Geoscientific Models Research Articles

Construction of an open knowledge framework for geoscientific models

AbstractGeoscientific models have rapidly developed in recent decades as effective tools to understand the past, perceive the present, and predict the future. However, with the increasing number of models available, discovering suitable ones and applying them properly in problem‐solving situations has become more challenging. Existing materials describing geoscientific models (e.g., articles, manuals, handbooks, metadata documents, and web pages) are scattered and varied in structure and content. To help model users from different disciplinary backgrounds find, access, implement, and reuse models more conveniently, we propose an open knowledge framework for geoscientific models. The knowledge framework includes three levels: a resource level for indicating where to find a model, a connection level for indicating what materials are related to a model, and an application level for indicating how the model is applied. Through such a three‐level framework, model users can collaboratively provide descriptive information for a model, link different materials to a model (e.g., data, references, and tools), and contribute experiences regarding model application in practical cases (as reusable solutions). Thus, a web‐based community can be formed to facilitate the better use of geoscientific models. This article introduces the Open Geographic Modeling and Simulation System (OpenGMS) as the implementation of this open knowledge framework. Case studies are given to showcase the effectiveness and capability of the proposed framework.

A novel Eulerian model based on central moments to simulate age and reactivity continua interacting with mixing processes

Abstract. In geoscientific models, simulating the properties associated with particles in a continuum can serve many scientific purposes, and this has commonly been addressed using Lagrangian models. As an alternative approach, we present an Eulerian method here: diffusion–advection–reaction type partial differential equations are derived for centralized moments, which can describe the distribution of properties associated with chemicals in reaction–transport models. When the property is age, the equations for centralized moments (unlike non-central moments) do not require terms to account for aging, making this method suitable for modeling age tracers. The properties described by the distributions may also represent kinetic variables affecting reaction rates. In practical applications, continuous distributions of ages and reactivities are resolved to simulate organic matter mineralization in surficial sediments, where macrofaunal and physical mixing processes typically dominate transport. In test simulations, mixing emerged as the predominant factor shaping reactivity and age distributions. Furthermore, the applications showcase the method's aptitude for modeling continua in mixed environments while also highlighting practical considerations and challenges.

Universal differential equations for glacier ice flow modelling

Abstract. Geoscientific models are facing increasing challenges to exploit growing datasets coming from remote sensing. Universal differential equations (UDEs), aided by differentiable programming, provide a new scientific modelling paradigm enabling both complex functional inversions to potentially discover new physical laws and data assimilation from heterogeneous and sparse observations. We demonstrate an application of UDEs as a proof of concept to learn the creep component of ice flow, i.e. a nonlinear diffusivity differential equation, of a glacier evolution model. By combining a mechanistic model based on a two-dimensional shallow-ice approximation partial differential equation with an embedded neural network, i.e. a UDE, we can learn parts of an equation as nonlinear functions that then can be translated into mathematical expressions. We implemented this modelling framework as ODINN.jl, a package in the Julia programming language, providing high performance, source-to-source automatic differentiation (AD) and seamless integration with tools and global datasets from the Open Global Glacier Model in Python. We demonstrate this concept for 17 different glaciers around the world, for which we successfully recover a prescribed artificial law describing ice creep variability by solving ∼ 500 000 ordinary differential equations in parallel. Furthermore, we investigate which are the best tools in the scientific machine learning ecosystem in Julia to differentiate and optimize large nonlinear diffusivity UDEs. This study represents a proof of concept for a new modelling framework aiming at discovering empirical laws for large-scale glacier processes, such as the variability in ice creep and basal sliding for ice flow, and new hybrid surface mass balance models.

Dynamically weighted ensemble of geoscientific models via automated machine-learning-based classification

Abstract. Despite recent developments in geoscientific (e.g., physics- or data-driven) models, effectively assembling multiple models for approaching a benchmark solution remains challenging in many sub-disciplines of geoscientific fields. Here, we proposed an automated machine-learning-assisted ensemble framework (AutoML-Ens) that attempts to resolve this challenge. Details of the methodology and workflow of AutoML-Ens were provided, and a prototype model was realized with the key strategy of mapping between the probabilities derived from the machine learning classifier and the dynamic weights assigned to the candidate ensemble members. Based on the newly proposed framework, its applications for two real-world examples (i.e., mapping global soil water retention parameters and estimating remotely sensed cropland evapotranspiration) were investigated and discussed. Results showed that compared to conventional ensemble approaches, AutoML-Ens was superior across the datasets (the training, testing, and overall datasets) and environmental gradients with improved performance metrics (e.g., coefficient of determination, Kling–Gupta efficiency, and root-mean-squared error). The better performance suggested the great potential of AutoML-Ens for improving quantification and reducing uncertainty in estimates due to its two unique features, i.e., assigning dynamic weights for candidate models and taking full advantage of AutoML-assisted workflow. In addition to the representative results, we also discussed the interpretational aspects of the used framework and its possible extensions. More importantly, we emphasized the benefits of combining data-driven approaches with physics constraints for geoscientific model ensemble problems with high dimensionality in space and nonlinear behaviors in nature.

