arXiv subjects

Istvan Szunyogh

Publications and source records attributed to Istvan Szunyogh.

9 recordsLinked to original sources

arXiv2025

Model Training, Data Assimilation, and Forecast Experiments with a Hybrid Atmospheric Model that Incorporates Machine Learning

The hybrid model combines the physics-based primitive-equations model SPEEDY with a machine learning-based (ML-based) model component, while ERA5 reanalyses provide the presumed true states of the atmosphere. Six-hourly simulated noisy observations are generated for a 30-year ML training period and a one-year testing period. These observations are assimilated with a Local Ensemble Transform Kalman Filter (LETKF), and a 10-day deterministic forecast is also started from each ensemble mean analysis of the testing period. In the first experiment, the physics-based model provides the background ensemble members and the 10-day deterministic forecasts. In the other three experiments, the hybrid model plays the same role as the physics-based model in the first experiment, but it is trained on a different data set in each experiment. These training data sets are analyses obtained by using the physics-based model (second experiment), the hybrid model of the previous experiment (third experiment), and for comparison, ERA5 reanalyses (fourth experiment). The results of the experiments show that hybridizing the model can substantially improve the accuracy of the analyses and forecasts. When the model is trained on ERA5 reanalyses, the biases of the analyses are negligible and the magnitude of the flow-dependent part of the analysis errors is greatly reduced. While the gains in analysis accuracy are distinctly more modest in the other two hybrid model experiments, the gains in forecast accuracy tend to be larger in those experiments after 1-3 forecast days. However, these extra gains of forecast accuracy are achieved, in part, by a modest gradual reduction of the spatial variability of the forecasts.

nlin.CD↗

arXiv2024

Prediction Beyond the Medium Range with an Atmosphere-Ocean Model that Combines Physics-based Modeling and Machine Learning

This paper explores the potential of a hybrid modeling approach that combines machine learning (ML) with conventional physics-based modeling for weather prediction beyond the medium range. It extends the work of Arcomano et al. (2022), which tested the approach for short- and medium-range weather prediction, and the work of Arcomano et al. (2023), which investigated its potential for climate modeling. The hybrid model used for the forecast experiments of the paper is based on the low-resolution, simplified parameterization atmospheric general circulation model SPEEDY. In addition to the hybridized prognostic variables of SPEEDY, the model has three purely ML-based prognostic variables: the 6h cumulative precipitation, the sea surface temperature, and the heat content of the top 300m deep layer of the ocean (a new addition compared to the model used in Arcomano et al., 2023). The model has skill in predicting the El Nino cycle and its global teleconnections with precipitation for 3-7 months depending on the season. The model captures equatorial variability of the precipitation associated with Kelvin and Rossby waves and MJO. Predictions of the precipitation in the equatorial region have skill for 15 days in the East Pacific and 11.5 days in the West Pacific. Though the model has low spatial resolution, for these tasks it has prediction skill comparable to what has been published for high-resolution, purely physics-based, conventional, operational forecast models.

physics.ao-ph↗

arXiv2021

Using Data Assimilation to Train a Hybrid Forecast System that Combines Machine-Learning and Knowledge-Based Components

We consider the problem of data-assisted forecasting of chaotic dynamical systems when the available data is in the form of noisy partial measurements of the past and present state of the dynamical system. Recently there have been several promising data-driven approaches to forecasting of chaotic dynamical systems using machine learning. Particularly promising among these are hybrid approaches that combine machine learning with a knowledge-based model, where a machine-learning technique is used to correct the imperfections in the knowledge-based model. Such imperfections may be due to incomplete understanding and/or limited resolution of the physical processes in the underlying dynamical system, e.g., the atmosphere or the ocean. Previously proposed data-driven forecasting approaches tend to require, for training, measurements of all the variables that are intended to be forecast. We describe a way to relax this assumption by combining data assimilation with machine learning. We demonstrate this technique using the Ensemble Transform Kalman Filter (ETKF) to assimilate synthetic data for the 3-variable Lorenz system and for the Kuramoto-Sivashinsky system, simulating model error in each case by a misspecified parameter value. We show that by using partial measurements of the state of the dynamical system, we can train a machine learning model to improve predictions made by an imperfect knowledge-based model.

cs.LG↗

arXiv2020

Combining Machine Learning with Knowledge-Based Modeling for Scalable Forecasting and Subgrid-Scale Closure of Large, Complex, Spatiotemporal Systems

We consider the commonly encountered situation (e.g., in weather forecasting) where the goal is to predict the time evolution of a large, spatiotemporally chaotic dynamical system when we have access to both time series data of previous system states and an imperfect model of the full system dynamics. Specifically, we attempt to utilize machine learning as the essential tool for integrating the use of past data into predictions. In order to facilitate scalability to the common scenario of interest where the spatiotemporally chaotic system is very large and complex, we propose combining two approaches:(i) a parallel machine learning prediction scheme; and (ii) a hybrid technique, for a composite prediction system composed of a knowledge-based component and a machine-learning-based component. We demonstrate that not only can this method combining (i) and (ii) be scaled to give excellent performance for very large systems, but also that the length of time series data needed to train our multiple, parallel machine learning components is dramatically less than that necessary without parallelization. Furthermore, considering cases where computational realization of the knowledge-based component does not resolve subgrid-scale processes, our scheme is able to use training data to incorporate the effect of the unresolved short-scale dynamics upon the resolved longer-scale dynamics ("subgrid-scale closure").

