Lu, Feiyu, April 2025: U-Net Kalman Filter (UNetKF): An example of machine learning-assisted data assimilation. Journal of Advances in Modeling Earth Systems, 17(4), DOI:10.1029/2023MS003979. Abstract
Machine learning techniques have seen a tremendous rise in popularity in weather and climate sciences. Data assimilation (DA), which combines observations and numerical models, has great potential to incorporate machine learning and artificial intelligence (ML/AI) techniques. In this paper, we use U-Net, a type of convolutional neutral network (CNN), to improve the localized error covariances for the Ensemble Kalman Filter (EnKF) algorithm. Using a 2-layer quasi-geostrophic model, U-Nets are trained using data from EnKF DA experiments. The trained U-Nets are then successfully implemented in U-Net Kalman Filter (UNetKF) experiments to predict localized error covariances that possess adaptive localization and some state-dependent features of the model error covariances. UNetKF is compared to traditional 3-dimensional variational (3DVar), ensemble 3DVar (En3DVar) and EnKF methods. The performance of UNetKF can match or exceed that of 3DVar, or En3DVar and EnKF for small to moderate ensemble sizes. We also demonstrate that trained U-Nets can be transferred to a higher-resolution model for UNetKF implementation, which again performs competitively to 3DVar and EnKF, particularly for small ensemble sizes.
The seasonal prediction skill of tropical cyclone (TC) activity is evaluated using the Seamless System for Prediction and Earth System Research (SPEAR), a modeling system developed at the Geophysical Fluid Dynamics Laboratory (GFDL) for experimental real-time seasonal forecasts. Compared with previous GFDL seasonal prediction models, SPEAR demonstrates improved skill in predicting TC activity for the western North Pacific, while exhibiting comparable or slightly degraded skill for the eastern North Pacific and North Atlantic. These changes in prediction skill do not always align with changes in prediction skill in large-scale variables, particularly over the North Atlantic. This study highlights that changes in the model’s response of TCs to large-scale variables, as well as the changes in the amplitude of interannual variations in TC genesis frequency, are crucial for the changes in TC prediction skill. Using the predicted sea surface temperatures from SPEAR as lower boundary conditions, the High-Resolution Forecast-Oriented Low Ocean Resolution (HiFLOR-S) model was employed to predict intense TCs, demonstrating skillful predictions of major hurricanes that are comparable to the previous HiFLOR coupled model predictions.
Balwada, Dhruv, Ryan Abernathey, Shantanu Acharya, Alistair Adcroft, Judith Brener, V Balaji, Mohamed Aziz Bhouri, Joan Bruna, Mitchell Bushuk, Will Chapman, Alex Connolly, Julie Deshayes, Carlos Fernandez-Granda, Pierre Gentine, Anastasiia Gorbunova, William Gregory, Arthur Guillaumin, Shubham Gupta, Marika M Holland, J Emmanuel Johnsson, Julien Le Sommer, Ziwei Li, Nora Loose, Feiyu Lu, Paul A O'Gorman, Pavel Perezhogin, Brandon G Reichl, Andrew C Ross, Aakash Sane, Sara Shamekh, Tarun Verma, Janni Yuval, Lorenzo Zampieri, Cheng Zhang, and Laure Zanna, December 2024: Learning machine learning with Lorenz-96. Journal of Open Source Education, 7(82), DOI:10.21105/jose.00241. Abstract
Machine learning (ML) is a rapidly growing field that is starting to touch all aspects of our lives, and science is not immune to this. In fact, recent
work in the field of scientific ML, i.e. combining ML and with conventional scientific problems, is leading to new breakthroughs in notoriously hard problems, which might have seemed too distant till a few years ago. One such age-old problem is that of turbulence closures in fluid flows. This closure or parameterization problem is particularly relevant for environmental fluids, which span a large range of scales from the size of the planet down to millimeters, and remains a big challenge in the way of improving forecasts of weather and projections of climate.
Humid heat extreme (HHE) is a type of compound extreme weather event that poses severe risks to human health. Skillful forecasts of HHE months in advance are crucial for developing strategies to enhance community resilience to extreme events1,2. This study demonstrates that the frequency of summertime HHE in the southeastern United States (SEUS) can be skillfully predicted 0–1 months in advance using the SPEAR (Seamless system for Prediction and EArth system Research) seasonal forecast system. Sea surface temperatures (SSTs) in the tropical North Atlantic (TNA) basin are identified as the primary driver of this prediction skill. The responses of large-scale atmospheric circulation and winds to anomalous warm SSTs in the TNA favor the transport of heat and moisture from the Gulf of Mexico to the SEUS. This research underscores the role of slowly varying sea surface conditions in modifying large-scale environments, thereby contributing to the skillful prediction of HHE in the SEUS. The results of this study have potential applications in the development of early warning systems for HHE.