Surrogate-assisted distributed swarm optimisation for computationally expensive geoscientific models

Evolutionary algorithms provide gradient-free optimisation which is beneficial for models that have difficulty in obtaining gradients; for instance, geoscientific landscape evolution models. However, such models are at times computationally expensive and even distributed swarm-based optimisation with parallel computing struggle. We can incorporate efficient strategies such as surrogate-assisted optimisation to address the challenges; however, implementing inter-process communication for surrogate-based model training is difficult. In this paper, we implement surrogate-based estimation of fitness evaluation in distributed swarm optimisation over a parallel computing architecture. We first test the framework on a set of benchmark optimisation problems and then apply to a geoscientifc model that features landscape evolution model. Our results demonstrate very promising results for benchmark functions and the Badlands landscape evolution model. We obtain a reduction in computationally time while retaining optimisation solution accuracy through the use of surrogates in a parallel computing environment. The major contribution of the paper is in the application of surrogate-based optimisation for geoscientific models which can in the future help in better understanding of paleoclimate and geomorphology.

A methodological framework for improving the performance of data-driven models: a case study for daily runoff prediction in the Maumee domain, USA

Abstract. Geoscientific models are simplified representations of complex earth and environmental systems (EESs). Compared with physics-based numerical models, data-driven modeling has gained popularity due mainly to data proliferation in EESs and the ability to perform prediction without requiring explicit mathematical representation of complex biophysical processes. However, because of the black-box nature of data-driven models, their performance cannot be guaranteed. To address this issue, we developed a generalizable framework for improving the efficiency and effectiveness of model training and the reduction of model overfitting. This framework consists of two parts: hyperparameter selection based on Sobol global sensitivity analysis and hyperparameter tuning using a Bayesian optimization approach. We demonstrated the framework efficacy through a case study of daily edge-of-field (EOF) runoff predictions by a tree-based data-driven model using the extreme gradient boosting (XGBoost) algorithm in the Maumee domain, USA. This framework contributes towards improving the performance of a variety of data-driven models and can thus help promote their applications in EESs.

The Intelligent Prospector v1.0: geoscientific model development and prediction by sequential data acquisition planning with application to mineral exploration

Abstract. Geoscientific models are based on geoscientific data; hence, building better models, in the sense of attaining better predictions, often means acquiring additional data. In decision theory, questions of what additional data are expected to best improve predictions and decisions is within the realm of value of information and Bayesian optimal survey design. However, these approaches often evaluate the optimality of one additional data acquisition campaign at a time. In many real settings, certainly in those related to the exploration of Earth resources, a large sequence of data acquisition campaigns possibly needs to be planned. Geoscientific data acquisition can be expensive and time-consuming, requiring effective measurement campaign planning to optimally allocate resources. Each measurement in a data acquisition sequence has the potential to inform where best to take the following measurements; however, directly optimizing a closed-loop measurement sequence requires solving an intractable combinatoric search problem. In this work, we formulate the sequential geoscientific data acquisition problem as a partially observable Markov decision process (POMDP). We then present methodologies to solve the sequential problem using Monte Carlo planning methods. We demonstrate the effectiveness of the proposed approach on a simple 2D synthetic exploration problem. Tests show that the proposed sequential approach is significantly more effective at reducing uncertainty than conventional methods. Although our approach is discussed in the context of mineral resource exploration, it likely has bearing on other types of geoscientific model questions.

