Deep learning model emulators for marine biogeochemistry forecasting from days to decades
📄 arXiv:2606.27168 · 📥 PDF · 2026-06-25 · nlin.CD
Authors: Jozef Skakala [arXiv · scholar] , Ieuan Higgs [arXiv · scholar] , David Moffat [arXiv · scholar]
🕰 Orloj analysis
Tento článek představuje hluboké učení emulátorů pro předpověď mořské biogeochemie, využívající LSTM a fyzikálně informované 1D CNN sítě. Emulátory úspěšně replikují komplexní model v jednorozměrném rámci, zůstávají stabilní po desetiletí a překonávají rodičovský model v přesnosti předpovědí klíčových ekosystémových proměnných.
💡 Jde o vysoce hodnotnou aplikaci hlubokého učení, která slibuje výrazné snížení výpočetních nákladů a zlepšení kvality předpovědí v mořské biogeochemii, s potenciálem pro širší dopad.
✓ falsifiable, modest_claims, limit_reductions, physics_informed_model, outperforms_baseline
⚠ simplified_one_dimensional_framework, performance_in_3d_applications_is_future_hope, largely_stable_implies_not_perfectly_stable, code_availability_not_mentioned
📄 Abstract
Deep-learning emulators have emerged as a promising approach for reducing the computational cost of Earth System Models while potentially improving forecasting skill. Here, we demonstrate the successful emulation of a high-complexity marine biogeochemistry model within a simplified one-dimensional water-column framework. We explore two emulator architectures: Long Short-Term Memory (LSTM) neural networks that emulate a selected subset of variables at daily resolution, and physics-informed one-dimensional Convolutional Neural Networks (1D CNNs) that emulate the full pelagic system throughout the water column also at daily resolution. Using ocean physics simulator inputs, both emulators remain largely stable over multi-decadal timescales and accurately reproduce the parent model in both decadal climate projections and short-range (10-day) forecasting applications. The former includes the ability to predict the timing of phytoplankton Spring blooms several years in advance. When trained on reanalysis data, the emulators substantially outperform the parent model's forecast skill score for several key ecosystem variables, including phytoplankton and zooplankton. If similar performance can be achieved in three-dimensional regional applications, these emulators could provide substantially higher-quality predictions at a fraction of the computational cost. We further apply novel explainability techniques to identify key drivers of emulator behaviour and gain insights into emergent ecosystem dynamics. Performance is evaluated using a range of metrics, including the reproduction of daily variability and extreme events. These approaches have considerable potential for future applications in operational forecasting, climate-scale simulations, and marine autonomous systems.