Arctic Mission Benefit Analysis: impact of sea ice thickness, freeboard, and snow depth products on sea ice forecast performance

Abstract

Assimilation of remote-sensing products of sea ice thickness (SIT) into sea ice-ocean models has been shown to improve the quality of sea ice forecasts. Key open questions are whether assimilation of lower-level data products such as radar freeboard (RFB) can further improve model performance and what performance gains can be achieved through joint assimilation of these data products in combination with a snow depth product. The Arctic Mission Benefit Analysis system was developed to address this type of question. Using the quantitative network design (QND) approach, the system can evaluate, in a mathematically rigorous fashion, the observational constraints imposed by individual and groups of data products. We demonstrate the approach by presenting assessments of the observation impact (added value) of different Earth observation (EO) products in terms of the uncertainty reduction in a 4-week forecast of sea ice volume (SIV) and snow volume (SNV) for three regions along the Northern Sea Route in May 2015 using a coupled model of the sea ice-ocean system, specifically the Max Planck Institute Ocean Model. We assess seven satellite products: three real products and four hypothetical products. The real products are monthly SIT, sea ice freeboard (SIFB), and RFB, all derived from CryoSat-2 by the Alfred Wegener Institute. These are complemented by two hypothetical monthly laser free-board (LFB) products with low and high accuracy, as well as two hypothetical monthly snow depth products with low and high accuracy.On the basis of the per-pixel uncertainty ranges provided with the CryoSat-2 SIT, SIFB, and RFB products, the SIT and RFB achieve a much better performance for SIV than the SIFB product. For SNV, the performance of SIT is only low, the performance of SIFB is higher and the performance of RFB is yet higher. A hypothetical LFB product with low accuracy (20 cm uncertainty) falls between SIFB and RFB in performance for both SIV and SNV. A reduction in the uncertainty of the LFB product to 2 cm yields a significant increase in performance.Combining either of the SIT or freeboard products with a hypothetical snow depth product achieves a significant performance increase. The uncertainty in the snow product matters: a higher-accuracy product achieves an extra performance gain. Providing spatial and temporal uncertainty correlations with the EO products would be beneficial not only for QND assessments, but also for assimilation of the products.

From Observation to Information and Users: The Copernicus Marine Service Perspective

“…Improvements in sea-ice concentrations data have also been assessed by OSSEs (Posey et al, 2015). Weighting the value from different observation types has been done by adjoint modeling (Kaminski et al, 2018). The assimilation of sea-ice drift has been less successful, so far, partly because of the short model memory of sea-ice drift and partly because of sea-ice models deficiencies (Stark et al, 2008;Sakov et al, 2012).…”

A Framework for the Development, Design and Implementation of a Sustained Arctic Ocean Observing System

“…In contrast to both OSEs and OSSEs, where statistical approach is an important component of the DA step, adjointbased methods utilize the dynamical information in the tangent linear and adjoint models of the underlying general circulation model (GCM). Through the equations which capture conservation and constitutive laws, propagation of information up-and down-stream of any quantity of interest (QoI) is used to (a) assess impactful regions where new observations can be potentially deployed (Marotzke et al, 1999;Zanna et al, 2010;Heimbach et al, 2011;Nguyen et al, 2017;Stammer et al, 2018); (b) assess the redundancy of existing observing networks (Köhl and Stammer, 2004;Moore et al, 2017b); (c) quantify the impacts of selected existing/new observational networks on reducing posterior uncertainties of the GCM control parameters and/or potential unobserved remote QoI (Moore et al, 2011(Moore et al, , 2017aBui-Thanh et al, 2012;Heimbach, 2014, 2018;Kaminski et al, 2015Kaminski et al, , 2018; (d) find an optimal observing network through Hessian-based OED that minimizes the posterior uncertainties as a function of the control parameters and/or targeted QoI (Alexanderian et al, 2016;Loose, 2019). The advantage of the adjoint-based methods is not only the quantification of uncertainty reduction of the GCM control parameters and/or any specific QoI to the observing network but also the identification of dynamical connection and causal relationship between them.…”

“…To date, OSE, OSSE, and adjoint-based OED have not been fully utilized for optimization of observational work due to the disadvantages associated with each of the methods as discussed above. As a result, there are only a few examples of OSEs employed to provide quantitative, traceable guidance on key aspects of observing system design (e.g., Panteleev et al, 2009;Jung et al, 2016;Kaminski et al, 2018;Stammer et al, 2018). An argument put forward against over-reliance on OSEs and OSSEs for system design is that measurements in a rapidly changing Arctic are typically meant to provide information that can help anticipate or prepare for major changes and transformations.…”

“…Advances in forward modeling and 4D-var DA have facilitated model guidance that is more tightly linked to, e.g., sensor and observing technology constraints such that OS(S)Es and OEDs can potentially help support the design of the sensor systems themselves (Kaminski et al, 2018). It should be expected that, due to the potentially limited and serious deficiencies, an optimal network based on these quantitative tools might not capture the dynamics when compared to field observations and other political/economic constraints.…”

Leveraging Uncertainty Quantification to Design Ocean Climate Observing Systems

“…Hessian UQ has been routinely applied in numerical weather prediction (NWP; Leutbecher, 2003) and, more broadly, in computational science and engineering (CSE; Bui-Thanh et al, 2012), but it has only seen limited use in the oceanographic community. Previous studies have applied Hessian UQ after severely reducing the dimension of the space of uncertain parameters in an ad-hoc manner (Kaminski et al, 2015;Kaminski et al, 2018), or in the dual form of "representers" (Bennett, 1985;Moore et al, 2017;Zhang et al, 2010). These examples have focused on regional ocean settings and on daily to monthly time scales.…”