An iterative process for efficient optimisation of parameters in geoscientific models: a demonstration using the Parallel Ice Sheet Model (PISM) version 0.7.3

Abstract. Physical processes within geoscientific models are sometimes described by simplified schemes known as parameterisations. The values of the parameters within these schemes can be poorly constrained by theory or observation. Uncertainty in the parameter values translates into uncertainty in the outputs of the models. Proper quantification of the uncertainty in model predictions therefore requires a systematic approach for sampling parameter space. In this study, we develop a simple and efficient approach to identify regions of multi-dimensional parameter space that are consistent with observations. Using the Parallel Ice Sheet Model to simulate the present-day state of the Antarctic Ice Sheet, we find that co-dependencies between parameters preclude any simple identification of a single optimal set of parameter values. Approaches such as large ensemble modelling are therefore required in order to generate model predictions that incorporate proper quantification of the uncertainty arising from the parameterisation of physical processes.

Prediction of Climate Change using SVM and Naïve Bayes Machine Learning Algorithms

Various reasons are there in failures of Intergovernmental Panel on Climate Change (IPCC) simulation model for prediction of climate change. For the better understanding of IPCC model’s failures by researchers, an improvement is qualitative and quantitative analysis is required and to be implemented. We come across a continuous crashes in simulation of Parallel Ocean Program (POP2) component of the Community Climate System Model (CCSM4), while measuring the impact of ocean model parameter uncertainties on weather simulations, during the period of uncertainty quantification (UQ) ensemble. This manuscript analyse the different machine learning algorithms, such as, Random forest, Linear Regression, k-means and naïve-bayes algorithms. From machine learning, a quality classifier called support vector machine (SVM) classification is used to predict and quantify the failures probability as a function of the values of POP2 parameters. Apart from quantification and prediction, this method performs a better understanding in simulation crashes in other complex geo-scientific models.

Azimuthal averaging–reconstruction filtering techniques for finite-difference general circulation models in spherical geometry

Abstract. When solving hydrodynamic equations in spherical or cylindrical geometry using explicit finite-difference schemes, a major difficulty is that the time step is greatly restricted by the clustering of azimuthal cells near the pole due to the Courant–Friedrichs–Lewy condition. This paper adapts the azimuthal averaging–reconstruction (ring average) technique to finite-difference schemes in order to mitigate the time step constraint in spherical and cylindrical coordinates. The finite-difference ring average technique averages physical quantities based on an effective grid and then reconstructs the solution back to the original grid in a piecewise, monotonic way. The algorithm is implemented in a community upper-atmospheric model, the Thermosphere–Ionosphere Electrodynamics General Circulation Model (TIEGCM), with a horizontal resolution up to 0.625∘×0.625∘ in geographic longitude–latitude coordinates, which enables the capability of resolving critical mesoscale structures within the TIEGCM. Numerical experiments have shown that the ring average technique introduces minimal artifacts in the polar region of general circulation model (GCM) solutions, which is a significant improvement compared to commonly used low-pass filtering techniques such as the fast Fourier transform method. Since the finite-difference adaption of the ring average technique is a post-solver type of algorithm, which requires no changes to the original computational grid and numerical algorithms, it has also been implemented in much more complicated models with extended physical–chemical modules such as the Coupled Magnetosphere–Ionosphere–Thermosphere (CMIT) model and the Whole Atmosphere Community Climate Model with thermosphere and ionosphere eXtension (WACCM-X). The implementation of ring average techniques in both models enables CMIT and WACCM-X to perform global simulations with a much higher resolution than that used in the community versions. The new technique is not only a significant improvement in space weather modeling capability, but it can also be adapted to more general finite-difference solvers for hyperbolic equations in spherical and polar geometries. Highlights. The ring average technique is adapted to solve the issue of clustered grid cells in polar and spherical coordinates with a finite-difference method. The ring average technique is applied to develop a 0.625∘×0.625∘ high-resolution TIEGCM and more complicated geoscientific models with polar and spherical coordinates as well as finite-difference numerical schemes. The high-resolution TIEGCM shows good capability in resolving mesoscale structures in the ionosphere–thermosphere (I–T) system.