Using a 1/12° regional model of the Northwest Atlantic Ocean (MOM6-NWA12), we downscale an ensemble of retrospective seasonal forecasts from a 1° global forecast model. To evaluate whether downscaling improved the forecast skill for surface temperature and salinity and bottom temperature, the global and downscaled forecasts are compared with each other and with a reference forecast of persistence using anomaly correlation. Both sets of forecasts are also evaluated on the basis of mean bias and ensemble spread. We find that downscaling significantly improved the forecast skill for monthly sea surface temperature anomalies in the Northeast US Large Marine Ecosystem, a region that global models have historically struggled to skillfully predict. The downscaled sea surface temperature (SST) predictions for this region were also more skillful than the persistence baseline across most initialization months and lead times. Although some of the SST prediction skill in this region stems from the recent rapid warming trend, prediction skill above persistence is generally maintained after removing the contribution of the trend, and patterns of skill suggestive of predictable processes are also preserved. While downscaling mainly improved the SST anomaly prediction skill in the Northeast US region, it improved bottom temperature and sea surface salinity anomaly skill across many of the marine ecosystems along the North American east coast. Although improvements in anomaly prediction via downscaling were ubiquitous, the effects of downscaling on prediction bias were mixed. Downscaling generally reduced the mean surface salinity biases found in the global model, particularly in regions with sharp salinity gradients (the Northern Gulf of Mexico and the Northeast US). In some cases, however, downscaling amplified the surface and bottom temperature biases found in the global predictions. We discuss several processes that are better resolved in the regional model and contribute to the improved skill, including the autumn reemergence of temperature anomalies and advection of water masses by coastal currents. Overall, the results show that a downscaled high-resolution model can produce improved seasonal forecast skill by representing fine-scale processes that drive predictability.
A key challenge with the wind energy utilization is that winds, and thus wind power, are highly variable on seasonal to interannual timescales because of atmospheric variability. There is a growing need of skillful seasonal wind energy prediction for energy system planning and operation. Here we demonstrate model’s capability in producing skillful seasonal wind energy prediction over the U.S. Great Plains during peak energy seasons (winter and spring), using seasonal prediction products from a climate model. The dominant source of that skillful prediction mainly comes from year-to-year variations of El Niño-Southern Oscillation in the tropical Pacific, which alters large-scale wind and storm track patterns over the United States. In the Southern Great Plains, the model can predict strong year-to-year wind energy changes with high skill multiple months in advance. Thus, this seasonal wind energy prediction capability offers potential benefits for optimizing wind energy utilization during peak energy production seasons.
Skillful prediction of wintertime cold extremes on seasonal time scales is beneficial for multiple sectors. This study demonstrates that North American cold extremes, measured by the frequency of cold days in winter, are predictable several months in advance in the Geophysical Fluid Dynamics Laboratory’s SPEAR (Seamless system for Prediction and EArth system Research) seasonal forecast system. Three predictable components of cold extremes over the North American continent are found to be skillfully predicted on seasonal scales. One is a trend-like component, which shows a continent-wide decrease in the frequency of cold extremes and is primarily attributable to external radiative forcing. This trend-like component is predictable at least 9 months ahead. The second predictable component displays a dipole structure over North America, with negative signs in the northwest and positive signs in the southeast. This dipole component is predictable with significant correlation skill for 2 months and is a response to the central Pacific ENSO (El Niño-Southern Oscillation) as revealed from SPEAR AMIP-style simulations. The third component with the largest loadings over Canada and the northern US shows significant correlations with snow anomalies over mid-to-high latitudes of the North American continent. Predictions using only the three predictable components yield higher/comparable skill relative to the SPEAR raw forecasts.
The Model-Analogs technique is used in the present study to assess the decadal sea surface temperature (SST) prediction skill over the Southern Ocean (SO). The Model-Analogs here is based on reanalysis products and model control simulations that have ∼1° ocean/ice (refined to 0.5° at high latitudes) components and 100 km atmosphere/land components. It is found that the model analog hindcasts show comparable skills with the initialized retrospective decadal hindcasts south of 50°S, with even higher skills over the Weddell Sea at longer lead years. The high SST skills primarily arise from the successful capture of SO deep convection states. This deep ocean memory and the associated decadal predictability are also clearly seen when we assess the Model-Analogs technique in a perfect model context. Within 30°S–50°S latitudinal band, the model analog hindcasts show low skills. When we include the externally forced signals estimated from the large ensemble simulations, the model analog hindcasts and initialized decadal hindcasts show identical skills. The Model-Analogs method therefore provides a great baseline for developing future decadal forecast systems. It is unclear whether such analog techniques would also be successful with models that explicitly resolve ocean mesoscale eddies or other small-scale processes. This area of research needs to be explored further.