cs.LG↗

arXiv2011

Ensemble regional data assimilation using joint states

We propose a data assimilation scheme that produces the analyses for a global and an embedded limited area model simultaneously, considering forecast information from both models. The purpose of the proposed approach is twofold. First, we expect that the global analysis will benefit from incorporation of information from the higher resolution limited area model. Second, our method is expected to produce a limited area analysis that is more strongly constrained by the large scale flow than a conventional limited area analysis. The proposed scheme minimizes a cost function in which the control variable is the joint state of the global and the limited area models. In addition, the cost function includes a constraint term that penalizes large differences between the global and the limited area state estimates. The proposed approach is tested by idealized experiments, using `toy' models introduced by Lorenz in 2005. The results of these experiments suggest that the proposed approach improves the global analysis within and near the limited area domain and the regional analysis near the lateral boundaries. These analysis improvements lead to forecast improvements in both the global and the limited area models.

nlin.CD↗

arXiv2011

On the propagation of information and the use of localization in ensemble Kalman filtering

Several localized versions of the ensemble Kalman filter have been proposed. Although tests applying such schemes have proven them to be extremely promising, a full basic understanding of the rationale and limitations of localization is currently lacking. It is one of the goals of this paper to contribute toward addressing this issue. The second goal is to elucidate the role played by chaotic wave dynamics in the propagation of information and the resulting impact on forecasts. To accomplish these goals, the principal tool used here will be analysis and interpretation of numerical experiments on a toy atmospheric model introduced by Lorenz in 2005. Propagation of the wave packets of this model is shown. It is found that, when an ensemble Kalman filter scheme is employed, the spatial correlation function obtained at each forecast cycle by averaging over the background ensemble members is short ranged, and this is in strong contrast to the much longer range correlation function obtained by averaging over states from free evolution of the model. Propagation of the effects of observations made in one region on forecasts in other regions is studied. The error covariance matrices from the analyses with localization and without localization are compared. From this study, major characteristics of the localization process and information propagation are extracted and summarized.

nlin.CD↗

arXiv2007

Comparison between Local Ensemble Transform Kalman Filter and PSAS in the NASA finite volume GCM: perfect model experiments

This paper explores the potential of Local Ensemble Transform Kalman Filter (LETKF) by comparing the performance of LETKF with an operational 3D-Var assimilation system, Physical-Space Statistical Analysis System (PSAS), under a perfect model scenario. The comparison is carried out on the finite volume Global Circulation Model (fvGCM) with 72 grid points zonally, 46 grid points meridionally and 55 vertical levels. With only forty ensemble members, LETKF obtains an analysis and forecasts with lower RMS errors than those from PSAS. The performance of LETKF is further improved, especially over the oceans, by assimilating simulated temperature observations from rawinsondes and conventional surface pressure observations instead of geopotential heights. An initial decrease of the forecast errors in the NH observed in PSAS but not in LETKF suggests that the PSAS analysis is less balanced. The observed advantage of LETKF over PSAS is due to the ability of the forty-member ensemble from LETKF to capture flow-dependent errors and thus create a good estimate of the true background uncertainty. Furthermore, localization makes LETKF highly parallel and efficient, requiring only 5 minutes per analysis in a cluster of 20 PCs with forty ensemble members.

physics.ao-ph↗

arXiv2006

Efficient Data Assimilation for Spatiotemporal Chaos: a Local Ensemble Transform Kalman Filter

Data assimilation is an iterative approach to the problem of estimating the state of a dynamical system using both current and past observations of the system together with a model for the system's time evolution. Rather than solving the problem from scratch each time new observations become available, one uses the model to ``forecast'' the current state, using a prior state estimate (which incorporates information from past data) as the initial condition, then uses current data to correct the prior forecast to a current state estimate. This Bayesian approach is most effective when the uncertainty in both the observations and in the state estimate, as it evolves over time, are accurately quantified. In this article, we describe a practical method for data assimilation in large, spatiotemporally chaotic systems. The method is a type of ``ensemble Kalman filter'', in which the state estimate and its approximate uncertainty are represented at any given time by an ensemble of system states. We discuss both the mathematical basis of this approach and its implementation; our primary emphasis is on ease of use and computational speed rather than improving accuracy over previously published approaches to ensemble Kalman filtering. We include some numerical results demonstrating the efficiency and accuracy of our implementation for assimilating real atmospheric data with the global forecast model used by the U.S. National Weather Service.

physics.data-an↗

arXiv2003

A Local Ensemble Kalman Filter for Atmospheric Data Assimilation

In this paper, we introduce a new, local formulation of the ensemble Kalman Filter approach for atmospheric data assimilation. Our scheme is based on the hypothesis that, when the Earth's surface is divided up into local regions of moderate size, vectors of the forecast uncertainties in such regions tend to lie in a subspace of much lower dimension than that of the full atmospheric state vector of such a region. Ensemble Kalman Filters, in general, assume that the analysis resulting from the data assimilation lies in the same subspace as the expected forecast error. Under our hypothesis the dimension of this subspace is low. This implies that operations only on relatively low dimensional matrices are required. Thus, the data analysis is done locally in a manner allowing massively parallel computation to be exploited. The local analyses are then used to construct global states for advancement to the next forecast time. The method, its potential advantages, properties, and implementation requirements are illustrated by numerical experiments on the Lorenz-96 model. It is found that accurate analysis can be achieved at a cost which is very modest compared to that of a full global ensemble Kalman Filter.

physics.ao-ph↗