Journal of Advances in Modeling Earth Systems最新文献

Climate change will impact mountain watershed streamflow both directly—with changing precipitation amounts and variability—and indirectly—through temperature shifts altering snowpack, melt, and evapotranspiration. To understand how these complex processes will affect ecosystem functioning and water resources, we need tools to distinguish connections between water sources (rain/snowmelt), groundwater storage, and exit fluxes (streamflow/evapotranspiration), and to determine how these connections change seasonally and as climate shifts. Here, we develop novel watershed-scale approaches to understand water source, storage, and exit flux connections using a dynamic-flux particle tracking model (EcoSLIM) applied in California's Cosumnes Watershed, which connects the Sierra Nevada and Central Valley. This work develops new visualizations and applications to provide mechanistic understanding that underpins the interpretation of isotopic field data at watershed scales to distinguish sources, flow paths, residence times, and storage selection. In our simulations, streamflow comes primarily from snow-derived water while evapotranspiration generally comes from rain. Most streamflow starts above 1,000 m while evapotranspiration is sourced relatively evenly across the watershed and is generally younger than streamflow. Modeled streamflow consists primarily of water sourced from precipitation in the previous 5 years but before the current water year, while ET consists primarily of water from precipitation in the current water year. ET, and to a lesser extent streamflow, are both younger than water in groundwater storage. However, snowmelt-derived streamflow preferentially discharges older water from snow-derived storage. Dynamic-flux particle tracking and new approaches presented here enable novel model-tracer comparisons in large-scale watersheds to better understand watershed behavior in a changing climate.

We present a parsimonious deep learning weather prediction model to forecast seven atmospheric variables with 3-hr time resolution for up to 1-year lead times on a 110-km global mesh using the Hierarchical Equal Area isoLatitude Pixelization (HEALPix). In comparison to state-of-the-art (SOTA) machine learning (ML) weather forecast models, such as Pangu-Weather and GraphCast, our DLWP-HPX model uses coarser resolution and far fewer prognostic variables. Yet, at 1-week lead times, its skill is only about 1 day behind both SOTA ML forecast models and the SOTA numerical weather prediction model from the European Center for Medium-Range Weather Forecasts. We report several improvements in model design, including switching from the cubed sphere to the HEALPix mesh, inverting the channel depth of the U-Net, and introducing gated recurrent units (GRU) on each level of the U-Net hierarchy. The consistent east-west orientation of all cells on the HEALPix mesh facilitates the development of location-invariant convolution kernels that successfully propagate weather patterns across the globe without requiring separate kernels for the polar and equatorial faces of the cube sphere. Without any loss of spectral power after the first 2 days, the model can be unrolled autoregressively for hundreds of steps into the future to generate realistic states of the atmosphere that respect seasonal trends, as showcased in 1-year simulations.

The representation of subgrid-scale orography is a challenge in the physical parameterization of orographic gravity-wave sources in weather forecasting. A significant hurdle is encoding as much physical information with as simple a representation as possible. Other issues include scale awareness, that is, the orographic representation has to change according to the grid cell size and usability on unstructured geodesic grids with non-quadrilateral grid cells. This work introduces a novel spectral analysis method approximating a scale-aware spectrum of subgrid-scale orography on unstructured geodesic grids. The dimension of the physical orographic data is reduced by more than two orders of magnitude in its spectral representation. Simultaneously, the power of the approximated spectrum is close to the physical value. The method is based on well-known least-squares spectral analyses. However, it is robust to the choice of the free parameters, and tuning the algorithm is generally unnecessary. Numerical experiments involving an idealized setup show that this novel spectral analysis performs significantly better than a straightforward least-squares spectral analysis in representing the physical energy of a spectrum. Studies involving real-world topographic data are conducted, and reasonable error scores within ±10% error relative to the maximum physical quantity of interest are achieved across different grid sizes and background wind speeds. The deterministic behavior of the method is investigated along with its principal capabilities and potential biases, and it is shown that the error scores can be iteratively improved if an optimization target is known. Discussions on the method's limitations and broader applicability conclude this work.

