Journal of Advances in Modeling Earth Systems最新文献

A critical challenge for machine-learning (ML) parameterization in global climate models (GCMs) is to achieve stable, accurate simulations under climates not seen during training. Previous studies have demonstrated promising offline performance and year-long online stability in aquaplanet simulations but have encountered difficulties in real geography and under climate warming. Here we report that a GCM with real geography configuration using neural-network-based cloud and convection parameterization, trained exclusively with present-day climate data, successfully performs a stable, decade-long simulation of a warm climate with +4 K sea surface temperature (SST). The neural network (NN) is based on Han et al. (2023, https://doi.org/10.1029/2022ms003508) with additional inputs. The simulation captures the global precipitation distribution, surface temperatures, vertical atmospheric structures, and extreme precipitation very well, closely matching simulations from both the superparameterized CAM (SPCAM) and the conventional CAM5 in the warm climate without accuracy degradation compared to those in the baseline climate. Moreover, it produces a climate response to +4 K SST in atmospheric thermodynamic states and circulations similar to those from SPCAM and CAM5. Prognostic ablation tests on NN input variables show that the NN without convective memory as input suffers from numerical instability, and the NN without considering radiative variables and land fraction as input, or with reduced training samples produce less accurate results. To our knowledge, this is the first time an ML parameterization successfully achieves online extrapolation to a warm climate without using additional warm-climate data for training. It demonstrates the potential of ML-driven parameterizations for credible long-term climate projections.

Processes at the air-sea interface govern the climate mean state and variability by determining the exchange of momentum, heat, and water between the atmosphere and ocean. Traditional climate models compute those exchanges across the air-sea interface by assuming an ocean surface with roughness determined by atmospheric wind and stability conditions, essentially assuming ocean surface waves are in equilibrium states. In reality, that is rarely the case. Such effects have been emphasized in numerical weather predictions for weather systems like tropical cyclones. An accurate representation of ocean surface waves requires a prognostic ocean surface wave model. The addition of WAVEWATCH III to the Community Earth System Model version 2 (CESM2) makes it possible to parameterize the impacts of ocean surface waves on the momentum and energy exchange. This study documents the implementation of a sea-state-dependent surface flux scheme in CESM2. It considers the effects of waves on ocean surface roughness and those of sea spray on sensible and latent heat. It is found that the new scheme significantly impacts mean atmospheric circulation and the upper ocean. The errors in mean atmospheric circulation and surface temperature patterns are reduced. The modified surface flux lowers the eddy-driven jet speed and weakens the Hadley circulation. Global sea surface temperature (SST) warm bias is reduced due to the cooling of the Southern Ocean and eastern boundary currents. In particular, some parts of eastern and central Pacific exhibit a weak cooling trend in the simulation for recent decades, reducing the existing SST trend bias in CESM2.

The China Meteorological Administration's mesoscale operational Global/Regional Assimilation and PrEdiction System (GRAPES_Meso5.1), coupled with the Chinese Unified Atmospheric Chemistry Environment (CUACE) model, has been enhanced to incorporate aerosol-cloud-radiation interactions, creating the CMA's first version of the chemistry-weather integrated model, GRAPES_Meso5.1/CUACE CW V1.0. We applied this model to examine the impacts of aerosols on weather forecasts during the 2016–2017 winter season across China. Results indicate that incorporating aerosol feedbacks improves forecasts of temperature and precipitation, particularly in several regions. The most pronounced improvement in 72-hr temperature forecasts occurred over the Beijing-Tianjin-Hebei (JJJ) region, where forecast errors were reduced by up to 30%. High cloud cover increased in the JJJ and Pearl River Delta regions. On a broader scale, the model reduced cumulative precipitation forecast errors by an average of 7.9–8.7 mm across China. While some of these improvements may fall within the range of internal atmospheric variability, the results suggest that aerosols can meaningfully influence convective and radiative processes relevant to short-term weather prediction. These findings underscore the potential benefits of integrating prognostic aerosol processes into numerical weather prediction systems, particularly for improving forecast accuracy under high-pollution conditions.