Surrogate-assisted Bayesian inversion for landscape and basin evolution models

Abstract. The complex and computationally expensive nature of landscape evolution models poses significant challenges to the inference and optimization of unknown model parameters. Bayesian inference provides a methodology for estimation and uncertainty quantification of unknown model parameters. In our previous work, we developed parallel tempering Bayeslands as a framework for parameter estimation and uncertainty quantification for the Badlands landscape evolution model. Parallel tempering Bayeslands features high-performance computing that can feature dozens of processing cores running in parallel to enhance computational efficiency. Nevertheless, the procedure remains computationally challenging since thousands of samples need to be drawn and evaluated. In large-scale landscape evolution problems, a single model evaluation can take from several minutes to hours and in some instances, even days or weeks. Surrogate-assisted optimization has been used for several computationally expensive engineering problems which motivate its use in optimization and inference of complex geoscientific models. The use of surrogate models can speed up parallel tempering Bayeslands by developing computationally inexpensive models to mimic expensive ones. In this paper, we apply surrogate-assisted parallel tempering where the surrogate mimics a landscape evolution model by estimating the likelihood function from the model. We employ a neural-network-based surrogate model that learns from the history of samples generated. The entire framework is developed in a parallel computing infrastructure to take advantage of parallelism. The results show that the proposed methodology is effective in lowering the computational cost significantly while retaining the quality of model predictions.

Improving AI System Awareness of Geoscience Knowledge: Symbiotic Integration of Physical Approaches and Deep Learning

AbstractModeling dynamic geophysical phenomena is at the core of Earth and environmental studies. The geoscientific community relying mainly on physical representations may want to consider much deeper adoption of artificial intelligence (AI) instruments in the context of AI's global success and emergence of big Earth data. A new perspective of using hybrid physics‐AI approaches is a grand vision, but actualizing such approaches remains an open question in geoscience. This study develops a general approach to improving AI geoscientific awareness, wherein physical approaches such as temporal dynamic geoscientific models are included as special recurrent neural layers in a deep learning architecture. The illustrative case of runoff modeling across the conterminous United States demonstrates that the physics‐aware DL model has enhanced prediction accuracy, robust transferability, and good intelligence for inferring unobserved processes. This study represents a firm step toward realizing the vision of tackling Earth system challenges by physics‐AI integration.

Validator – a Web-Based Interactive Tool for Validation of Ocean Models at Oceanographic Stations

Numerical ocean models, like other geoscientific models, are a strongly simplified representation of real oceans. They are used as tools to answer research questions about the real-world systems. Therefore, their thorough validation is essential to ensure that the conclusions drawn from the model experiment are valid in reality. We demonstrate a software which allows an interactive model validation through a web interface based on the R Shiny framework. At pre-defined stations, different kinds of plots can be rendered within a few seconds, according to the user’s choice, allowing a live validation of different model parameters even in model simulations which are still running. This makes it different from validation approaches which generate a pre-defined set of plots after the calculations have finished and make it particularly useful for model tuning purposes. Observation data can be read in from text files or can be extracted from a database. Once set up, the validation tool requires no technical skills to use. It can be used for single- or multi-model validation and allows saving the generated plots as high-resolution images suitable for use in scientific publications. A Linux operating system is required for the Validator app, but via a virtual machine, the software can run on Windows or MacOS hosts as well. A Dockerfile is supplied which allows to test the software with example data without installation. Funding statement: This software was developed at the Leibniz Institute for Baltic Sea Research Warnemunde (IOW) on institutional funding.

A run control framework to streamline profiling, porting, and tuning simulation runs and provenance tracking of geoscientific applications

Abstract. Geoscientific modeling is constantly evolving, with next-generation geoscientific models and applications placing large demands on high-performance computing (HPC) resources. These demands are being met by new developments in HPC architectures, software libraries, and infrastructures. In addition to the challenge of new massively parallel HPC systems, reproducibility of simulation and analysis results is of great concern. This is due to the fact that next-generation geoscientific models are based on complex model implementations and profiling, modeling, and data processing workflows. Thus, in order to reduce both the duration and the cost of code migration, aid in the development of new models or model components, while ensuring reproducibility and sustainability over the complete data life cycle, an automated approach to profiling, porting, and provenance tracking is necessary. We propose a run control framework (RCF) integrated with a workflow engine as a best practice approach to automate profiling, porting, provenance tracking, and simulation runs. Our RCF encompasses all stages of the modeling chain: (1) preprocess input, (2) compilation of code (including code instrumentation with performance analysis tools), (3) simulation run, and (4) postprocessing and analysis, to address these issues. Within this RCF, the workflow engine is used to create and manage benchmark or simulation parameter combinations and performs the documentation and data organization for reproducibility. In this study, we outline this approach and highlight the subsequent developments scheduled for implementation born out of the extensive profiling of ParFlow. We show that in using our run control framework, testing, benchmarking, profiling, and running models is less time consuming and more robust than running geoscientific applications in an ad hoc fashion, resulting in more efficient use of HPC resources, more strategic code development, and enhanced data integrity and reproducibility.

