Addressing global challenges such as coastal hazards and climate change requires innovative tools capable of analyzing complex environmental drivers, including waves, storm surges, and cyclones, across varying scales. These tools are vital for predicting floods, assessing risks, and planning adaptive responses. BlueMath-Hub has been developed as a global collaborative initiative to provide accessible, customizable solutions for both researchers and practitioners. It aims to simplify the use of advanced statistical and numerical models, fostering creative and scalable approaches in coastal science and engineering. BlueMath, the core of this platform, is an open-source repository of Python tools accessible via a cloud-based Jupyter Hub environment. It integrates statistical methods and numerical model wrappers within a modular framework. The system includes: (a) BlueMath-Toolkit, providing tools for data mining, interpolation, and model integration; (b) BlueMath-Statistical Downscaling, focusing on extreme events and generalized models; (c) BlueMath-Hybrid Downscaling, combining statistical and numerical approaches for optimized solutions; and (d) BlueMath-Climate Services, supporting integrated applications such as compound flooding assessments. BlueMath is continuously evolving, with its tools already applied in research, publications, and training. By lowering barriers to entry and enabling collaborative workflows, BlueMath-Hub supports the development of innovative solutions to mitigate the impacts of a changing climate.

Wave climatologies from historical and projected simulations of the ACCESS1.0, MIROC5 and CNRM-CM5 Global Circulation Models (GCM) were sourced from the Coordinated Ocean Wave Climate Project (COWCLIP) and downscaled using the SWAN wave model. Biases between GCM's historical simulations and a regional hindcast were assessed, and the two best-performing models (ACCESS1.0, MIROC5) had their projections analysed. A general increase in wave height and period was observed along the south/west, together with a decrease in (Formula presented.) along the north/east coasts. The projected near-term (NEA21C) period shows mostly a (Formula presented.) increase, whilst for the long-term (END21C) period, increased and decreased (Formula presented.) are present. The areas of statistically significant changes are larger in the END21C than in the NEA21C period. The wave direction change is counter-clockwise along the west and clockwise along the east coasts. This study is a first assessment of historical and projected GCM-forced waves along New Zealand and the database we generated can be of great value for renewable energy research, risk assessment and the mitigation of future coastal hazards.

A high resolution partitioned wave hindcast of New Zealand waters is presented together with validation results against 9 buoy deployments that are representative of the local wave climate. An analysis of integrated and partitioned mean wave parameters was conducted together with a study of the correlation between wave height anomalies and atmospheric indices. A directional spatial analysis was also performed to identify and quantify wave systems that are not detectable by in situ buoys or hindcast integrated parameters. The presented study outlines a framework of how partitions can be analyzed in order to improve our understanding of the local wave climate in terms of wind-sea and swell waves.

Beaches around the world continuously adjust to daily and seasonal changes in wave and tide conditions, which are themselves changing over longer time-scales. Different approaches to predict multi-year shoreline evolution have been implemented; however, robust and reliable predictions of shoreline evolution are still problematic even in short-term scenarios (shorter than decadal). Here we show results of a modelling competition, where 19 numerical models (a mix of established shoreline models and machine learning techniques) were tested using data collected for Tairua beach, New Zealand with 18 years of daily averaged alongshore shoreline position and beach rotation (orientation) data obtained from a camera system. In general, traditional shoreline models and machine learning techniques were able to reproduce shoreline changes during the calibration period (1999–2014) for normal conditions but some of the model struggled to predict extreme and fast oscillations. During the forecast period (unseen data, 2014–2017), both approaches showed a decrease in models’ capability to predict the shoreline position. This was more evident for some of the machine learning algorithms. A model ensemble performed better than individual models and enables assessment of uncertainties in model architecture. Research-coordinated approaches (e.g., modelling competitions) can fuel advances in predictive capabilities and provide a forum for the discussion about the advantages/disadvantages of available models.

Ocean-basin-scale climate variability produces shifts in wave climates and water levels affecting the coastlines of the basin. Here we present a hybrid shoreline change—foredune erosion model (A COupled CrOss-shOre, loNg-shorE, and foreDune evolution model, COCOONED) intended to inform coastal planning and adaptation. COCOONED accounts for coupled longshore and cross-shore processes at different timescales, including sequencing and clustering of storm events, seasonal, interannual, and decadal oscillations by incorporating the effects of integrated varying wave action and water levels for coastal hazard assessment. COCOONED is able to adapt shoreline change rates in response to interactions between longshore transport, cross-shore transport, water level variations, and foredune erosion. COCOONED allows for the spatial and temporal extension of survey data using global data sets of waves and water levels for assessing the behavior of the shoreline at multiple time and spatial scales. As a case study, we train the model in the period 2004–2014 (11 years) with seasonal topographic beach profile surveys from the North Beach Sub-cell (NBSC) of the Columbia River Littoral Cell (Washington, USA). We explore the shoreline response and foredune erosion along 40 km of beach at several timescales during the period 1979–2014 (35 years), revealing an accretional trend producing reorientation of the beach, cross-shore accretional, and erosional periods through time (breathing) and alternating beach rotations that are correlated with climate indices.

Understanding marine climate variability is important for coastal planning and marine operations. It is also particularly challenging for complicated settings (e.g., islands) and data-poor regions. The aim of this work is to establish a relationship between daily synoptic atmospheric patterns, and wave and storm surge conditions around New Zealand waters, based on instrumental and reanalysis data. The daily predictor we developed is able to represent sea and swell wave conditions as well as storm surge variability over different temporal scales. However, when climate variability is analysed on a longer temporal period, based on the 20th century reanalysis, large inhomogeneities are found. This highlights the dangers related to assessing climate variability, especially in data-poor regions (such as New Zealand), where inhomogeneities could be interpreted as actual changes.

