Expected performance of future MAGIC data-assimilated Terrestrial Water Storage (TWS) products

The planned MAGIC mission, a collaboration between ESA and NASA, is expected to deliver 
an extended record of the global mass transport time series with improved accuracy, 
temporal and spatial resolutions. ESA’s involvement in MAGIC is through its Next
Generation Gravity Mission (NGGM) and NASA contributes with its GRACE-C mission. One 
of the key deliverables is terrestrial water storage (TWS), which is vital for assessing 
changes in climate and managing water resources efficiently. As an essential climate 
variable, TWS plays an important role in providing information regarding extreme events. 
While the GRACE and its Follow-On mission distribute global TWS anomalies (TWSA), their 
coarse spatial resolution (around 150,000 km²) constrains detailed analysis of smaller 
basins. Given that freshwater is often sourced from localized aquifer systems, enhancing 
spatial resolution is necessary for effective local water management. Furthermore, 
improvements in spatio-temporal resolution would allow advances in early flood warning 
applications. To overcome these limitations, data assimilation (DA) techniques have been 
developed to combine GRACE observations with land surface models (LSMs), making it 
possible to downscale and disaggregate TWSA information into its individual components.

This study evaluates the performance of data assimilation utilising GRACE-type and MAGIC 
error information within the LSMs NOAH-MP and CLSM, with a focus on two regions in 
South America and Europe. The model runs cover the period from January 2003 to 
December 2006, using data generated during the ESA Science Support study for MAGIC 
Phase A. The data is based on closed-loop simulations with a 30 day repeat orbit, containing 
the hydrology, ice and solid Earth (HIS) signal along with atmosphere-ocean errors, ocean 
tide errors and instrument noise. In total 12 years of monthly data were produced spanning 
from January 1995 to December 2006 with spherical harmonic coefficients up to degree and 
order 90. A reference HIS signal, acquired from the ESA ESM over the same period is used 
to compute retrieval errors. 

The results demonstrate that MAGIC data assimilation offers advantages across climatically 
different regions independent of LSMs chosen. Furthermore, this study indicates that unlike 
previous data assimilation studies, it will be possible to assimilate MAGIC into smaller basins 
sizes, seen by the relative improvements of MAGIC DA over GRACE-type DA. Lastly, it is 
shown that in case of MAGIC DA post-processing can be considerably reduced, such as 
removing the need of filtering up to degree and order 60. Thus, leakage quantification issues 
due to the applied filter would be alleviated achieving more straightforward uncertainty 
quantification.

Overall, MAGIC data assimilation offers substantial improvements in TWS estimation and 
trend correction compared to GRACE-type data assimilation, demonstrating its potential for 
improving hydrological applications