GPU-accelerated atmospheric chemical kinetics in the ECHAM/MESSy (EMAC) Earth system model (version 2.52)

Abstract. This paper presents an application of GPU accelerators in Earth system modeling. We focus on atmospheric chemical kinetics, one of the most computationally intensive tasks in climate–chemistry model simulations. We developed a software package that automatically generates CUDA kernels to numerically integrate atmospheric chemical kinetics in the global climate model ECHAM/MESSy Atmospheric Chemistry (EMAC), used to study climate change and air quality scenarios. A source-to-source compiler outputs a CUDA-compatible kernel by parsing the FORTRAN code generated by the Kinetic PreProcessor (KPP) general analysis tool. All Rosenbrock methods that are available in the KPP numerical library are supported.Performance evaluation, using Fermi and Pascal CUDA-enabled GPU accelerators, shows achieved speed-ups of 4. 5 × and 20. 4 × , respectively, of the kernel execution time. A node-to-node real-world production performance comparison shows a 1. 75 × speed-up over the non-accelerated application using the KPP three-stage Rosenbrock solver. We provide a detailed description of the code optimizations used to improve the performance including memory optimizations, control code simplification, and reduction of idle time. The accuracy and correctness of the accelerated implementation are evaluated by comparing to the CPU-only code of the application. The median relative difference is found to be less than 0.000000001 % when comparing the output of the accelerated kernel the CPU-only code.The approach followed, including the computational workload division, and the developed GPU solver code can potentially be used as the basis for hardware acceleration of numerous geoscientific models that rely on KPP for atmospheric chemical kinetics applications.

The introspective may achieve more: Enhancing existing Geoscientific models with native-language emulated structural reflection

Geoscientific models manage myriad and increasingly complex data structures as trans-disciplinary models are integrated. They often incur significant redundancy with cross-cutting tasks. Reflection, the ability of a program to inspect and modify its structure and behavior at runtime, is known as a powerful tool to improve code reusability, abstraction, and separation of concerns. Reflection is rarely adopted in high-performance Geoscientific models, especially with Fortran, where it was previously deemed implausible. Practical constraints of language and legacy often limit us to feather-weight, native-language solutions. We demonstrate the usefulness of a structural-reflection-emulating, dynamically-linked metaObjects, gd. We show real-world examples including data structure self-assembly, effortless input/output (IO) and upgrade to parallel I/O, recursive actions and batch operations. We share gd and a derived module that reproduces MATLAB-like structure in Fortran and C++. We suggest that both a gd representation and a Fortran-native representation are maintained to access the data, each for separate purposes. Embracing emulated reflection allows generically-written codes that are highly re-usable across projects.

Determining the relationship of thermal conductivity and compressional wave velocity of common rock types as a basis for reservoir characterization

A comprehensive dataset detailing thermal conductivity and acoustic (compressional) wave velocity of 1430 oven-dry rock samples from clastic sedimentary (sandstone, arkose, greywacke), carbonatic (limestone, marl, dolomite, marble, coquina), plutonic (gabbro, gabbrodiorite, diorite, granodiorite, granite) and volcanic (basalt, andesite, rhyolite) rock types is presented. Correlation of thermal conductivity, compressional wave velocity and porosity are discussed in detail for each tested rock type. The study confirms that thermal conductivity of dry rocks can be predicted from acoustic velocity for porous rock types such as volcanites and sandstones, while non- and low-porous rocks show no to minor trends. With a prediction accuracy ±0.5Wm−1K−1 and a confidence of >80% for sediments and mafic volcanites the calculated data is far more comprehensive than data collected from literature, and is likely accurate enough for most first exploration approaches or geoscientific models before detailed site-scale investigation or modelling is conducted.To investigate the effect of water saturation on thermal conductivity and compressional wave velocity 118 sedimentary samples (arkose and fine-, medium- and coarse sandstones) were saturated in de-aired water and the heat conduction and acoustic velocity were remeasured. The obtained data shows that both thermal conductivity and compressional wave velocity of saturated samples markedly increase in contrast to dry samples. The extent of the thermal conductivity and compressional wave velocity gain is mainly controlled by porosity. Thermal conductivity of saturated samples increases twice as much for higher porous samples than for low porous fine and medium sandstones. In contrast, the gain of compressional wave velocity of saturated sandstones decreases with increasing porosity.