Research over the past decade has demonstrated that dynamical forecast systems can skillfully predict pan-Arctic sea ice extent (SIE) on the seasonal time scale; however, there have been fewer assessments of prediction skill on user-relevant spatial scales. In this work, we evaluate regional Arctic SIE predictions made with the Forecast-Oriented Low Ocean Resolution (FLOR) and Seamless System for Prediction and Earth System Research (SPEAR_MED) dynamical seasonal forecast systems developed at the NOAA/Geophysical Fluid Dynamics Laboratory. Compared to FLOR, we find that the recently developed SPEAR_MED system displays improved skill in predicting regional detrended SIE anomalies, partially owing to improvements in sea ice concentration (SIC) and thickness (SIT) initial conditions. In both systems, winter SIE is skillfully predicted up to 11 months in advance, whereas summer minimum SIE predictions are limited by the Arctic spring predictability barrier, with typical skill horizons of roughly 4 months. We construct a parsimonious set of simple statistical prediction models to investigate the mechanisms of sea ice predictability in these systems. Three distinct predictability regimes are identified: a summer regime dominated by SIE and SIT anomaly persistence; a winter regime dominated by SIE and upper-ocean heat content (uOHC) anomaly persistence; and a combined regime in the Chukchi Sea, characterized by a trade-off between uOHC-based and SIT-based predictability that occurs as the sea ice edge position evolves seasonally. The combination of regional SIE, SIT, and uOHC predictors is able to reproduce the SIE skill of the dynamical models in nearly all regions, suggesting that these statistical predictors provide a stringent skill benchmark for assessing seasonal sea ice prediction systems.
This study shows that the frequency of North American summertime (June–August) heat extremes is skillfully predicted several months in advance in the newly developed Geophysical Fluid Dynamics Laboratory (GFDL) Seamless System for Prediction and Earth System Research (SPEAR) seasonal forecast system. Using a statistical optimization method, the average predictability time, we identify three large-scale components of the frequency of North American summer heat extremes that are predictable with significant correlation skill. One component, which is related to a secular warming trend, shows a continent-wide increase in the frequency of summer heat extremes and is highly predictable at least 9 months in advance. This trend component is likely a response to external radiative forcing. The second component is largely driven by the sea surface temperatures in the North Pacific and North Atlantic and is significantly correlated with the central U.S. soil moisture. The second component shows largest loadings over the central United States and is significantly predictable 9 months in advance. The third component, which is related to the central Pacific El Niño, displays a dipole structure over North America and is predictable up to 4 months in advance. Potential implications for advancing seasonal predictions of North American summertime heat extremes are discussed.
The Kuroshio Extension (KE), an eastward-flowing jet located in the Pacific western boundary current system, exhibits prominent seasonal-to-decadal variability, which is crucial for understanding climate variations in the northern midlatitudes. We explore the representation and prediction skill for the KE in the GFDL SPEAR (Seamless System for Prediction and Earth System Research) coupled model. Two different approaches are used to generate coupled reanalyses and forecasts: 1) restoring the coupled model’s SST and atmospheric variables toward existing reanalyses, or 2) assimilating SST and subsurface observations into the coupled model without atmospheric assimilation. Both systems use an ocean model with 1° resolution and capture the largest sea surface height (SSH) variability over the KE region. Assimilating subsurface observations appears to be essential to reproduce the narrow front and related oceanic variability of the KE jet in the coupled reanalysis. We demonstrate skillful retrospective predictions of KE SSH variability in monthly (up to 1 year) and annual-mean (up to 5 years) KE forecasts in the seasonal and decadal prediction systems, respectively. The prediction skill varies seasonally, peaking for forecasts initialized in January and verifying in September due to the winter intensification of North Pacific atmospheric forcing. We show that strong large-scale atmospheric anomalies generate deterministic oceanic forcing (i.e., Rossby waves), leading to skillful long-lead KE forecasts. These atmospheric anomalies also drive Ekman convergence and divergence, which forms ocean memory, by sequestering thermal anomalies deep into the winter mixed layer that re-emerge in the subsequent autumn. The SPEAR forecasts capture the recent negative-to-positive transition of the KE phase in 2017, projecting a continued positive phase through 2022.
Quantifying the response of atmospheric rivers (ARs) to radiative forcing is challenging due to uncertainties caused by internal climate variability, differences in shared socioeconomic pathways (SSPs), and methods used in AR detection algorithms. In addition, the requirement of medium-to-high model resolution and ensemble sizes to explicitly simulate ARs and their statistics can be computationally expensive. In this study, we leverage the unique 50-km large ensembles generated by a Geophysical Fluid Dynamics Laboratory next-generation global climate model, Seamless system for Prediction and EArth system Research, to explore the warming response in ARs. Under both moderate and high emissions scenarios, increases in AR-day frequency emerge from the noise of internal variability by 2060. This signal is robust across different SSPs and time-independent detection criteria. We further examine an alternative approach proposed by Thompson et al. (2015), showing that unforced AR variability can be approximated by a first-order autoregressive process. The confidence intervals of the projected response can be analytically derived with a single ensemble member.
The continuing decline of the summertime sea ice cover has reduced the sea ice path that must be traversed to Arctic destinations and through the Arctic between the Atlantic and Pacific Oceans, stimulating interest in trans–Arctic Ocean routes. Seasonal prediction of the sea ice cover along these routes could support the increasing summertime ship traffic taking advantage of recent low ice conditions. We introduce the minimum Arctic sea ice path (MIP) between Atlantic and Pacific Oceans as a shipping-relevant metric that is amenable to multidecadal hindcast evaluation. We show, using 1992–2017 retrospective predictions, that bias correction is necessary for the GFDL Seamless System for Prediction and Earth System Research (SPEAR) forecast system to improve upon damped persistence seasonal forecasts of summertime daily MIP between the Atlantic and Pacific Oceans both east and west of Greenland, corresponding roughly to the Northeast and Northwest Passages. Without bias correction, only the Northwest Passage MIP forecasts have lower error than a damped persistence forecast. Using the forecast ensemble spread to estimate a lower bound on forecast error, we find large opportunities for forecast error reduction, especially at lead times of less than 2 months. Most of the potential improvement remains after linear removal of climatological and trend biases, suggesting that significant error reduction might come from improved initialization and simulation of subannual variability. Using a different passive microwave sea ice dataset for calculating error than was used for data assimilation increases the raw forecast errors but not the trend anomaly forecast errors.
