The proposal can be found here.

The main aim: The proposal’s current main aim—”to examine the coupled interactions between submesoscale turbulent processes and biogeochemistry in the upper ocean, with a specific focus on tracer patchiness”—is somewhat broad and method-focused. It lacks precise, scientific phrasing that explicitly states what will be established or quantified and does not tightly link to a clearly defined knowledge gap.
Improved aim: “To determine the effects of submesoscale turbulent mixing on the spatial and spectral characteristics of biogeochemical tracer patchiness in the upper ocean and quantify its implications for Earth system model parameterizations.”
Why better: This revised aim uses specific, scientific language (“determine,” “quantify”) that conveys clear, measurable goals rather than describing the process. It focuses on what will be achieved—advancing understanding of tracer patchiness and informing model improvement—thus clarifying the research purpose. Additionally, it explicitly ties to a knowledge gap regarding submesoscale turbulence effects that are critical for accurate Earth system modeling, fulfilling the best practice of logically linking the aim to a known challenge.

Problem statement and knowledge gap: The current introduction presents relevant background information and scientific questions, but it is overly detailed and method-focused before stating the core problem and knowledge gap. It does not concisely highlight the urgent need to understand tracer patchiness driven by unresolved small-scale turbulence in Earth system models, nor does it clearly frame the challenge of computational cost that prevents studying these processes realistically. The narrative would benefit from a more structured flow: from broad importance, to the critical knowledge gap, to the challenge that impedes progress, followed by the innovative approach to address it, then finally the main aim.
Rewritten suggestion: “Reactive biogeochemical tracers such as CO2 and phytoplankton in the upper ocean play a crucial role in the global carbon cycle and climate regulation. These tracers exhibit spatial patchiness influenced by turbulent mixing at submesoscales, yet Earth system models (ESMs) currently fail to resolve these small-scale processes, limiting the accuracy of climate predictions. The main obstacle is the prohibitive computational cost of coupling complex biogeochemical models with high-resolution turbulence simulations. To overcome this, we propose employing novel model reduction techniques adapted from combustion science to enable fully coupled, submesoscale-resolving large eddy simulations (LES) of ocean biogeochemistry. The main aim is to determine the effects of submesoscale turbulent mixing on the spatial and spectral characteristics of biogeochemical tracer patchiness and quantify its implications for improving ESM parameterizations.”
Why better: This version starts broadly and succinctly states the importance of the problem and knowledge gap, directly addresses the challenge hindering progress, introduces the innovative solution clearly, and leads logically into the precise, scientific main aim. It aligns well with best practices for compelling proposal introductions, improving clarity and persuasiveness while ensuring the reader immediately understands the significance and novelty of the proposed work.

Specific objectives: The current proposal clearly states three specific objectives aligned with the main aim, but some objectives still lean towards describing methods rather than outcomes. Also, the objectives could be phrased more concisely with a sharper focus on what will be determined or quantified scientifically.
Improved formulation:
1. To establish a computationally efficient framework for fully-coupled LES of upper-ocean turbulence and biogeochemistry by developing and implementing reduced biogeochemical models accelerated on GPUs.
2. To quantify the impact of submesoscale turbulent processes, including wave-driven Langmuir turbulence and vertical convection, on the evolution, spatial distribution, and spectral characteristics of biogeochemical tracer patchiness under idealized ocean conditions.
3. To determine the contributions of submesoscale turbulence to biogeochemical tracer patchiness in realistic ocean scenarios through LES simulations validated by observational data from the Drake Passage.
Why better: These revised objectives emphasize precise, outcome-oriented goals using active verbs like “establish,” “quantify,” and “determine,” which clearly specify what will be achieved rather than how. Each objective aligns directly with the main aim, covers necessary steps in logical order from model development to idealized studies to real-world validation, and avoids redundancy by clearly differentiating process study from computational implementation and real-world application. This approach improves clarity, scientific focus, and feasibility representation, aligning with best practices for strong proposals.