Atmospheric Kelvin waves (KWs) are influenced by both tropical convective heating and dry dynamical processes, but the relative contributions of these mechanisms remain uncertain. This study examines the effects of various processes on stationary and transient KWs in ERA5 data using a global, multivariate framework that identifies KWs based on their spatial structure. The nonlinear dynamical momentum and temperature tendencies, that is, those arising from advection, are quantified together with tendencies due to physical parametrizations derived from short-term ERA5 forecasts. A large part of the KW signal is stationary; this wave pattern is associated with non-radiative diabatic heating and warm vertical advection over the Maritime Continent. These processes act as KW energy sinks because the temperature tendencies are phase-shifted relative to the stationary KW pattern. For both stationary and transient KWs, the largest energy source is the meridional advection of zonal momentum, which induces easterly momentum tendencies in the upper troposphere, thereby strengthening the KW easterlies in the eastern hemisphere. The main KW energy sink originates from momentum dissipation. The day-to-day variability of KW energy tendencies is dominated by non-radiative diabatic heating. The developed global numerical framework paves the way for future investigations of wave interactions in the tropics and extratropics.

A satellite-based temperature-dependent parameterization of the Wegener-Bergeron-Findeisen (WBF) process that takes into account the subgrid-scale variability of cloud thermodynamic phase within mixed-phase clouds is developed and implemented in version 5.3 of the Community Atmosphere Model (CAM5.3). Its impact on cloud microphysical and macrophysical properties in experiments with prescribed sea surface temperature and sea ice concentrations as well as the cloud feedback response to a global warming perturbation is investigated. The parameterization significantly improves overestimates in the mass of ice within mixed-phase clouds and ice effective radius relative to satellite observations, the former being superior to tuning the WBF process with a multiplicative constant. The parameterization also reduces overall biases in cloud fraction with respect to satellite observations, however, is due to compensating biases in existing simulated low biases in low-level cloud cover and new increased biases in non-low-level cloud cover. The increased bias in non-low-level cloud cover is due to decreases in the rate of autoconversion of cloud ice that is a side effect of the WBF parameterization. While the WBF parameterization can significantly impact the magnitude of model biases in cloud properties and the cloud feedback, it does not significantly change their spatial distribution. Before observational constraints on WBF process rates become available, it is recommended that temperature-dependent scalings of the WBF process are used to account for the subgrid-scale variability of cloud phase rather than a constant scaling parameter as the former type of parameterization can more realistically simulate cloud properties relative to satellite observations.

This paper introduces the Global Multilayer Canopy OPTimization (GMC-OPT) model, designed to scale sub-daily leaf-level carbon fluxes to the canopy level. The model integrates three core components: canopy radiative transfer, optimization-based physiology, and energy balance. This study highlights the model's simulation of gross primary productivity (GPP), with a novel focus on vertically resolved radiation fields, photosynthetic capacity, and associated physiological processes. Specifically, the model accounts for: (a) sub-daily stomatal conductance optimization; (b) sub-daily timing of photosynthetic capacity optimization, and (c) vertical positioning of seasonal leaf turnover. After benchmarking against flux tower GPP data, GMC-OPT achieves high agreement with the tower GPP at annual, monthly, and hourly scales. Model calibration suggests a lower-than-expected efficiency in converting absorbed photosynthetically active radiation (APAR), primarily because not all APAR reaches chlorophyll. In addition, the model predicts a decrease in photosynthetic capacity from the top to the bottom of the canopy, with vertical acclimation becoming less responsive under high light intensities at the upper canopy. The model further reveals distinct plant functional type strategies on seasonal acclimation due to vertical leaf turnover. Tree-dominated biomes such as needleleaf and mixed forests tend to prioritize light harvesting, while non-tree biomes do not. Deciduous broadleaf forests maintain a relatively constant leaf and canopy photosynthetic capacity through leaf turnover, whereas biomes such as evergreen broadleaf forests and woody savannas often reallocate nutrients within existing leaves through acclimation. GMC-OPT promises a powerful diagnostic tool for exploring interactions among carbon, water, nutrients, and energy in a changing environment.