A subseasonal-to-seasonal (S2S) prediction system was recently developed using the GFDL Seamless System for Prediction and Earth System Research (SPEAR) global coupled model. Based on 20-yr hindcast results (2000–19), the boreal wintertime (November–April) Madden–Julian oscillation (MJO) prediction skill is revealed to reach 30 days measured before the anomaly correlation coefficient of the real-time multivariate (RMM) index drops to 0.5. However, when the MJO is partitioned into four distinct propagation patterns, the prediction range extends to 38, 31, and 31 days for the fast-propagating, slow-propagating, and jumping MJO patterns, respectively, but falls to 23 days for the standing MJO. A further improvement of MJO prediction requires attention to the standing MJO given its large gap with its potential predictability (38 days). The slow-propagating MJO detours southward when traversing the Maritime Continent (MC), and confronts the MC prediction barrier in the model, while the fast-propagating MJO moves across the central MC without this prediction barrier. The MJO diversity is modulated by stratospheric quasi-biennial oscillation (QBO): the standing (slow-propagating) MJO coincides with significant westerly (easterly) phases of QBO, partially explaining the contrasting MJO prediction skill between these two QBO phases. The SPEAR model shows its capability, beyond the propagation, in predicting their initiation for different types of MJO along with discrete precursory convection anomalies. The SPEAR model skillfully predicts the observed distinct teleconnections over the North Pacific and North America related to the standing, jumping, and fast-propagating MJO, but not the slow-propagating MJO. These findings highlight the complexities and challenges of incorporating MJO prediction into the operational prediction of meteorological variables.
The rapid day-to-day temperature swings associated with extratropical storm tracks can cause cascading infrastructure failure and impact human outdoor activities, thus research on seasonal prediction and predictability of extreme temperature swings is of huge societal importance. To measure the extreme surface air temperature (SAT) variations associated with the winter extratropical storm tracks, a Temperature Swing Index (TSI) is formulated as the standard deviation of 24-h-difference-filtered data of the 6-hourly SAT. The dominant term governing the TSI variability is shown to be proportional to the product of eddy heat flux and mean temperature gradient. The seasonal prediction skill of the winter TSI over North America was assessed using Geophysical Fluid Dynamics Laboratory's new seasonal prediction system. The locations with skillful TSI prediction show a geographic pattern that is distinct from the pattern of skillful seasonal mean SAT prediction. The prediction of TSI provides additional predictable climate information beyond the traditional seasonal mean temperature prediction. The source of the seasonal TSI prediction can be attributed to year-to-year variations of the El Niño-Southern Oscillation (ENSO), North Pacific Oscillation (NPO), and Pacific/North American (PNA) teleconnection. Over the central United States, the correlation skill of TSI prediction reaches 0.75 with strong links to observed ENSO, NPO, and PNA, while the skill of seasonal SAT prediction is relatively low with a correlation of 0.36. As a first attempt of diagnosing the combined predictability of the first-order (the seasonal mean) and second-order (TSI) statistics for SAT, this study highlights the importance of the eddy-mean flow interaction perspective for understanding the seasonal climate predictability in the extra tropics. These results point toward providing skillful prediction of higher-order statistical information related to winter temperature extremes, thus enriching the seasonal forecast products for the research community and decision makers.
One of the most puzzling observed features of recent climate has been a multidecadal surface cooling trend over the subpolar Southern Ocean (SO). In this study we use large ensembles of simulations with multiple climate models to study the role of the SO meridional overturning circulation (MOC) in these sea surface temperature (SST) trends. We find that multiple competing processes play prominent roles, consistent with multiple mechanisms proposed in the literature for the observed cooling. Early in the simulations (twentieth century and early twenty-first century) internal variability of the MOC can have a large impact, in part due to substantial simulated multidecadal variability of the MOC. Ensemble members with initially strong convection (and related surface warming due to convective mixing of subsurface warmth to the surface) tend to subsequently cool at the surface as convection associated with internal variability weakens. A second process occurs in the late-twentieth and twenty-first centuries, as weakening of oceanic convection associated with global warming and high-latitude freshening can contribute to the surface cooling trend by suppressing convection and associated vertical mixing of subsurface heat. As the simulations progress, the multidecadal SO variability is suppressed due to forced changes in the mean state and increased oceanic stratification. As a third process, the shallower mixed layers can then rapidly warm due to increasing forcing from greenhouse gas warming. Also, during this period the ensemble spread of SO SST trend partly arises from the spread of the wind-driven Deacon cell strength. Thus, different processes could conceivably have led to the observed cooling trend, consistent with the range of possibilities presented in the literature. To better understand the causes of the observed trend, it is important to better understand the characteristics of internal low-frequency variability in the SO and the response of that variability to global warming.
