GUAECAST V2.0 (2019–2022) – Expanded Coverage and Multi-Grid Nesting

Broader Geographic Coverage: Released in 2019, GUAECAST V2.0 marked a major expansion of the forecast system’s scope and capabilities. The most visible change was the broadening of the model domain beyond Guaecá to cover a large portion of the São Paulo state coastline, as well as the adjacent southern coast of Rio de Janeiro. In particular, detailed high-resolution forecasts were now produced for the municipalities of Santos, Guarujá, São Sebastião, and Ubatuba (encompassing many popular surf breaks along the northern São Paulo coast). The project’s name “GUAECAST” was retained for continuity, even though its coverage grew far beyond the original Guaecá Beach focus.

Nested Grid System: To achieve this wider coverage without sacrificing nearshore resolution, V2.0 implemented a nesting hierarchy of five grids with different scales: - The outermost grid (Grid 1) covers the broad region including the entire coast of São Paulo and the southern part of Rio de Janeiro state. This coarse grid receives open-ocean boundary conditions from the NOAA/EMC global wave model (WW3), similarly to V1.0. - A second grid (Grid 2) of intermediate resolution focuses on the central and northern São Paulo coast, spanning approximately from Peruíbe in the south up to the Ubatuba (SP) – Paraty (RJ) region in the north. Grid 2 is nested inside Grid 1, receiving its boundary wave conditions from the outer grid (Figure 1). - Three innermost local grids (Grids 3–5) provide high-resolution coverage (~400 m cell spacing) for each of the key municipalities: one grid for Santos/Guarujá area, one for São Sebastião (including Guaecá Beach), and one for Ubatuba (Figure 3). These local grids are nested within the intermediate grid, allowing detailed wave transformation in bays, around islands, and near complex coastal topography.

This multi-grid approach allowed GUAECAST V2.0 to maintain fine resolution in surf zones while keeping computational demands manageable over the larger domain. The SWAN model was run on each grid, cascading information from the global model down to regional and then local scales.

Forcing and Inputs: All grids in V2.0 employed atmospheric forcing from NOAA’s GFS for winds, ensuring consistent wind input across scales. In the high-resolution local grids, water level variations were included through both astronomical tide and meteorological surge components. The surge component was derived from a separate regional hydrodynamic model (a personal project under development) to account for weather-induced sea level changes (e.g. due to pressure and wind setup). By incorporating these water level changes, V2.0 improved the modeling of wave breaking and refraction, which are sensitive to water depth, thus refining the nearshore wave predictions. The model outputs from the local grids were used to generate site-specific wave forecasts (maps, time-series, and statistics) at various surf spots along the coast (illustrated in Figures 4–6), greatly enhancing the utility of GUAECAST for local surfers and coastal managers.

Operational Enhancements: GUAECAST V2.0 ran continuously year-round from 2019 onward, moving beyond the seasonal operation of V1. A dedicated website was launched to disseminate the forecasts in real time, featuring interactive maps and charts for the covered regions. This represented a significant operational improvement, making high-quality wave forecasts readily accessible to the public and stakeholders in the São Paulo coastal community.

Intermediate Updates (V2.x): Throughout the V2.0 era, incremental updates were applied to keep the system up-to-date and accurate. The bathymetry of the model grids was periodically refreshed as new or improved data became available, since coastal bathymetric changes (from sediment transport or engineering works) can affect wave propagation. Additionally, adaptations were made to the model forcing and grid configurations in response to upgrades in the NOAA global wave model. Notably, in 2021 NOAA integrated its WaveWatch III into the unified GFS atmospheric model (GFS v16), effectively changing the resolution and forecast length of the boundary wave data. The WW3 global grid resolution was improved from ~0.5° to ~0.25°, doubling the spatial detail, and the wave forecast horizon was extended from ~7 days to 16 days. GUAECAST’s outer grid and boundary conditions were adjusted accordingly to ingest this new GFS-Wave data stream. This ensured continuity of the forecasts and took advantage of the higher-resolution wave input now available. (The coarse GUAECAST outer grid was still much finer than 0.25° in coastal areas, but the global update provided better spectral boundary information, potentially improving the accuracy of swells entering the model domain.)

