Italian coastal wave climate dataset — bias-adjusted reanalysis, hindcasts and EURO-CORDEX projections

Dataset Overview

This dataset provides bias-adjusted wave data for 15 buoy locations along the Italian coastline, together with bias-corrected historical and future wave projections. The dataset includes corrections applied to multiple reanalysis and model sources—ERA5, CMEMS, MeteOcean-UniGe 10 km regular grid, and MeteOcean-UniGe unstructured grid.

The MeteOcean-UniGe 10km regular and unstructured grid are produced by the MeteOcean research group of the University of Genova (www.meteocean.science), providing wave information with different resolution for the Mediterranean Sea using the third-generation wave model Wavewatch III. 

CMEMS dataset: This database contains raw reanalysis data provided by E.U. Copernicus Marine Service Information; https://doi.org/10.48670/mds-00376.
This database contains modified reanalysis data generated using E.U. Copernicus Marine Service Information; https://doi.org/10.48670/mds-00376.

ERA5: This database contains raw reanalysis data provided by Copernicus Climate Change Service; https://doi.org/10.24381/cds.adbb2d47. 
This database contains modified Copernicus Climate Change Service information. Neither the European Commission nor ECMWF is responsible for any use that may be made of the Copernicus information or data it contains.

The bias adjustment is performed, for the significant wave height (Hs), using the DM-seasonal method (Distribution Mapping by seasons) with a mixed distribution approach, employing an extreme-value-tail distribution to improve the representation of storm extremes. The corrected MeteOcean-UniGe 10 km dataset is then used as a reference to correct historical (baseline) and future (RCP8.5) simulations from 21 EURO-CORDEX RCM–GCM combinations.

Additionally, the dataset includes results of a Bivariate Dynamical Optimal Transport Correction (dOTC) applied jointly to significant wave height (Hs) and peak period (Tp) for the MeteOcean-UniGe 10 km grid, together with the corresponding bias-corrected projections for all 21 climate models.

This dataset supports coastal risk-impact studies, wave climate assessment, and long-term hazard projections along the Italian coasts.

Data Structure

The dataset is provided as a collection of ZIP archives, each containing text files (.txt) corresponding to the processed time series.

All ZIP files follow the structure:

Where:

<data>

• raw

• bias-adj-Hs-DMseasonal

• bias-adj-Hs_Tp-dOTC

<model>

• CMEMS

• ERA5

• hindcast_MeteOcean-UniGe-ww3-10km

• hindcast_MeteOcean-UniGe-ww3-unst

• projections_MeteOcean-UniGe-ww3-10km

Raw data ZIP archives

• raw_ERA5.zip

• raw_CMEMS.zip

• raw_hindcast_MeteOcean-UniGe-ww3-10km.zip

• raw_hindcast_MeteOcean-UniGe-ww3-unst.zip

• raw_projections_MeteOcean-UniGe-ww3-10km.zip

Each hindcast ZIP contains 15 .txt files, one per buoy location. The projections ZIP contains two folders:

Each folder contains 315 .txt files (= 15 locations × 21 RCM–GCM models).

DM-seasonal bias-adjusted (Hs) ZIP archives

• bias-adj-Hs-DMseasonal_CMEMS.zip

• bias-adj-Hs-DMseasonal_ERA5.zip

• bias-adj-Hs-DMseasonal_hindcast_MeteOcean-UniGe-ww3-10km.zip

• bias-adj-Hs-DMseasonal_hindcast_MeteOcean-UniGe-ww3-unst.zip

• bias-adj-Hs-DMseasonal_projections_MeteOcean-UniGe-ww3-10km.zip

Hindcast ZIPs: 15 .txt files. Projections ZIP: two folders (baseline, rcp85), each with 315 .txt files

dOTC bivariate Hs–Tp ZIP archives

• bias-adj-Hs_Tp-dOTC_hindcast_MeteOcean-UniGe-ww3-10km.zip

• bias-adj-Hs_Tp-dOTC_projections_MeteOcean-UniGe-ww3-10km.zip

Hindcast: 15 .txt files. Projections: two folders, each with 315 .txt files

File format description

Each .txt file includes:

• a header listing variables, units, and metadata

• time series of:

  • timestamps

  • Hs (DM-seasonal and dOTC)

  • Tp (dOTC only)

  • additional variables for raw data (e.g., mean wave period and wave direction)

Dataset version v2 changes

In this new version of the dataset, we corrected an error in the reference locations used for several raw and bias-corrected products. The original configuration relied on incorrect reference points, which affected both the extracted raw time series and the subsequent bias-correction procedures. We therefore re-extracted all relevant datasets using the correct reference locations (Catania, Ponza, and Palermo, depending on the product) and reran the entire bias-correction workflow. This update applies consistently across RAW datasets (ERA5, CMEMS, WW3-unst, WW3-10km, and projections) as well as all BC-DTOC and BC-EQM products, ensuring spatial consistency and improved reliability of the corrected outputs.