The current GFDL seasonal prediction system, the Seamless System for Prediction and Earth System Research (SPEAR), has shown skillful prediction of Arctic sea ice extent with atmosphere and ocean constrained by observations. In this study we present improvements in subseasonal and seasonal predictions of Arctic sea ice by directly assimilating sea ice observations. The sea ice initial conditions from a data assimilation (DA) system that assimilates satellite sea ice concentration (SIC) observations are used to produce a set of reforecast experiments (IceDA) starting from the first day of each month from 1992 to 2017. Our evaluation of daily sea ice extent prediction skill concludes that the SPEAR system generally outperforms the anomaly persistence forecast at lead times beyond 1 month. We primarily focus our analysis on daily gridcell-level sea ice fields. SIC DA improves prediction skill of SIC forecasts prominently in the June-, July-, August-, and September-initialized reforecasts. We evaluate two additional user-oriented metrics: the ice-free probability (IFP) and ice-free date (IFD). IFP is the probability of a grid cell experiencing ice-free conditions in a given year, and IFD is the first date on which a grid cell is ice free. A combined analysis of IFP and IFD demonstrates that the SPEAR model can make skillful predictions of local ice melt as early as May, with modest improvements from SIC DA.
The low Antarctic sea ice extent following its dramatic decline in late 2016 has persisted over a multiyear period. However, it remains unclear to what extent this low sea ice extent can be attributed to changing ocean conditions. Here, we investigate the causes of this period of low Antarctic sea ice extent using a coupled climate model partially constrained by observations. We find that the subsurface Southern Ocean played a smaller role than the atmosphere in the extreme sea ice extent low in 2016, but was critical for the persistence of negative anomalies over 2016–2021. Prior to 2016, the subsurface Southern Ocean warmed in response to enhanced westerly winds. Decadal hindcasts show that subsurface warming has persisted and gradually destabilized the ocean from below, reducing sea ice extent over several years. The simultaneous variations in the atmosphere and ocean after 2016 have further amplified the decline in Antarctic sea ice extent.
Compared to the Arctic, seasonal predictions of Antarctic sea ice have received relatively little attention. In this work, we utilize three coupled dynamical prediction systems developed at the Geophysical Fluid Dynamics Laboratory to assess the seasonal prediction skill and predictability of Antarctic sea ice. These systems, based on the FLOR, SPEAR_LO, and SPEAR_MED dynamical models, differ in their coupled model components, initialization techniques, atmospheric resolution, and model biases. Using suites of retrospective initialized seasonal predictions spanning 1992–2018, we investigate the role of these factors in determining Antarctic sea ice prediction skill and examine the mechanisms of regional sea ice predictability. We find that each system is capable of skillfully predicting regional Antarctic sea ice extent (SIE) with skill that exceeds a persistence forecast. Winter SIE is skillfully predicted 11 months in advance in the Weddell, Amundsen/Bellingshausen, Indian, and west Pacific sectors, whereas winter skill is notably lower in the Ross sector. Zonally advected upper-ocean heat content anomalies are found to provide the crucial source of prediction skill for the winter sea ice edge position. The recently developed SPEAR systems are more skillful than FLOR for summer sea ice predictions, owing to improvements in sea ice concentration and sea ice thickness initialization. Summer Weddell SIE is skillfully predicted up to 9 months in advance in SPEAR_MED, due to the persistence and drift of initialized sea ice thickness anomalies from the previous winter. Overall, these results suggest a promising potential for providing operational Antarctic sea ice predictions on seasonal time scales.
Atmospheric rivers (ARs) exert significant socioeconomic impacts in western North America, where 30% of the annual precipitation is determined by ARs that occur in less than 15% of wintertime. ARs are thus beneficial to water supply but can produce extreme precipitation hazards when making landfall. While most prevailing research has focused on the subseasonal (<5 weeks) prediction of ARs, only limited efforts have been made for AR forecasts on multiseasonal timescales (>3 months) that are crucial for water resource management and disaster preparedness. Through the analysis of reanalysis data and retrospective predictions from a new seasonal-to-decadal forecast system, this research shows the existing potential of multiseasonal AR frequency forecasts with predictive skills 9 months in advance. Additional analysis explores the dominant predictability sources and challenges for multiseasonal AR prediction.
Midlatitude baroclinic waves drive extratropical weather and climate variations, but their predictability beyond 2 weeks has been deemed low. Here we analyze a large ensemble of climate simulations forced by observed sea surface temperatures (SSTs) and demonstrate that seasonal variations of baroclinic wave activity (BWA) are potentially predictable. This potential seasonal predictability is denoted by robust BWA responses to SST forcings. To probe regional sources of the potential predictability, a regression analysis is applied to the SST-forced large ensemble simulations. By filtering out variability internal to the atmosphere and land, this analysis identifies both well-known and unfamiliar BWA responses to SST forcings across latitudes. Finally, we confirm the model-indicated predictability by showing that an operational seasonal prediction system can leverage some of the identified SST-BWA relationships to achieve skillful predictions of BWA. Our findings help to extend long-range predictions of the statistics of extratropical weather events and their impacts.
