Repository logo

Browsing by Author "Hurrell, James W., committee member"

Now showing 1 - 4 of 4
  • Thumbnail Image
    ItemOpen Access
    Application of neural networks to subseasonal to seasonal predictability in present and future climates
    (Colorado State University. Libraries, 2022) Mayer, Kirsten J., author; Barnes, Elizabeth A., advisor; Hurrell, James W., committee member; Maloney, Eric D., committee member; Anderson, Charles, committee member
    The Earth system is known for its lack of predictability on subseasonal to seasonal timescales (S2S; 2 weeks to a season). Yet accurate predictions on these timescales provide crucial, actionable lead times for agriculture, energy, and water management sectors. Fortunately, specific Earth system states – deemed forecasts of opportunity – can be leveraged to improve prediction skill. Our current understanding of these opportunities are rooted in our knowledge of the historical climate. Depending on societal actions, the future climate could vary drastically, and these possible futures could lead to varying changes to S2S predictability. In recent years, neural networks have been successfully applied to weather and climate prediction. With the rapid development of neural network explainability techniques, the application of neural networks now provides an opportunity to further understand our climate system as well. The research presented here demonstrates the utility of explainable neural networks for S2S prediction and predictability changes under future climates. The first study presents a novel approach for identifying forecasts of opportunity in observations using neural network confidence. It further demonstrates that neural networks can be used to gain physical insight into predictability, through neural network explainability techniques. We then employ this methodology to explore S2S predictability differences in two future scenarios: under anthropogenic climate change and stratospheric aerosol injection (SAI). In particular, we explore subseasonal predictability and forecasts of opportunity changes under anthropogenic warming compared to a historical climate in the CESM2-LE. We then investigate how future seasonal predictability may differ under SAI compared to a future without SAI deployment in the ARISE-SAI simulations. We find differences in predictability between the historical and future climates and the two future scenarios, respectively, where the largest differences in skill generally occur during forecasts of opportunity. This demonstrates that the forecast of opportunity approach, presented in the first study, is useful for identifying differences in future S2S predictability that may not have been identified if examining predictability across all predictions. Overall, these results demonstrate that neural networks are useful tools for exploring subseasonal to seasonal predictability, its sources, and future changes.
  • Thumbnail Image
    ItemOpen Access
    Links between atmospheric cloud radiative effects and tropical circulations
    (Colorado State University. Libraries, 2021) Needham, Michael R., author; Randall, David A., advisor; Hurrell, James W., committee member; Gao, Xinfeng, committee member
    Atmospheric cloud radiative effects (ACRE) quantify the radiative heating or cooling due to clouds within the atmosphere. In this study, a framework is developed with which to analyze the ways that ACRE impact large-scale circulations in humid and dry regions of the tropics. The frame-work is applied to a set of simulations from a global atmospheric model configured with uniform tropical sea surface temperatures, following the protocol of the Radiative Convective Equilibrium Model Intercomparison Project. It is found that humid regions export energy and import moisture, and that ACRE in extremely humid regions are strong enough to change the sign of the net radiation tendency. This net heating drives a feedback in which large-scale ascent moistens the troposphere by lifting latent energy from near the surface. Moisture at these higher levels then forms clouds which in turn reinforce the ACRE, continuing the process. The relevance of this feedback to the germinal study of Riehl and Malkus (1958) is discussed. Additionally, the analysis method reveals a simple relationship between cloud radiative effects and column relative humidity in the idealized model. The same relationship is also observed in cloud radiative effects calculated from satellite observations. This suggests a simple way to estimate the cloud radiative effect at the top of the atmosphere. The estimated cloud radiative effect may be useful in estimating the ACRE, which is harder to infer from measurements using previous methods. The estimation shows some skill at estimating the cloud radiative effect in humid regions across the tropics on time scales of one month or longer. The method is found to be extremely effective at estimating observed cloud radiative effects in the equatorial west Pacific. Weaknesses of the estimation method in relation to marine stratus clouds are discussed.
