A New Methodology to Process the Total Solar Irradiance observations Using Machine Learning and Data Fusion

Abstract:

Across the last decades, various space missions have measured the
total solar irradiance (TSI) such as the Variability of Irradiance and
Gravity Oscillations (VIRGO) experiment on the Solar and Heliospheric
Observatory (SOHO) starting in 1996. Since the beginning of its
recording time, one challenge is to correct the measurements from the
degradation of the TSI sensors in space. Various groups
have proposed different methodologies to produce a continuous TSI time
series (TSI composite) which is essential to monitor the sun activity
and its influence on the Earth’s climate.

However, the benchmark to test all those solutions is source of a
debate in the community. Moreover, the input data for the TSI composite
are the degradation-corrected measurements provided by each individual
instrument team. Here, we propose a different approach using a
machine learning and data fusion algorithm to produce automatically the
degradation-corrected TSI time series based on a small number of generic
assumptions. The algorithm is applied to the VIRGO/PMO6, VIRGO/DIARAD
and PREMOS/PMO6 data. The time series agree between each other in terms
of mean value with a difference of ~ 0.14 W/m2 (PREMOS), ~ 0.23 W/m2
(VIRGO) and ~ -0.18 W/m2 (DIARAD). Finally, taking a conservative value
of 0.3 W/m2 between our different TSI products, induces a variation of
the global mean surface temperature of ~ 0.02 K based on global climate
simulations, which is within the uncertainties of simulated global mean
surface temperatures, hence not impacting significantly any climate
forcing scenarios.