We document the development and simulation characteristics of the next generation modeling system for seasonal to decadal prediction and projection at the Geophysical Fluid Dynamics Laboratory (GFDL). SPEAR (Seamless System for Prediction and EArth System Research) is built from component models recently developed at GFDL ‐ the AM4 atmosphere model, MOM6 ocean code, LM4 land model and SIS2 sea ice model. The SPEAR models are specifically designed with attributes needed for a prediction model for seasonal to decadal time scales, including the ability to run large ensembles of simulations with available computational resources. For computational speed SPEAR uses a coarse ocean resolution of approximately 1.0o (with tropical refinement). SPEAR can use differing atmospheric horizontal resolutions ranging from 1o to 0.25o. The higher atmospheric resolution facilitates improved simulation of regional climate and extremes. SPEAR is built from the same components as the GFDL CM4 and ESM 4 models, but with design choices geared toward seasonal to multidecadal physical climate prediction and projection. We document simulation characteristics for the time‐mean climate, aspects of internal variability, and the response to both idealized and realistic radiative forcing change. We describe in greater detail one focus of the model development process that was motivated by the importance of the Southern Ocean to the global climate system. We present sensitivity tests that document the influence of the Antarctic surface heat budget on Southern Ocean ventilation and deep global ocean circulation. These findings were also useful in the development processes for the GFDL CM4 and ESM 4 models.
Lu, Lv, Shaoqing Zhang, Stephen G Yeager, Gokhan Danabasoglu, Ping Chang, Lixin Wu, Xiaopei Lin, Anthony Rosati, and Feiyu Lu, September 2020: Impact of Coherent Ocean Stratification on AMOC Reconstruction by Coupled Data Assimilation with a Biased Model. Journal of Climate, 33(17), DOI:10.1175/JCLI-D-19-0735.1. Abstract
The Atlantic meridional overturning circulation (AMOC) is of great importance in Earth’s climate system, and reconstructing its structure and variability by combining observations with a coupled model is a key step in understanding historical and future states of AMOC. However, models always have systematic errors called bias owing to imperfect numerical representation of the real world. Model bias and the sparse nature of ocean observations, particularly in deep oceans, make it difficult to generate a complete historical picture of AMOC structure and variability. Here, two coupled models that are biased with respect to each other are used to design “twin” experiments to systematically study the influence of model bias on AMOC reconstruction. One model is used to produce the “observations” that sample the “true” solution of the AMOC to be reconstructed, while the other model is used to incorporate the “observations” to reconstruct the “truth” through coupled data assimilation (CDA). The degree to which the “truth” is recovered by a CDA scheme assesses the critical role of coherent (both upper- and deep-ocean incorporate enough observations to mitigate stratification instability) ocean stratification on AMOC reconstruction. Results show that balancing restoration of climatology and assimilation of observations is vital to better reconstruct AMOC structure and variability, given that most ocean observations are only available in the upper 2000 m. The gained results serve as a guideline in ocean-state estimation with a balance of deep restoring and upper data constraint for climate prediction initialization, especially for decadal predictions.
The next‐generation seasonal prediction system is built as part of the Seamless System for Prediction and EArth System Research (SPEAR) at the Geophysical Fluid Dynamics Laboratory (GFDL) of the National Oceanic and Atmospheric Administration (NOAA). SPEAR is an effort to develop a seamless system for prediction and research across time scales. The ensemble‐based ocean data assimilation (ODA) system is updated for Modular Ocean Model Version 6 (MOM6), the ocean component of SPEAR. Ocean initial conditions for seasonal predictions, as well as an ocean state estimation, are produced by the MOM6 ODA system in coupled SPEAR models. Initial conditions of the atmosphere, land, and sea ice components for seasonal predictions are constructed through additional nudging experiments in the same coupled SPEAR models. A bias correction scheme called ocean tendency adjustment (OTA) is applied to coupled model seasonal predictions to reduce model drift. OTA applies the climatological temperature and salinity increments obtained from ODA as three‐dimensional tendency terms to the MOM6 ocean component of the coupled SPEAR models. Based on preliminary retrospective seasonal forecasts, we demonstrate that OTA reduces model drift—especially sea surface temperature (SST) forecast drift—in coupled model predictions and improves seasonal prediction skill for applications such as El Niño–Southern Oscillation (ENSO).