Satellite observations of sea surface temperature (SST) and ocean chlorophyll are critical for validating Earth system models (ESMs). However, missing satellite data due to cloud cover, sea ice, and low solar angle can introduce sampling bias that distorts model–observation comparisons. Here, we quantify satellite sampling bias in Moderate Resolution Imaging Spectroradiometer (MODIS) SST and chlorophyll and demonstrate how accounting for this bias changes our estimates of model performance. We apply realistic MODIS sampling to modeled SST and chlorophyll from an ocean-only hindcast simulation (2003–2016) of the Community Earth System Model. These model outputs are compared to real-world MODIS observations to examine how selective sampling affects the magnitude and spatial patterns of the apparent model bias. We find that model bias generally exceeds sampling bias, though the relative importance of the two depends on the spatial and temporal scale. Sampling bias is most pronounced at high-latitudes and in persistently cloudy regions, where it can impact annual means and apparent long-term trends. Accounting for sampling bias reduces model–observation differences in the multi-year means: the root mean square error decreases from 0.976 to 0.792°C for SST and from 0.635 to 0.624 (log-transformed units) for chlorophyll. However, in some regions, correcting for satellite sampling bias increases model bias. These results demonstrate that while sampling bias is generally a small uncertainty compared to model bias, it can meaningfully influence model evaluation and should be considered in assessments of ESM performance for SST and chlorophyll.

The tropical atmosphere plays an important role in transporting energy poleward and driving the global circulation. However, understanding and simulating this fundamental aspect of our climate remains difficult due to its sensitivity to convective parameterizations and horizontal resolution. This study focuses on benchmarking the resolution dependence of tropical poleward energy transport in two aquaplanet atmospheric general circulation models with disabled convective parameterizations: a nonhydrostatic high-resolution (100–6 km) finite-volume cubed-sphere model with a full physics package and a lower-resolution (300–100 km) hydrostatic spectral model with idealized moist physics. Despite differences in their physics and numerics, both models demonstrate that column-integrated poleward moist static energy transport by the mean meridional circulation increases with resolution in the deep tropics, while transport by transient eddies decreases. These changes are associated with enhanced gross moist stability that switches from negative to positive due to an increasingly top-heavy mean circulation and reduced eddy activity diffusing water vapor along an unchanging mean moisture gradient. Further analysis rules out extratropical baroclinic eddies and radiation as the main drivers of these changes. Instead, the resolution dependence of both the mean meridional circulation and transient eddies appears to reflect the resolution dependence of tropical explicit (unparameterized) deep convection. We speculate the multiscale interactions of convection allow for a coupling between gross moist stability and eddy moisture flux, leading to their concurrent changes with resolution. We discuss the implications of this resolution dependence for developing theories and models of the tropical atmosphere.

Clouds play a crucial role in regulating the hydrologic cycle and Earth's radiative energy budget, yet they are often poorly represented in global climate models (GCMs). This study applies deep machine learning (DML) to develop a physical parameterization of volumetric cloud fraction (VCF), the fraction of a 3-D grid volume occupied by clouds using satellite lidar-radar measurements. The DML learns the complex relationships between observed VCF profiles and collocated meteorological variables from MERRA-2 reanalysis data. Our results show that the neural network (NN), particularly a sequence-to-sequence long short-term memory (LSTM) network with a customized loss function, effectively captures underlying cloud physical processes. The DML prediction outperforms MERRA-2 reanalysis in representing low-level clouds in tropical and subtropical regions and low- and middle-level clouds over midlatitude storm-track and improves VCF histograms. These improvements are reflected in vertical distributions of zonally, meridionally, and globally averaged VCFs, geographic distributions of low-, middle-, and high-level clouds, and seasonal variations in monthly mean VCF. Furthermore, the DML predictions effectively capture El Niño-Southern Oscillation (ENSO) and other interannual variations. The NN parameterization is further evaluated through sensitivity analysis, where a single predictor is perturbed at a time. This reveals that relative humidity (RH) is the dominant factor influencing variations in globally averaged VCF at low and middle altitudes, followed by temperature. At higher altitudes, temperature becomes the primary driver of VCF through its effect on RH. Increases in pressure vertical velocity (ω) are associated with decreases in VCF, though their effect is minor compared to RH and temperature.