A recent 35-year endpoint shoreline change analysis revealed significant counterclockwise rotations occurring in north-central Oregon, USA, littoral cells that extend 10s of kilometers in length. While the potential for severe El Niños to contribute to littoral cell rotations at seasonal to interannual scale was previously recognized, the dynamics resulting in persistent (multidecadal) rotation were unknown, largely due to a lack of historical wave conditions extending back multiple decades and the difficulty of separating the timescales of shoreline variability in a high energy region. This study addresses this question by (1) developing a statistical downscaling framework to characterize wave conditions relevant for longshore sediment transport during data-poor decades and (2) applying a one-line shoreline change model to quantitatively assess the potential for such large embayed beaches to rotate. A climate INdex was optimized to capture variability in longshore wave power as a proxy for potential LOngshore Sediment Transport (LOST_IN), and a procedure was developed to simulate many realizations of potential wave conditions from the index. Waves were transformed dynamically with Simulating Waves Nearshore to the nearshore as inputs to a one-line model that revealed shoreline rotations of embayed beaches at multiple time and spatial scales not previously discernible from infrequent observations. Model results indicate that littoral cells respond to both interannual and multidecadal oscillations, producing comparable shoreline excursions to extreme El Niño winters. The technique quantitatively relates morphodynamic forcing to specific climate patterns and has the potential to better identify and quantify coastal variability on timescales relevant to a changing climate.

Shifts in the frequency of typical meteorological patterns in an ocean basin, over interannual to decadal time scales, cause shifts in the patterns of wave generation. Therefore, ocean basin-scale climate shifts produce shifts in the wave climates affecting the coastlines of the basin. We present a hybrid methodology for downscaling observed (or predicted) climate shifts into local nearshore wave climates and then into the associated coastline responses. A series of statistical analyses translate observed (or predicted) distributions of meteorological states into the deep water wave climate affecting a coastal region and dynamical modeling combined with statistical analyses transform the deep water wave climate into the nearshore wave climate affecting a particular coastline. Finally, dynamical modeling of coastline evolution hindcasts (or predicts) how coastline shapes respond to climate shifts. As a case study, we downscale from meteorological hindcast in the North Atlantic basin since 1870 to the responses of the shape of the coast of the Carolinas, USA. We test the hindcasts using shoreline change rates calculated from historical shorelines, because shifts in coastline shape equate to changes in the alongshore pattern of shoreline change rates from one historical period to another. Although limited by the availability of historical shorelines (and complicated by historical inlet openings), the observations are consistent with the predicted signal of ocean basin-scale climate change. The hybrid downscaling methodology, applied to the output of global climate models, can be used to help forecast future patterns of shoreline change related to future climate change scenarios.

This paper proposes a buoy independent directional calibration methodology that uses satellite altimeter data to correct the simultaneous wind-sea and all swell partitions available from wave hindcasts. The proposed technique was applied to a 20-year wave hindcast in the New Zealand region and showed promising results. The method could potentially be applied worldwide to fully assess systematic errors in wind-sea and swell partitions.

Characterization of multimodal directional wave spectra is important for many offshore and coastal applications, such as marine forecasting, coastal hazard assessment, and design of offshore wave energy farms and coastal structures. However, the multivariate and multiscale nature of wave climate variability makes this complex problem tractable using computationally expensive numerical models. So far, the skill of statistical-downscaling model-based parametric (unimodal) wave conditions is limited in large ocean basins such as the Pacific. The recent availability of long-term directional spectral data from buoys and wave hindcast models allows for development of stochastic models that include multimodal sea-state parameters. This work introduces a statistical downscaling framework based on weather types to predict multimodal wave spectra (e.g., significant wave height, mean wave period, and mean wave direction from different storm systems, including sea and swells) from large-scale atmospheric pressure fields. For each weather type, variables of interest are modeled using the categorical distribution for the sea-state type, the Generalized Extreme Value (GEV) distribution for wave height and wave period, a multivariate Gaussian copula for the interdependence between variables, and a Markov chain model for the chronology of daily weather types. We apply the model to the southern California coast, where local seas and swells from both the Northern and Southern Hemispheres contribute to the multimodal wave spectrum. This work allows attribution of particular extreme multimodal wave events to specific atmospheric conditions, expanding knowledge of time-dependent, climate-driven offshore and coastal sea-state conditions that have a significant influence on local nearshore processes, coastal morphology, and flood hazards.

Interest in understanding long-term coastal morphodynamics has recently increased as climate change impacts become perceptible and accelerated. Multiscale, behavior-oriented and process-based models, or hybrids of the two, are typically applied with deterministic approaches which require considerable computational effort. In order to reduce the computational cost of modeling large spatial and temporal scales, input reduction and morphological acceleration techniques have been developed. Here we introduce a general framework for reducing dimensionality of wave-driver inputs to morphodynamic models. The proposed framework seeks to account for dependencies with global atmospheric circulation fields and deals simultaneously with seasonality, interannual variability, long-term trends, and autocorrelation of wave height, wave period, and wave direction. The model is also able to reproduce future wave climate time series accounting for possible changes in the global climate system. An application of long-term shoreline evolution is presented by comparing the performance of the real and the simulated wave climate using a one-line model.