Sun, Jingzhe, Zhengyu Liu, Feiyu Lu, Weimin Zhang, and Shaoqing Zhang, June 2020: Strongly Coupled Data Assimilation Using Leading Averaged Coupled Covariance (LACC). Part III: Assimilation of Real World Reanalysis. Monthly Weather Review, 148(6), DOI:10.1175/MWR-D-19-0304.1. Abstract
Recent studies proposed LACC (leading averaged coupled covariance) as an effective strongly coupled data assimilation (SCDA) method to improve the coupled state estimation over weakly coupled data assimilation (WCDA) in a coupled general circulation model (CGCM). This SCDA method, however, has been previously evaluated only in the perfect model scenario. Here, as a further step towards evaluating LACC for real world data assimilation, LACC is evaluated for the assimilation of reanalysis data in a CGCM. Several criterions are used to evaluate LACC against the benchmark WCDA. It is shown that despite significant model bias, LACC can improve the coupled state estimation over WCDA. Compared to WCDA, LACC increases the globally averaged anomaly correlation coefficients (ACCs) of sea surface temperature (SST) by 0.036 and atmosphere temperature at the bottom level (Ts) by 0.058. However, there also exist regions where WCDA outperforms LACC. Although the reduction in the anomaly root-mean-square error (RMSE) is not as consistently clear as the increase in ACC, LACC can largely correct the biased model climatology.
Zhang, Shaoqing, Zhengyu Liu, X-F Zhang, Xinrong Wu, G Han, Y Zhao, X Yu, C Liu, Y Liu, S Wu, and Feiyu Lu, et al., June 2020: Coupled data assimilation and parameter estimation in coupled ocean–atmosphere models: a review. Climate Dynamics, 54(11-12), DOI:10.1007/s00382-020-05275-6. Abstract
Recent studies have started to explore coupled data assimilation (CDA) in coupled ocean–atmosphere models because of the great potential of CDA to improve climate analysis and seamless weather–climate prediction on weekly-to-decadal time scales in advanced high-resolution coupled models. In this review article, we briefly introduce the concept of CDA before outlining its potential for producing balanced and coherent weather–climate reanalysis and minimizing initial coupling shocks. We then describe approaches to the implementation of CDA and review progress in the development of various CDA methods, notably weakly and strongly coupled data assimilation. We introduce the method of coupled model parameter estimation (PE) within the CDA framework and summarize recent progress. After summarizing the current status of the research and applications of CDA-PE, we discuss the challenges and opportunities in high-resolution CDA-PE and nonlinear CDA-PE methods. Finally, potential solutions are laid out.
Liu, Zhengyu, C He, and Feiyu Lu, August 2018: Local and Remote Responses of Atmospheric and Oceanic Heat Transports to Climate Forcing: Compensation versus Collaboration. Journal of Climate, 31(16), DOI:10.1175/JCLI-D-17-0675.1. Abstract
We present a theoretical study on local and remote responses of atmosphere and ocean meridional heat transports (AHT and OHT, respectively) to climate forcing in a coupled energy balance model. We show that, in general, a surface heat flux forces opposite AHT and OHT responses in the so-called compensation response, while a net heat flux into the coupled system forces AHT and OHT responses of the same direction in the so-called collaboration response. Furthermore, unless the oceanic thermohaline circulation is significantly changed, a remote climate response far away from the forcing region tends to be dominated by the collaboration response, because of the effective propagation of a coupled ocean–atmosphere energy transport mode of collaboration structure. The relevance of our theory to previous CGCM experiments is also discussed. Our theoretical result provides a guideline for understanding of the response of heat transports and the associated climate changes.
Lu, Feiyu, and Zhengyu Liu, November 2018: Assessing Extratropical Influence on Observed El Nino-Southern Oscillation Events Using Regional Coupled Data Assimilation. Journal of Climate, 31(21), DOI:10.1175/JCLI-D-17-0849.1. Abstract
The extratropical influence on the observed events of El Niño–Southern Oscillation (ENSO) variability from 1948 to 2015 is assessed by constraining the extratropical atmospheric variability in a coupled general circulation model (CGCM) using the regional coupled data assimilation (RCDA) method. The ensemble-mean ENSO response to extratropical atmospheric forcing, which is systematically and quantitatively studied through a series of RCDA experiments, indicates robust extratropical influence on some observed ENSO events. Furthermore, an event-by-event quantitative analysis shows significant differences of the extratropical influence among the observed ENSO events, both in its own strength and in its relation to tropical precursors such as the equatorial Pacific heat content anomaly. This study provides the first dynamic quantitative assessment of the extratropical influence on observed ENSO variability on an event-by-event basis.
Lu, Feiyu, Zhengyu Liu, Y Liu, Shaoqing Zhang, and R Jacob, May 2017: Understanding the control of extratropical atmospheric variability on ENSO using a coupled data assimilation approach. Climate Dynamics, 48(9), DOI:10.1007/s00382-016-3256-7. Abstract
The control of extratropical atmospheric variability on ENSO variability is studied in a coupled general circulation model (CGCM) utilizing an ensemble-based coupled data assimilation (CDA) method in the perfect-model framework. Assimilation is limited to the desired model components (e.g. atmosphere) and spatial areas (e.g. the extratropics) to study the ensemble-mean model response (e.g. tropical response to “observed” extratropical atmospheric variability). The CDA provides continuously “corrected” extratropical atmospheric forcing and boundary conditions for the tropics and the use of ensemble optimizes the observational forcing signal over internal variability in the model component or region without assimilation. The experiments demonstrate significant control of extratropical atmospheric forcing on ENSO variability in the CGCM. When atmospheric “observations” are assimilated only poleward of 20° in both hemispheres, most ENSO events in the “observation” are reproduced and the error of the Nino3.4 index is reduced by over 40 % compared to the ensemble control experiment that does not assimilate any observations. Further experiments with the assimilation in each hemisphere show that the forced ENSO variability is contributed roughly equally and independently by the Southern and Northern Hemisphere extratropical atmosphere. Further analyses of the ENSO events in the southern hemisphere forcing experiment reveal robust precursors in both the extratropical atmosphere over southeastern Pacific and equatorial Pacific thermocline, consistent with previous studies of the South Pacific Meridional Mode and the discharge-recharge paradigm, respectively. However, composite analyses based on each precursor show that neither precursor alone is sufficient to trigger ENSO onset by itself and therefore neither alone could serve as a reliable predictor. Additional experiments with northern hemisphere forcing, ocean assimilation or different latitudes are also performed.