Novelty and significance: The proposal clearly highlights novelty in applying combustion-inspired chemical kinetic model reduction and GPU-accelerated solvers for the first time to ocean biogeochemical models within high-fidelity, submesoscale-resolving LES. It emphasizes the unique integration of reduced BGC models with two-way coupled turbulent flow simulations, enabling studies of tracer patchiness at spatial and spectral scales previously inaccessible. However, the narrative could better integrate this novelty more consistently into the research logic, explicitly contrasting prior limitations and the new transformative impact throughout, using strong language like “For the first time” and “Unlike previous studies.”
Concrete improvement suggestion: “Unlike prior studies that have been limited by computational costs to oversimplified or uncoupled approaches, this project will, for the first time, leverage chemical kinetics model reduction techniques adapted from combustion science to enable fully coupled, GPU-accelerated LES of upper-ocean turbulence and complex biogeochemical tracers at submesoscales. This novel integration overcomes previous computational barriers, allowing the first detailed quantification of how submesoscale turbulent mixing drives tracer patchiness in realistic ocean scenarios. The outcomes will significantly advance our fundamental understanding of upper-ocean biogeochemical dynamics and provide unprecedented data to refine Earth system model parameterizations.”
Why better: This phrasing better stresses the novelty repeatedly and embeds it tightly within the research rationale. It clearly contrasts with prior work, explicitly states why the novelty matters, and highlights the expected step-change in capability and scientific insight, all of which strengthen the proposal’s persuasive impact and intellectual merit argument.

Feasibility: The proposal convincingly argues feasibility by detailing strong preliminary work using NCAR LES and GPU-accelerated solvers, showing model reduction techniques applied successfully in combustion contexts, and demonstrating prior simulation results of tracer evolution at submesoscales. It also specifies access to suitable petascale supercomputers like Cheyenne and Summit with GPU capabilities, supported by existing allocations, affirming resource availability.
However, the proposal could further strengthen feasibility by explicitly discussing potential risks, such as uncertainties in model reduction accuracy or computational costs possibly exceeding estimates, and outlining concrete mitigation strategies (e.g., fallback approaches or adaptive model complexity). Including a concise paragraph on risk assessment and management would directly address common reviewer concerns and enhance confidence in project success.
Adding such a section might read: “Potential risks include the reduced BGC models not capturing all relevant biogeochemical dynamics and computational costs surpassing initial estimates. To mitigate these risks, we will perform rigorous validation of reduced models against full-model benchmarks and observational data (including Drake Passage time series). Should computational costs exceed projections, we will prioritize simulations of key tracers and scenarios using scalable GPU resources and adaptive model fidelity. Prior experience with NCAR LES and GPU acceleration underpins our confidence in overcoming these challenges.”
This addition would provide a realistic assessment demonstrating preparedness for challenges and conform to best practices in feasibility presentation, helping reviewers trust the project’s successful execution.

Impact and significance: The proposal effectively articulates the transformative impact of enabling high-resolution, fully coupled LES of submesoscale turbulent and biogeochemical processes, which has not been possible before due to computational constraints. It clearly links anticipated outcomes—such as better understanding of tracer patchiness and improved Earth system model parameterizations—to significant advances in climate modeling accuracy. The proposal also compellingly describes broad impacts on diverse scientific communities, workforce development through training underrepresented STEM students, outreach to K–12 and the general public, and societal benefits from more accurate climate impact predictions that could influence policy and mitigation.
To further enhance impact clarity and persuasiveness, the proposal could include a more explicit statement early on in the introduction or impact section about how these advances will enable new capabilities that are currently impossible, such as simulating coupled biogeochemical-physical processes at operationally relevant scales for climate prediction models. Explicitly contrasting the current state-of-the-art with the proposed novel capabilities would sharpen the presentation of significance.
Suggested addition: “For the first time, this project will enable the simulation of fully coupled, reactive biogeochemical tracers with submesoscale physical turbulence at spatial and temporal scales relevant to Earth system models, overcoming long-standing computational barriers. This advance will provide unprecedented insight into the mechanisms driving tracer patchiness and allow development of accurate subgrid-scale parameterizations, directly improving the fidelity of climate predictions—critical for informing policy decisions and societal response to climate change.”
This concrete framing of what becomes newly possible and why it matters would strengthen the perceived significance by emphasizing the research’s enabling and transformative nature, thus matching best practice to connect impact tightly to objectives and problem motivation.

Narrative: Good research proposals follow a structured narrative that logically builds a compelling case for the research. They guide the reader through a progression of ideas that naturally lead from background knowledge to research objectives, novelty, feasibility, and impact. Here comes an attempt to capture the current proposal in the framework that most successful proposals have in common.