In summary, GUAECAST V2.0 transformed the project from a single-beach prototype into a regional operational system. It demonstrated the feasibility of nested high-resolution wave modeling on a multi- city scale, providing consistent forecasts along a ~400 km stretch of coastline. The lessons learned in managing a multi-grid, multi-region model set the stage for the next advancement of the project.

GUAECAST V3.0 (2023–Present) – Extended Forecasts and Updates

By 2023, the GUAECAST system had matured into a robust forecasting tool. Version 3.0, released in 2023, introduced further enhancements aimed at extending the forecast range and updating data sources: -

Extended Forecast Horizon: A primary improvement in V3.0 was the extension of the forecast period from 7 days to 14 days (two weeks). This upgrade was made possible by the aforementioned improvements in NOAA’s global wave model. Since the unified GFS-Wave now provides wave predictions out to 16 days, GUAECAST was reconfigured to generate a two-week wave forecast for its domains. The ability to forecast swell events up to 14 days in advance represents a significant benefit for planning purposes, giving surfers, port authorities, and coastal managers more lead time to prepare for conditions. The model still outputs high-frequency results (hourly to 3-hourly) for short- term accuracy, but now with a longer outlook into the future.

Data and Model Updates: GUAECAST V3.0 also incorporated new datasets and visual improvements. The bathymetric grids for all model domains were refreshed using the latest available surveys and depth data, improving the fidelity of wave propagation and breaking simulations (especially in areas where coastal morphology may have changed since the previous version). On the output side, the public-facing website was enhanced with updated map visuals and charts, improving the clarity and detail of forecast information presented to users. For example, new interactive maps show the nested model domains and allow users to toggle between regional and local forecasts, and improved charting of wave height and period makes it easier to interpret the 14-day outlook.

Continuity in Coverage: Spatially, V3.0 maintains the broad coverage established in V2.0. The focus remains on the Baixada Santista (Santos/Guarujá region), São Sebastião (central north coast, including Ilhabela and nearby beaches), and Ubatuba (far north coast) areas in São Paulo, with the outer grid still covering the wider South Brazil coastal region. Thus, V3.0 did not add new geographic regions, but rather enhanced the depth and usefulness of forecasts for the existing coverage area. The multi-grid SWAN modeling approach, nested in NOAA’s global model, continues to be the backbone of the system.

Current Configuration Visualization: For reference, the current GUAECAST model configuration is illustrated in a series of figures on the project website. In particular, Figure 7 highlights the stark contrast in scale between the coarse global WW3 model grid and the high-resolution GUAECAST grids. The global model’s cells (on the order of tens of kilometers) appear as a much coarser mesh, whereas the innermost GUAECAST grid around Guaecá Beach contains cells on the order of a few hundred meters. This comparison underscores why downscaling is essential for accurate local wave forecasts: the global model alone would average over coastal complexities, while the nested high-resolution model can simulate wave refraction, diffraction, and sheltering effects caused by islands, headlands, and bathymetric variations. V3.0 fully leverages this nested approach with state-of-the-art inputs, providing a scientifically robust and operationally useful wave forecasting system for the São Paulo coast.

Comparative Summary Table of GUAECAST Versions

Figure 1: Broadest grid of GUAECAST, covering the south of the state of Rio de Janeiro and the state of São Paulo, with open boundaries indicated in red. The coarser grid is the global wave model WaveWatch 3 from NOAA/EMC.

Figure 2: Intermediate grid of GUAECAST, covering the center and north of the state of São Paulo, with open boundaries indicated in red. The coarser grid is the GUAECAST broadest grid (SP/RJ - Figure 1).

Figure 3: Local grids of GUAECAST, covering Santos/Guarujá, São Sebastião, and Ubatuba, with open boundaries indicated in red. The coarser grid is the GUAECAST intermediate grid (Figure 2).

Figure 4: GUAECAST local grid for Santos/Guarujá. Numbers indicate beach locations for forecast analysis.

Figure 5: GUAECAST local grid for São Sebastião. Numbers indicate beach locations for forecast analysis.

Figure 6: GUAECAST local grid for Ubatuba. Numbers indicate beach locations for forecast analysis.

Figure 7: Comparison between GUAECAST high-resolution grids and WW3 global grid. Top-left panel zooms into Guaecá Beach region.