Climate models have long-standing difficulties simulating the South Pacific Convergence Zone (SPCZ) and its variability. For example, the default Zhang-McFarlane (ZM) convection scheme in the Community Atmosphere Model version 5 (CAM5) produces too much light precipitation and too little heavy precipitation in the SPCZ, with this bias toward light precipitation even more pronounced in the SPCZ than in the tropics as a whole. Here, we show that implementing a recently developed convection scheme in the CAM5 yields significant improvements in the simulated SPCZ during austral summer and discuss the reasons behind these improvements. In addition to intensifying both mean rainfall and its variability in the SPCZ, the new scheme produces a larger heavy rainfall fraction that is more consistent with observations and state-of-the-art reanalyses. This shift toward heavier, more variable rainfall increases both the magnitude and altitude of diabatic heating associated with convective precipitation, intensifying lower tropospheric convergence and increasing the influence of convection on the upper-level circulation. Increased diabatic production of potential vorticity in the upper troposphere intensifies the distortion effect exerted by convection on transient Rossby waves that pass through the SPCZ. Weaker distortion effects in simulations using the ZM scheme allow waves to propagate continuously through the region rather than dissipating locally, further reducing updrafts and weakening convection in the SPCZ. Our results outline a dynamical framework for evaluating model representations of tropical–extratropical interactions within the SPCZ and clarify why convective parameterizations that produce “top-heavy” profiles of deep convective heating better represent the SPCZ and its variability.

The effects of small-scale topography-induced land surface heterogeneity are not well represented in current Earth System Models (ESMs). In this study, a new topography-based subgrid structure referred to as topographic units (TGU) designed to better capture subgrid topographic effects, and methods to downscale atmospheric forcing to the land TGUs have been implemented in the Energy Exascale Earth System Model (E3SM) Land Model (ELM). Effects of the subgrid scheme and downscaling methods on ELM simulated land surface processes are evaluated over the conterminous United States (CONUS). For this purpose, ELM simulations are performed using two configurations without (NoD ELM) and with (D ELM) downscaling, both using TGUs derived for the 0.5-degree grids and the same land surface parameters. Simulations using the two ELM configurations are compared over the CONUS domain, regional levels, and at observational sites (e.g., SNOTEL). The CONUS-level results suggest that D ELM simulates more snowfall and snow water equivalent (SWE), higher runoff, and less ET during spring and summer. Regional-level results suggest more pronounced impacts of downscaling over regions dominated by higher elevation TGUs and regions with maximum precipitation occurring during cool seasons. Results at the SNOTEL sites suggest that D ELM has superior capability of reproducing the observed SWE at 83% of the sites, with more pronounced performance over topographically heterogeneous TGUs with their maximum precipitation occurring during cool seasons. The results highlight the importance of improving representation of small-scale surface heterogeneity in ESMs and motivate future research to understand their effects on land-atmosphere interactions, streamflow, and water resources management over mountainous regions.

Machine learning (ML) is used to build a bulk microphysical parameterization including ice processes. Simulations of the Lagrangian super-particle model McSnow are used as training data. The ML performs a coarse-graining of the particle-resolved microphysics to multi-category two-moment bulk equations. Besides mass and number, prognostic particle properties (P3) like melt water, rime mass, and rime volume are predicted by the ML-based bulk model. The ML-based scheme is tested with simulations of increasing complexity. As a box model, the ML-based bulk scheme can reproduce the simulations of McSnow quite accurately. In 3d idealized squall line simulations, the ML-based P3-like scheme provides a more realistic extended stratiform region when compared to the standard two-moment bulk scheme in ICON. In a realistic case study, the ML-based scheme runs stably, but can not significantly improve the results. This shows that ML can be used to coarse-grain super-particle simulations to a bulk scheme of arbitrary complexity.

Tropical forest diversity governs forest structures, compositions, and influences the ecosystem response to environmental changes. Better representation of forest diversity in ecosystem demography (ED) models within Earth system models is thus necessary to accurately capture and predict how tropical forests affect Earth system dynamics subject to climate changes. However, achieving forest coexistence in ED models is challenging due to their computational expense and limited understanding of the mechanisms governing forest functional diversity. This study applies the advanced Multi-Objective Population-based Parallel Local Surrogate-assisted search (MOPLS) optimization algorithm to simultaneously calibrate ecosystem fluxes and coexistence of two physiologically distinct tropical forest species in a size- and age-structured ED model with realistic representation of wood harvest. MOPLS exhibits satisfactory model performance, capturing hydrological and biogeochemical dynamics observed in Barro Colorado Island, Panama, and robustly achieving coexistence for the two representative forest species. This demonstrates its effectiveness in calibrating tropical forest coexistence. The optimal solution is applied to investigate the recovery trajectories of forest biomass after various intensities of clear-cut deforestation. We find that a 20% selective logging can take approximately 40 years for aboveground biomass to return to the initial level. This is due to the slow recovery rate of late successional trees, which only increases by 4% over the 40-year period. This study lays the foundation to calibrate coexistence in ED models. MOPLS can be an effective tool to help better represent tropical forest diversity in Earth system models and inform forest management practices.