  • Thumbnail Image
    ItemOpen Access
    Towards using neural networks for geoscientific discovery
    (Colorado State University. Libraries, 2020) Toms, Benjamin A., author; Barnes, Elizabeth A., advisor; Ebert-Uphoff, Imme, committee member; Hurrell, James W., committee member; Thompson, David W. J., committee member
    How can we use computational methods to extract physically meaningful patterns from geoscientific data? This question has been asked in some form for decades within the geoscientific community, with many landmark discoveries resulting from the novel application of computational methods to a geoscientific dataset. For example, the Madden-Julian Oscillation was discovered through Fourier transforms of tropical time-series, while the defining structures of the Northern Hemispheric annular modes were first captured using principal component analysis. These discoveries rooted in computational methods have since driven decades of geoscientific research and innovation, and are only two of among many similar examples. It is therefore clear that computational science and geoscience are inextricably intertwined, and so the continued advancement of both fields in tandem is beneficial to future geoscientific discovery. Many methods exist to discover patterns within geoscientific data, although each is limited by its own set of assumptions. The most common assumption is that of linearity, which oftentimes conflicts with our understanding that the earth system can be both dynamically and statistically nonlinear. However, a recently popularized subset of methods within the computer science community known as neural networks can identify nonlinear patterns and are therefore potentially powerful tools for geoscientific discovery. Neural networks learn how to map one dataset to another using a combination of nonlinear relationships, and are generalizable to a broad range of tasks including forecasting and identifying patterns within images. Regardless of the application, a common limitation of neural networks has been the difficulty to understand how and why they make their decisions. Therefore, while they have been used in geoscience for more than two decades, they have mostly been applied when accuracy is valued more than understanding, such as for making forecasts. Within this dissertation, we first propose a framework for how neural networks can be used for geoscientific discovery by applying recently invented methods from the computer science community. We focus on methods that explain which aspects of the input dataset are useful for the neural network when making connections to the output dataset. This framework enables physical interpretations of how and why neural networks make decisions, since the geoscientist that designs the neural network is likely familiar with the physical meaning of each input. In the first study of the dissertation, we outline the framework and apply it to two simple tasks to ensure the neural network interpretations abide by our current understanding of the earth system. The interpretable neural networks successfully identify the pattern of the El Niño Southern Oscillation and oceanic patterns that lend seasonal predictability, which lends confidence that the framework is reliable. In the second study, we then further test the methods by applying them to a more spatially and temporally complex oscillation called the Madden-Julian Oscillation (MJO). The interpretable neural networks correctly identify the known spatial structures and seasonality of the MJO, and also suggest that the MJO is nonlinear and expresses its nonlinearity through the uniqueness of each event. The final study assesses whether the proposed framework can be used to identify predictable patterns of earth-system variability within climate models through its application to decadal predictability. We find that the interpretable neural networks identify known modes of oceanic decadal variability that contribute to predictability of continental surface temperatures. The interpretations can also be used to identify distinct regimes of predictability, wherein spatially and temporally unique oceanic modes contribute predictability for the same location at different times. From a broader perspective, these studies suggest that neural networks are a viable tool for geoscientific discovery and are particularly useful given their ability to capture nonlinear, time-evolving patterns. It is likely that new neural network algorithms and methods for their interpretation will continue to be developed by the computer science community, and so this research provides a guideline for how such methods can be gainfully applied within the geosciences.
  • Thumbnail Image
    ItemOpen Access
    When is the unpredictable (slightly more) predictable? An assessment of opportunities for skillful decadal climate prediction using explainable neural networks
    (Colorado State University. Libraries, 2023) Gordon, Emily M., author; Barnes, Elizabeth A., advisor; Hurrell, James W., committee member; Rugenstein, Maria, committee member; Anderson, Charles, committee member
    Predicting climate variability on decadal (2-10 year) timescales can have huge implications for society because it can provide better estimates of both global trends as well as regional climate variability for crucial, actionable lead times. The key to skillful decadal prediction is understanding and predicting oceanic variability. However, predictable signals in the ocean can be masked by the inherent noise in the system, and therefore, skillful prediction on decadal timescales is challenging. Machine learning, with its ability to extract nonlinear signals from large sets of noisy data, has been shown a powerful tool for predicting and understanding processes across weather and climate applications. In this dissertation, I explore applications of machine learning to decadal prediction. First, I present a machine learning approach to predicting the Pacific decadal oscillation (PDO) with artificial neural networks (ANNs) within the Community Earth System Model version 2 (CESM2) pre-industrial control simulation. Predicting PDO transitions and understanding the associated mechanisms has proven a critical but challenging task in climate science. As a form of decadal variability, the PDO is associated with both large- scale climate shifts and regional climate predictability. I show that ANNs predict PDO persistence and transitions at lead times of 12 months onward. Using layer-wise relevance propagation to investigate the ANN predictions, I demonstrate that the ANNs utilize oceanic patterns that have been previously linked to predictable PDO behavior. ANNs recognize a build-up of ocean heat content in the off-equatorial western Pacific 12–27 months before a transition occurs. The ANNs also distinguish transition mechanisms between positive-to-negative sign transitions, and negative-to-positive transitions. Secondly, I demonstrate a technique for incorporating an uncertainty estimate into the prediction of a regression neural network, allowing the identification of predictable sea surface temperature (SST) anomalies on decadal timescales in the CESM2 pre-industrial control simulation. Predictability in SSTs can be masked by unpredictable variability, and one approach to extracting predictable signals is to investigate state-dependent predictability – how differences in prediction skill depend on the initial state of the system. I leverage the network's prediction of uncertainty to examine state-dependent predictability in SSTs by focusing on predictions with the lowest uncertainty. In particular, I study two regions of the global ocean–the North Atlantic and North Pacific–and find that skillful initial states identified by the neural network correspond to particular phases of low frequency variability in the North Pacific and North Atlantic oceans. Finally, I examine the potential role of predictable internal variability in a future, warmer climate by designing an interpretable neural network that can be decomposed to examine the relative contributions of external forcing and internal variability to future regional decadal SST trend predictions. I show that there is additional prediction skill to be garnered from internal variability in the CESM2 Large Ensemble in the near-term climate (2020-2050), even in a relatively high forcing future scenario. This predictability is especially apparent in the North Atlantic, North Pacific and Tropical Pacific Oceans as well as in the Southern Ocean. I further investigate how prediction skill covaries across the ocean and find three regions with distinct coherent prediction skill driven by internal variability. SST trend predictability is found to be associated with consistent patterns of interannual and decadal variability for the grid points within each region.