Reactive biogeochemical tracers such as CO2 and phytoplankton in the upper ocean play a crucial role in the global carbon cycle and climate regulation. These tracers exhibit spatial patchiness influenced by turbulent mixing at submesoscales, yet Earth system models (ESMs) currently fail to resolve these small-scale processes, limiting the accuracy of climate predictions. The main obstacle is the prohibitive computational cost of coupling complex biogeochemical models with high-resolution turbulence simulations. To overcome this, we propose employing novel model reduction techniques adapted from combustion science to enable fully coupled, submesoscale-resolving large eddy simulations (LES) of ocean biogeochemistry. 
The main aim is to determine the effects of submesoscale turbulent mixing on the spatial and spectral characteristics of biogeochemical tracer patchiness in the upper ocean and quantify its implications for improving Earth system model parameterizations.
To achieve this aim, we will pursue the following specific objectives:
1. To establish a computationally efficient framework for fully-coupled LES of upper-ocean turbulence and biogeochemistry by developing and implementing reduced biogeochemical models accelerated on GPUs.
2. To quantify the impact of submesoscale turbulent processes, including wave-driven Langmuir turbulence and vertical convection, on the evolution, spatial distribution, and spectral characteristics of biogeochemical tracer patchiness under idealized ocean conditions.
3. To determine the contributions of submesoscale turbulence to biogeochemical tracer patchiness in realistic ocean scenarios through LES simulations validated by observational data from the Drake Passage.
Unlike prior studies that have been limited by computational costs to oversimplified or uncoupled approaches, this project will, for the first time, leverage chemical kinetics model reduction techniques adapted from combustion science to enable fully coupled, GPU-accelerated LES of upper-ocean turbulence and complex biogeochemical tracers at submesoscales. This novel integration overcomes previous computational barriers, allowing the first detailed quantification of how submesoscale turbulent mixing drives tracer patchiness in realistic ocean scenarios. The outcomes will significantly advance our fundamental understanding of upper-ocean biogeochemical dynamics and provide unprecedented data to refine Earth system model parameterizations.
We build on robust preliminary work including simulations with NCAR LES incorporating GPU-accelerated solvers and successful application of model reduction techniques in combustion chemistry. Access to petascale supercomputers such as Cheyenne and Summit, with existing allocations and support for GPU-accelerated code, ensures availability of state-of-the-art computational resources. We acknowledge potential risks including uncertainties in reduced model fidelity and computational cost overruns. To mitigate these, we will rigorously validate reduced models against full-complexity benchmarks and observational datasets and adapt computational strategies to prioritize critical simulations with scalable GPU resources.
For the first time, this project will enable the simulation of fully coupled, reactive biogeochemical tracers with submesoscale physical turbulence at spatial and temporal scales relevant to Earth system models, overcoming long-standing computational barriers. This advance will provide unprecedented insight into the mechanisms driving tracer patchiness and allow development of accurate subgrid-scale parameterizations, directly improving the fidelity of climate predictions—critical for informing policy decisions and societal response to climate change.
Moreover, the project will foster workforce development by training underrepresented STEM students and engage diverse communities through outreach activities, amplifying the broader societal impact of advancing climate science. Overall, by bridging fundamental ocean turbulence dynamics with biogeochemical complexity using cutting-edge computational methods, this research promises to deliver transformative improvements in Earth system modeling and enhance our ability to anticipate and respond to global climate challenges.

Structure in the detailed project description: The detailed work plan would benefit greatly from restructuring its methods section around the three specific objectives, each forming a main section with clear subsection headings corresponding to key subtasks and methods. This would improve clarity by explicitly linking methods and simulations to specific objectives and providing a logical progression from model reduction (Objective 1), through process studies in idealized physical scenarios (Objective 2), to realistic simulations and validation (Objective 3).
Each objective section should start with a short introduction that summarizes the objective’s goal and ends with an explicit statement of expected outcomes. Subheadings within each section should organize tasks clearly, for example: model reduction approach, solver implementation, unit problem testing (Objective 1); LES setup, turbulence-biogeochemistry coupling mechanisms, tracer diagnostics (Objective 2); real-world scenario setup, observational data integration, spectral and statistical analyses (Objective 3). This structure enhances readability and shows a coherent research plan.
Moreover, a final integrative section on data analysis and interpretation, explicitly tied to all objectives, would help clarify how simulation results will be combined to answer the core scientific questions. Lastly, a brief discussion of risk management and contingency plans should be included, ideally following the respective objective(s) or in a dedicated subsection to reinforce feasibility.
This restructuring would mirror best practices seen in highly rated proposals, helping reviewers easily trace how each objective is addressed methodologically and how outcomes build cumulatively towards the main aim.

