thoroughly. One aim of FIDUCEO is to improve the calibration of these early observations in order to create new fundamental and thematic climate data records.

The Meteosat First Generation (MFG) satellites have been acquiring a continuous record of Earth observations for more than 30 years. The Meteosat Visible and Infrared Imager (MVIRI) on-board the MFG geostationary satellites acquires radiance twice per hour in a broad reflectance channel referred to as the visible (VIS) band. The temporal frequency of observations allows us to disentangle surface and atmospheric effects and makes the MVIRI data pre-eminently suited for producing long-term data records of surface albedo and aerosol optical depth.

But when they are examined more closely, the surface albedo data records that have been generated from MVIRI observations so far have temporal inconsistencies in them that are due to an inaccurate pre-launch spectral response characterisation. This inaccurate characterisation has restricted the use of calibrated MVIRI data records for climate applications, including retrievals of aerosol optical depth.

FIDUCEO aims to reduce and quantify the uncertainty on the instrument spectral response characterisation in order to facilitate the creation of long-term, consistent, and-high-quality data records of surface albedo and aerosol optical depth for climate applications. Within FIDUCEO, a novel method for recovering the VIS spectral response and its uncertainty is being developed. In addition, the method is aiming to recover and monitor the spectral degradation of the MVIRI VIS response with time.

This recovery method is based on using top-of-atmosphere simulated spectral radiance over pseudo-invariant calibration sites like deep convective clouds, bright deserts, and ocean targets, viewed under different illuminating and atmospheric conditions. We combine these simulated spectral radiance data with other simulated and actual sensor recordings. Inverse modelling methods are then applied to recover the sensor spectral response and ageing characteristics, which include uncertainties and covariance.

The figure above illustrates the spectral response functions from different inverse modelling runs. The models include different Bernstein polynomials with four, five and six basis components to model the pre-launch spectral response and a simple physical ‘grey’ ageing model. Refining the physical ageing model is an ongoing activity. The accuracy of the modelling will be assessed by comparing our test results to the spectral response of the Meteosat Second Generation (MSG) Spinning Enhanced Visible and Infrared Imager (SEVIRI) high-resolution visible (HRV) band, which is essentially similar to the spectral response of the MVIRI VIS band, but has been accurately measured pre-launch.

MVIRI VIS observations have been calibrated vicariously at EUMETSAT using a method that is based on simulated radiance over bright deserts and ocean surface targets, but does not take into account sensor ageing. EUMETSAT will apply the spectral response and ageing characteristics and uncertainties recovered by FIDUCEO, once established, for recalibrating the MVIRI VIS observations and enabling FIDUCEO to generate new fundamental (radiance) and thematic (albedo and aerosol) climate data records from MFG VIS observations.

Add new comment