Multi-scale Two-step Assimilation for SWOT Ocean Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current ocean state forecasting systems, such as the Navy Coastal Ocean Model (NCOM), struggle to effectively assimilate the high-density Surface Water Ocean Topography (SWOT) observations, particularly at smaller scales, leading to underutilization of valuable data and reduced forecast accuracy.
Innovation Solution
A multi-scale two-step data assimilation method is implemented, where a large-scale correction is made based on a long observation time window using a large-scale decorrelation length scale, followed by a small-scale correction using a short observation time window and small-scale decorrelation length scale, to generate a more accurate ocean state forecast.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single-scale assimilation system is used, then large-scale ocean features are accurately captured, but small-scale features are underutilized and forecast accuracy is reduced
Solution Approach 1:
The assimilation system is segmented into two distinct scales: large-scale assimilation first corrects mesoscale features using a larger decorrelation length scale, then small-scale assimilation corrects submesoscale features using a smaller decorrelation length scale. This segmentation allows each scale to be optimized independently, resolving the contradiction between capturing small-scale details and maintaining large-scale accuracy.
Solution Approach 2:
Large-scale assimilation is performed as a preliminary step before small-scale assimilation. The large-scale correction is applied to the background state first, creating an improved initial condition that preserves large-scale accuracy. Then small-scale assimilation builds upon this corrected background, ensuring small-scale features are captured without compromising large-scale fidelity.
2Quantity of substance
If high-density SWOT observations are assimilated using traditional methods, then data coverage is increased, but data underutilization occurs at smaller scales
Solution Approach 1:
The system applies different assimilation parameters locally to different spatial scales. Large-scale regions use a larger decorrelation length scale (e.g., 100 km) to capture mesoscale eddies, while small-scale regions use a smaller decorrelation length scale (e.g., 20 km) to capture submesoscale features. This local adaptation allows high-density SWOT observations to be fully utilized across all scales.
Solution Approach 2:
The assimilation system dynamically adjusts the decorrelation length scale parameter based on the spatial scale being assimilated. The system transitions from a static single-scale approach to a dynamic multi-scale approach where the decorrelation length scale is optimized for each assimilation step, enabling versatile utilization of high-density observations across varying spatial scales.
3Measurement precision
If a multi-scale two-step assimilation is implemented, then small-scale features are captured, but system complexity increases
Solution Approach 1:
The complex multi-scale assimilation problem is segmented into two manageable sequential steps: large-scale assimilation followed by small-scale assimilation. Each step uses a single decorrelation length scale parameter, simplifying the computational complexity while still achieving multi-scale resolution. The segmentation breaks down the complex problem into simpler sub-problems that can be solved sequentially.
4Reliability
If large-scale correction is applied first, then large-scale accuracy is maintained, but small-scale corrections may be compromised
Solution Approach 1:
Large-scale assimilation is performed as a preliminary action to establish an accurate large-scale background state. This preliminary correction ensures large-scale accuracy is maintained. Then small-scale assimilation is performed as a subsequent action that builds upon the corrected background, allowing small-scale corrections to be applied without compromising large-scale fidelity. The sequential ordering resolves the contradiction by ensuring each scale is corrected in the optimal sequence.
Data Source
AI summary
A method of forecasting an ocean state via a multi-scale two-step assimilation of Surface Water Ocean Topography (SWOT) observations. The method may include receiving data associated with a prior ocean state forecast associated with SWOT observations, determining a large-scale increment state variable based on a large scale correction associated with the prior ocean state forecast, and determining a small scale initial input value based on (i) a combination of the background state associated with the prior ocean state forecast and (ii) the determined large-scale increment state variable. The method may include generating, based on the determined small scale initial input value, a small scale correction associated with the prior ocean state forecast, determining a small-scale increment state variable based on the small scale correction, and generating a current ocean state forecast based on at least some of this information.