This work quantitatively evaluates the fidelity with which the Northern Annular Mode (NAM), Southern Annular Mode (SAM), Pacific-North American pattern (PNA), El Niño-Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO) and Atlantic Multidecadal Oscillation (AMO) and the first-order mode interactions are represented in Earth System Model (ESM) output from the CMIP6 archive. Several skill metrics are used as part of a differential credibility assessment (DCA) of both spatial and temporal characteristics of the modes across ESMs, ESM families and specific ESM realizations relative to ERA5. The spatial patterns and probability distributions are generally well represented but skill scores that measure the degree to which the frequencies of maximum variance are captured are consistently lower for most ESMs and climate modes. Substantial variability in skill scores manifests across realizations from individual ESMs for the PNA and oceanic modes. Further, the ESMs consistently overestimate the strength of the NAM-PNA first-order interaction and underestimate the NAM-AMO connection. These results suggest that the choice of ESM and ESM realizations will continue to play a critical role in determining climate projections at the global and regional scale at least in the near-term.

Wind gusts, and in particular intense gusts, are societally relevant but extremely challenging to forecast. This study systematically assesses the skill enhancement that can be achieved using artificial neural networks (ANNs) for forecasting of wind gust occurrence and magnitude. Geophysical predictors from the ERA5 reanalysis are used in conjunction with an auto-regressive term in regression and ANN models with different predictors, and varying model complexity. Models are derived and assessed for the warm (April–September) and cold (October–March) seasons for three high passenger-volume airports in the USA. Model uncertainty is assessed by deriving models for 1000 different randomly selected training (70%) and testing (30%) subsets. Gust prediction fidelity in independent test samples is critically dependent on inclusion of an autoregressive term. Gust occurrence probabilities derived using 5-layer ANNs exhibit consistently higher fidelity than those from regression models and shallower ANNs. Inclusion of the autoregressive term and increasing the number of hidden layers in ANNs from 1 to 5 also improve the model performance for gust magnitudes (lower RMSE, increased correlation, and model standard deviations that more closely approximate observed values). Deeper ANNs (e.g., 20 hidden layers) exhibit higher skill in forecasting strong (17 to 25.7 ms−1) and damaging (≥ 25.7 ms−1) wind gusts. However, such deep networks exhibit evidence of overfitting and still substantially underestimate (by 50%) the frequency of strong and damaging wind gusts at the three airports considered herein.

Wind is an important atmospheric variable that is receiving increased attention as the world seeks to shift from carbon-based fuels in order to mitigate climate change. This has resulted in increased need for more temporally and spatially continuous wind information, which is often met through the use of reanalysis data. However, limited work has been done to assess the long-term accuracy of the wind data against observations in the context of specific applications. This study focuses on the representation of daily and monthly average 10-m wind speed data in the upper Midwest by six global reanalysis datasets. The accuracy of the datasets was assessed using several measures of skill, as well as the associated wind speed distributions and long-term trends. While it was found that higher resolution and complexity in more recent generations of reanalyses produced more accurate simulations of wind in the region, important biases remained. High variability in the observed data resulted in lower correlations at the monthly time scale. As with previous research, linear trends calculated from the reanalyzed wind speeds were significantly underestimated compared to observed trends. While it is expected that future improvements in model resolution, physics, and data assimilation will further improve wind representation in reanalyses, accounting for the differences between the available datasets and their associated biases will be important for potential applications of the output, particularly wind resource assessment.

Short‐term forecasting of wind gusts, particularly those of higher intensity, is of great societal importance but is challenging due to the presence of multiple gust generation mechanisms. Wind gust observations from eight high‐passenger‐volume airports across the continental United States (CONUS) are summarized and used to develop predictive models of wind gust occurrence and magnitude. These short‐term (same hour) forecast models are built using multiple logistic and linear regression, as well as artificial neural networks (ANNs) of varying complexity. A suite of 19 upper‐air predictors drawn from the ERA5 reanalysis and an autoregressive (AR) term are used. Stepwise procedures instruct predictor selection, and resampling is used to quantify model stability. All models are developed separately for the warm (April–September) and cold (October–March) seasons. Results show that ANNs of 3–5 hidden layers (HLs) generally exhibit higher hit rates than logistic regression models and also improve skill with respect to wind gust magnitudes. However, deeper networks with more HLs increase false alarm rates in occurrence models and mean absolute error in magnitude models due to model overfitting. For model skill, inclusion of the AR term is critical while the majority of the remaining skill derives from wind speeds and lapse rates. A predictive ceiling is also clearly demonstrated, particularly for the strong and damaging gust magnitudes, which appears to be partially due to ERA5 predictor characteristics and the presence of mixed wind climates.

Climate modes play an important role in weather and climate variability over multiple spatial and temporal scales. This research assesses Earth system model (ESM) projections of the spatiotemporal characteristics of key internal climate modes (NAM, SAM, PNA, ENSO, PDO, and AMO) under high (SSP585) and low (SSP126) radiative forcing scenarios and contextualizes those projections using historical fidelity. Time series analyses are used to assess trends and mode phase characteristics are summarized for the historical period and for the end of the twenty-first century. Spatial patterns are compared to infer morphological changes. Shifts in the power spectra are used to examine changes in variability at subannual, interannual, and interdecadal scales. Changes in time-lagged correlations are used to capture the evolution of first-order interactions. While differences in historical skill are predominantly ESM dependent, changing mode characteristics in a warmer climate also exhibit variability between individual ensemble realizations. NAM, SAM, and ENSO tend to evolve toward increased prevalence of the positive phase up to 2100 across the multimodel ensemble while the PNA and PDO exhibit little trend but increasing phase intensity. AMO characteristics are shown to depend on the method used to remove the external signal. ESMs that show higher historical fidelity tend to show more modest changes in those modes under global nonstationarity. Changes in mode interactions are found to be highly ESM dependent but exhibit broadly similar behavior to historical relationships. These findings have implications for our understanding of internal variability and make clear that the choice of ESM, and even the ESM realization, matters for applications of climate projections. Significance Statement Internal modes of variability are important to understand due to their impact on local, regional, and global weather and climate patterns. Future climate changes will not only be affected by the variability arising from these modes, but the modes will themselves change in response to the changing climate. Spatial and temporal aspects of the modes are assessed from projections of future climate and related to how well they are captured in the historical climate. This yields some measure of confidence in the changes exhibited by the models. In most cases, when historically skillful models exhibit changes that are different from those produced by less skillful models, they tend to produce more modest changes. These results, as well as the variability between model outcomes, mean decisions on which ESM to use for projections of the future climate matter significantly.