This study proposes deterministic and stochastic energy-aware hybrid models that should enable simulations of idealized and primitive-equations Geophysical Fluid Dynamics (GFD) models at low resolutions without compromising on quality compared with high-resolution runs. Such hybrid models bridge the data-driven and physics-driven modeling paradigms by combining regional stability and classical GFD models at low resolution that cannot reproduce high-resolution reference flow features (large-scale flows and small-scale vortices) which are, however, resolved. Hybrid models use an energy-aware correction of advection velocity and extra forcing compensating for the drift of the low-resolution model away from the reference phase space. The main advantages of hybrid models are that they allow for physics-driven flow recombination within the reference energy band, reproduce resolved reference flow features, and produce more accurate ensemble forecasts than their classical GFD counterparts. Hybrid models offer appealing benefits and flexibility to the modeling and forecasting communities, as they are computationally cheap and can use both numerically-computed flows and observations from different sources. All these suggest that the hybrid approach has the potential to exploit low-resolution models for long-term weather forecasts and climate projections thus offering a new cost effective way of GFD modeling. The proposed hybrid approach has been tested on a three-layer quasi-geostrophic model for a beta-plane Gulf Stream flow configuration. The results show that the low-resolution hybrid model reproduces the reference flow features that are resolved on the coarse grid and also gives a more accurate ensemble forecast than the physics-driven model.

Data assimilation (DA) plays a pivotal role in diverse applications, ranging from weather forecasting to trajectory planning for autonomous vehicles. A prime example is the widely used ensemble Kalman filter (EnKF), which relies on the Kalman filter's linear update equation to correct each of the ensemble forecast member's state with incoming observations. Recent advancements have witnessed the emergence of deep learning approaches in this domain, primarily within a supervised learning framework. However, the adaptability of such models to untrained scenarios remains a challenge. In this study, we introduce a new DA strategy that utilizes reinforcement learning (RL) to apply state corrections using full or partial observations of the state variables. Our investigation focuses on demonstrating this approach to the chaotic Lorenz 63 and 96 systems, where the agent's objective is to maximize the geometric series with terms that are proportional to the negative root-mean-squared error (RMSE) between the observations and corresponding forecast states. Consequently, the agent develops a correction strategy, enhancing model forecasts based on available observations. Our strategy employs a stochastic action policy, enabling a Monte Carlo-based DA framework that relies on randomly sampling the policy to generate an ensemble of assimilated realizations. Numerical results demonstrate that the developed RL algorithm performs favorably when compared to the EnKF. Additionally, we illustrate the agent's capability to assimilate non-Gaussian observations, addressing one of the limitations of the EnKF.

Large-scale dynamical and thermodynamical processes are common environmental drivers of high-impact weather systems causing extreme weather events. However, such large-scale environmental conditions often display systematic biases in climate simulations, posing challenges to evaluating high-impact weather systems and extreme weather events. In this paper, a machine learning (ML) approach was employed to bias correct the large-scale wind, temperature, and humidity simulated by the atmospheric component of the Energy Exascale Earth System Model (E3SM) at ∼1° resolution. The usefulness of the ML approach for extreme weather analysis was demonstrated with a focus on three high-impact weather systems, including tropical cyclones (TCs), extratropical cyclones (ETCs), and atmospheric rivers (ARs). We show that the ML model can effectively reduce climate bias in large-scale wind, temperature, and humidity while preserving their responses to imposed climate change perturbations. The bias correction is found to directly improve water vapor transport associated with ARs, and representations of thermodynamical flows associated with ETCs. When the bias-corrected large-scale winds are used to drive a synthetic TC track forecast model over the Atlantic basin, the resulting TC track density agrees better with that of the TC track model driven by observed winds. In addition, the ML model insignificantly interferes with the mean climate change signals of large-scale storm environments as well as the occurrence and intensity of three weather systems. This study suggests that the proposed ML approach can be used to improve the downscaling of extreme weather events by providing more realistic large-scale storm environments simulated by low-resolution climate models.