This paper tests the idea of substituting the atmospheric observations with atmospheric reanalysis when setting up a coupled data assimilation system. The paper focuses on the quantification of the effects on the oceanic analysis resulted from this substitution and designs four different assimilation schemes for such a substitution. A coupled Lorenz96 system is constructed and an ensemble Kalman filter is adopted. The atmospheric reanalysis and oceanic observations are assimilated into the system and the analysis quality is compared to a benchmark experiment where both atmospheric and oceanic observations are assimilated. Four schemes are designed for assimilating the reanalysis and they differ in the generation of the perturbed observation ensemble and the representation of the error covariance matrix. The results show that when the reanalysis is assimilated directly as independent observations, the root-mean-square error increase of oceanic analysis relative to the benchmark is less than 16% in the perfect model framework; in the biased model case, the increase is less than 22%. This result is robust with sufficient ensemble size and reasonable atmospheric observation quality (e.g., frequency, noisiness, and density). If the observation is overly noisy, infrequent, sparse, or the ensemble size is insufficiently small, the analysis deterioration caused by the substitution is less severe since the analysis quality of the benchmark also deteriorates significantly due to worse observations and undersampling. The results from different assimilation schemes highlight the importance of two factors: accurate representation of the error covariance of the reanalysis and the temporal coherence along each ensemble member, which are crucial for the analysis quality of the substitution experiment.
Lu, Feiyu, Zhengyu Liu, Shaoqing Zhang, and Y Liu, September 2015: Strongly Coupled Data Assimilation Using Leading Averaged Coupled Covariance (LACC). Part I: Simple Model Study. Monthly Weather Review, 143(9), DOI:10.1175/MWR-D-14-00322.1. Abstract
This paper studies a new Leading Averaged Coupled Covariance (LACC) method for the strongly coupled data assimilation (SCDA). The SCDA not only uses the coupled model to generate the forecast and assimilates observations into multiple model components like the weakly coupled version (WCDA), but also applies cross update using the coupled covariance between variables from different model components. The cross update could potentially improve the balance and quality of the analysis, but its implementation has remained a great challenge in practice due to different timescales between model components.
In a typical extra-tropical coupled system, the ocean-atmosphere correlation shows a strong asymmetry with the maximum correlation occurring when the atmosphere leads the ocean by about the decorrelation time of the atmosphere. The LACC method utilizes such asymmetric structure by using the leading forecasts and observations of the fast atmospheric variable for cross update, therefore increasing the coupled correlation and enhancing the signal-to-noise ratio in calculating the coupled covariance. Here it is applied to a simple coupled model with the Ensemble Kalman Filter (EnKF). With the LACC method, the SCDA reduces the analysis error of the oceanic variable by over 20% compared to the WCDA and 10% compared to the SCDA using simultaneous coupled covariance. The advantage of the LACC method is more notable when the system contains larger errors, such as in the cases with smaller ensemble size, bigger timescale difference or model biases.
Lu, Feiyu, Zhengyu Liu, Shaoqing Zhang, Y Liu, and R Jacob, November 2015: Strongly Coupled Data Assimilation Using Leading Averaged Coupled Covariance (LACC). Part II: CGCM experiments. Monthly Weather Review, 143(11), DOI:10.1175/MWR-D-15-0088.1. Abstract
This paper uses a fully coupled general circulation model (CGCM) to study the Leading Averaged Coupled Covariance (LACC) method in a strongly coupled data assimilation (SCDA) system. Our previous study in a simple coupled climate model (Lu et al. 2015) has shown that, by calculating the coupled covariance using the leading averaged atmospheric states, the LACC method enhances the signal-to-noise ratio and improves the analysis quality of the slow model component compared to both the traditional weakly coupled data assimilation without cross-component adjustments (WCDA) and the regular SCDA using the simultaneous coupled covariance (SimCC).
Here in Part II, we test the LACC method with a CGCM in a perfect-model framework. By adding the observational adjustments from the low-level atmosphere temperature to the sea surface temperature (SST), the SCDA using LACC significantly reduces the SST error compared to WCDA over the globe; it also improves from the SCDA using SimCC, which performs better than the WCDA only in the deep tropics. The improvement in SST analysis is a result of the enhanced signal-to-noise ratio in the LACC method, especially in the extra-tropical regions. The improved SST analysis also benefits the subsurface ocean temperature and low-level atmosphere temperature analyses through dynamic and statistical processes.