Methodological detail: The proposal generally provides a strong level of methodological detail and justification, but some areas could be clarified and focused further to align better with highly rated proposals.
For example, while the plan to use model reduction from combustion science for biogeochemical (BGC) models and GPU acceleration is explained, the proposal could more explicitly justify why the selected reduction methods (e.g., Directed Relation Graph with Error Propagation) and time integration schemes (fourth-order Runge-Kutta-Chebyshev) are particularly suited for this problem, highlighting advantages such as handling stiffness and computational efficiency. Adding brief rationale for each choice helps reviewers appreciate appropriateness.
In the process study (Objective 2), more detail on how coupling between turbulence and BGC tracers will be quantified is needed—e.g., describing the calculation of spectral diagnostics, probability density functions, and variance measures including how these are interpreted relative to tracer patchiness hypotheses. Connecting methods to hypotheses explicitly would improve clarity.
Similarly, for the real-world simulations (Objective 3), while the use of Drake Passage observations for forcing and validation is good, the proposal could specify which observational datasets and parameters will be used and detail the spectral or statistical methods that will quantify agreement between simulations and observations.
Concrete suggestions:
– Add short justifications for the selected chemical kinetics reduction and solver choices, e.g., “DRGEP is chosen for its efficiency and proven accuracy in reducing large, stiff chemical kinetic models, making it suitable for complex BGC systems with many species.”
– Explicitly connect analysis methods to expected outcomes, e.g., “Spectral diagnostics such as wavenumber scaling exponents will quantify the spatial heterogeneity of tracers, linking directly to hypotheses regarding submesoscale turbulence impact on tracer patchiness.”
– Specify observational products used for validation (e.g., pCO2 measurements from the Drake Passage Time-series Program) and explain how simulation outputs will be compared quantitatively, improving confidence in real-world relevance.
These changes would enhance comprehension of why and how methods are chosen and employed, providing a clearer line of evidence supporting the project’s aim and objectives, matching the detail and justification level expected in high-quality proposals.

Timeline and planning: While the proposal includes a detailed enumeration of simulations and associated resource needs, it lacks an explicit, visual timeline (e.g., Gantt chart) clearly sequencing the specific objectives and milestones over the three-year duration. The current presentation disperses scheduling information across text and tables, making it difficult to quickly assess the logical order and time allocation per objective.
To improve, provide a concise, dedicated timeline figure or table that maps each objective and its key subtasks across project years and quarters, highlighting dependencies and overlapping activities. For example:
– Year 1: Focus on Objective 1 model reduction and solver development, including unit problem testing.
– Year 2: Conduct Objective 2 idealized LES process studies with reduced models, analyzing different turbulence regimes.
– Year 3: Perform Objective 3 realistic Drake Passage simulations and observational validation; integrate data analysis and finalize publications.
This sequencing matches logical dependencies (model development before large-scale simulations) and allows reviewers to verify that milestones are spread realistically. Explicit assignment of responsibilities to team members for each objective and milestone in the timeline enhances clarity and accountability.
Such a visual, logically organized timeline will make the work plan more comprehensible and demonstrate careful resource and task planning reflective of best practices in successful proposals.

Other: The proposal includes a detailed and well-structured description of prior NSF support, showcasing relevant expertise and accomplishments that support the feasibility and competence of the research team. It has a strong broader impacts section engaging diverse educational levels from K–12 to graduate students and the general public, as well as societal benefits via improved climate modeling relevant for policy. Although extensive and well integrated, the proposal would benefit from more explicit presentation of specific hypotheses and their linkage to objectives and methods to sharpen focus on the scientific questions being addressed. Additionally, consolidating some methodological details around key evaluation techniques and validation data sets would enhance clarity without overwhelming the reader.
Overall, the remaining sections complement the main research narrative well, demonstrating solid intellectual merit and thoughtful broader impact plans, fulfilling best practices summarized in the guidance documents. No major unaddressed deficiencies remain beyond those discussed in earlier steps.
Evaluation complete. Recommendations given in prior steps cover the main areas for improvement for a competitive, highly rated proposal.