Application of Sensitivity Kernels for Accurate Surface Mass Change Estimation in the Svalbard Region Using GRACE Data

The estimation of ice mass variations using GRACE data is highly sensitive to evaluation parameters and processing options. Factors that contribute significantly to the variability  of the results include the selection of spherical harmonic solutions, corrections for the geocentre motion, adjustments for inaccuracies in low-degree coefficients, application of auxiliary models to mitigate geophysical signal contamination, and the processing method. Döhne et al. (2023) propose sensitivity kernels as a robust representation of the processing method of surface mass changes, regardless of the general processing approach and the input data set. Sensitivity kernels allow the direct calculation of integrated surface mass changes within a given region and can directly be tailored to effectively account for inherent solution errors and leakage effects. In addition, they provide a consistent and reliable foundation for the evaluation. In this work, we apply this methodology to analyse surface mass changes in the Svalbard region. We test various input parameters, including spherical harmonic solutions or geocentre motion datasets, to determine whether the method produces consistent results regardless of the inputs. We also examine whether using a time-variable GRACE error covariance matrix to account for inherent solution errors yields different results compared to using its approximate representation. Our results highlight the applicability of tailored sensitivity kernels in regional studies and underline their potential to improve the consistency of GRACE-based assessments across different glaciated regions.