Bit Grooming: statistically accurate precision-preserving quantization with compression, evaluated in the netCDF Operators (NCO, v4.4.8+)

Abstract. Geoscientific models and measurements generate false precision (scientifically meaningless data bits) that wastes storage space. False precision can mislead (by implying noise is signal) and be scientifically pointless, especially for measurements. By contrast, lossy compression can be both economical (save space) and heuristic (clarify data limitations) without compromising the scientific integrity of data. Data quantization can thus be appropriate regardless of whether space limitations are a concern. We introduce, implement, and characterize a new lossy compression scheme suitable for IEEE floating-point data. Our new Bit Grooming algorithm alternately shaves (to zero) and sets (to one) the least significant bits of consecutive values to preserve a desired precision. This is a symmetric, two-sided variant of an algorithm sometimes called Bit Shaving that quantizes values solely by zeroing bits. Our variation eliminates the artificial low bias produced by always zeroing bits, and makes Bit Grooming more suitable for arrays and multi-dimensional fields whose mean statistics are important. Bit Grooming relies on standard lossless compression to achieve the actual reduction in storage space, so we tested Bit Grooming by applying the DEFLATE compression algorithm to bit-groomed and full-precision climate data stored in netCDF3, netCDF4, HDF4, and HDF5 formats. Bit Grooming reduces the storage space required by initially uncompressed and compressed climate data by 25–80 and 5–65 %, respectively, for single-precision values (the most common case for climate data) quantized to retain 1–5 decimal digits of precision. The potential reduction is greater for double-precision datasets. When used aggressively (i.e., preserving only 1–2 digits), Bit Grooming produces storage reductions comparable to other quantization techniques such as Linear Packing. Unlike Linear Packing, whose guaranteed precision rapidly degrades within the relatively narrow dynamic range of values that it can compress, Bit Grooming guarantees the specified precision throughout the full floating-point range. Data quantization by Bit Grooming is irreversible (i.e., lossy) yet transparent, meaning that no extra processing is required by data users/readers. Hence Bit Grooming can easily reduce data storage volume without sacrificing scientific precision or imposing extra burdens on users.

Correction of approximation errors with Random Forests applied to modelling of cloud droplet formation

Abstract. In atmospheric models, due to their computational time or resource limitations, physical processes have to be simulated using reduced (i.e. simplified) models. The use of a reduced model, however, induces errors to the simulation results. These errors are referred to as approximation errors. In this paper, we propose a novel approach to correct these approximation errors. We model the approximation error as an additive noise process in the simulation model and employ the Random Forest (RF) regression algorithm for constructing a computationally low cost predictor for the approximation error. In this way, the overall simulation problem is decomposed into two separate and computationally efficient simulation problems: solution of the reduced model and prediction of the approximation error realisation. The approach is tested for handling approximation errors due to a reduced coarse sectional representation of aerosol size distribution in a cloud droplet formation calculation as well as for compensating the uncertainty caused by the aerosol activation parameterization itself. The results show a significant improvement in the accuracy of the simulation compared to the conventional simulation with a reduced model. The proposed approach is rather general and extension of it to different parameterizations or reduced process models that are coupled to geoscientific models is a straightforward task. Another major benefit of this method is that it can be applied to physical processes that are dependent on a large number of variables making them difficult to be parameterized by traditional methods.

Failure analysis of parameter-induced simulation crashes in climate models

Abstract. Simulations using IPCC (Intergovernmental Panel on Climate Change)-class climate models are subject to fail or crash for a variety of reasons. Quantitative analysis of the failures can yield useful insights to better understand and improve the models. During the course of uncertainty quantification (UQ) ensemble simulations to assess the effects of ocean model parameter uncertainties on climate simulations, we experienced a series of simulation crashes within the Parallel Ocean Program (POP2) component of the Community Climate System Model (CCSM4). About 8.5% of our CCSM4 simulations failed for numerical reasons at combinations of POP2 parameter values. We applied support vector machine (SVM) classification from machine learning to quantify and predict the probability of failure as a function of the values of 18 POP2 parameters. A committee of SVM classifiers readily predicted model failures in an independent validation ensemble, as assessed by the area under the receiver operating characteristic (ROC) curve metric (AUC > 0.96). The causes of the simulation failures were determined through a global sensitivity analysis. Combinations of 8 parameters related to ocean mixing and viscosity from three different POP2 parameterizations were the major sources of the failures. This information can be used to improve POP2 and CCSM4 by incorporating correlations across the relevant parameters. Our method can also be used to quantify, predict, and understand simulation crashes in other complex geoscientific models.