---

name: ocean-data description: Download and analyze oceanographic data from Copernicus Marine Service and Argo floats using copernicusmarine, gsw, and xarray. tags:

• oceanography
• copernicus
• argo
• xarray
• earth-science version: "1.0.0" authors:
• name: "Rosetta Skills Contributors" github: "@xjtulyc" license: MIT platforms:
• claude-code
• codex
• gemini-cli
• cursor dependencies:
• copernicusmarine>=1.0
• gsw>=3.6
• xarray>=2023.6
• pandas>=2.0
• matplotlib>=3.7
• numpy>=1.24
• cartopy>=0.22 last_updated: "2026-03-17" status: stable

---

Ocean Data Analysis with Copernicus Marine Service and Argo Floats

Retrieve, process, and visualize ocean physical and biogeochemical data using the Copernicus Marine Service (CMEMS) Python client, Argo float profiles, and the TEOS-10 Gibbs SeaWater (GSW) toolbox.

---

When to Use This Skill

• You need sea surface temperature, salinity, or current fields for a specific region and time period from the Copernicus Marine Service catalog.
• You want to download individual Argo float profiles and compute derived quantities such as mixed layer depth (MLD) or potential density.
• You need to produce T-S diagrams, eddy kinetic energy (EKE) maps, or mixed-layer climatologies for research or operational purposes.
• You are building ocean-climate pipelines that combine model output with in-situ observations.

---

Background & Key Concepts

Copernicus Marine Service (CMEMS)

The Copernicus Marine Environment Monitoring Service provides free, quality-controlled ocean datasets spanning physical, biogeochemical, and sea-ice variables. Products include reanalysis, near-real-time, and forecast datasets. The copernicusmarine Python package (v1+) replaced the legacy motuclient and ftplib workflows.

Argo Float Program

Argo is a global array of ~4000 autonomous profiling floats that measure temperature and salinity from 2000 m to the surface every ~10 days. Data are freely available via GDAC servers (Ifremer / BODC) and through the copernicusmarine catalog.

TEOS-10 / GSW Toolbox

The Thermodynamic Equation of Seawater 2010 (TEOS-10) defines Absolute Salinity (SA) and Conservative Temperature (CT) as the preferred variables. The gsw Python library converts Practical Salinity and in-situ temperature to TEOS-10 variables and computes derived quantities such as potential density, buoyancy frequency, and mixed layer depth.

Mixed Layer Depth (MLD)

MLD is commonly estimated using a density threshold criterion: the depth at which potential density exceeds the surface value by 0.03 kg/m³ (de Boyer Montégut et al., 2004).

Eddy Kinetic Energy (EKE)

EKE quantifies mesoscale variability: EKE = 0.5 * (u'^2 + v'^2), where u' and v' are anomalies of eastward and northward geostrophic currents relative to a long-term mean.

---

Environment Setup

Install dependencies

pip install "copernicusmarine>=1.0" "gsw>=3.6" "xarray>=2023.6" \
    "pandas>=2.0" "matplotlib>=3.7" "numpy>=1.24" "cartopy>=0.22"

Authenticate with Copernicus Marine Service

Register for a free account at https://marine.copernicus.eu/ then store your credentials:

# Option 1 – interactive login (writes ~/.copernicusmarine/credentials)
copernicusmarine login

# Option 2 – environment variables (CI/CD friendly)
export COPERNICUSMARINE_USERNAME="<your-username>"
export COPERNICUSMARINE_PASSWORD=$(cat ~/.cmems_passwd)   # read from file, never hardcode

import os
import copernicusmarine as cm

username = os.getenv("COPERNICUSMARINE_USERNAME", "")
password = os.getenv("COPERNICUSMARINE_PASSWORD", "")

# Verify login works
cm.login(username=username, password=password, overwrite_configuration_file=True)

---

Core Workflow

Step 1 – Browse the Catalog and Download SST Data

import copernicusmarine as cm
import xarray as xr

# List available datasets matching a keyword
catalog = cm.describe(contains=["SST", "Mediterranean"])
for entry in catalog.products[:5]:
    print(entry.product_id, "-", entry.title)

# Download a subset of the CMEMS global SST L4 product
ds = cm.open_dataset(
    dataset_id="cmems_obs-sst_glo_phy_l4_my_0.25deg",
    variables=["analysed_sst", "analysis_error"],
    minimum_longitude=-20.0,
    maximum_longitude=40.0,
    minimum_latitude=25.0,
    maximum_latitude=50.0,
    start_datetime="2023-01-01T00:00:00",
    end_datetime="2023-03-31T23:59:59",
)
print(ds)
# Convert from Kelvin to Celsius
ds["sst_celsius"] = ds["analysed_sst"] - 273.15
ds["sst_celsius"].attrs["units"] = "degC"

Step 2 – Download Argo Float Profiles and Compute MLD

import gsw
import numpy as np
import xarray as xr
import copernicusmarine as cm

# Download Argo BGC profiles for the North Atlantic (float WMO 6902880)
argo = cm.open_dataset(
    dataset_id="cmems_obs-ins_glo_phy-temp-sal_nrt_argo_P1D-m",
    variables=["TEMP", "PSAL", "PRES", "LATITUDE", "LONGITUDE", "TIME"],
    minimum_longitude=-40.0,
    maximum_longitude=-20.0,
    minimum_latitude=40.0,
    maximum_latitude=60.0,
    start_datetime="2023-06-01T00:00:00",
    end_datetime="2023-08-31T23:59:59",
)

def compute_mld(temp, psal, pres, lat, threshold=0.03):
    """
    Compute mixed layer depth using a density threshold criterion.

    Parameters
    ----------
    temp  : array-like, in-situ temperature (ITS-90, °C)
    psal  : array-like, Practical Salinity (PSS-78)
    pres  : array-like, sea pressure (dbar)
    lat   : float, latitude (degrees N)
    threshold : float, density difference criterion (kg/m³), default 0.03

    Returns
    -------
    mld : float, mixed layer depth (m)
    """
    SA = gsw.SA_from_SP(psal, pres, 0.0, lat)
    CT = gsw.CT_from_t(SA, temp, pres)
    sigma0 = gsw.sigma0(SA, CT)          # potential density anomaly (kg/m³)

    # Reference density at shallowest valid level
    valid = np.isfinite(sigma0)
    if valid.sum() < 3:
        return np.nan
    ref_density = sigma0[valid][0]

    # Find first depth where density exceeds reference + threshold
    exceeds = np.where((sigma0 - ref_density) > threshold)[0]
    if len(exceeds) == 0:
        return float(pres[valid][-1])    # whole profile is mixed
    return float(pres[exceeds[0]])

# Apply to each profile
mld_values = []
for i in range(min(50, argo.dims.get("N_PROF", 0))):
    profile = argo.isel(N_PROF=i)
    mld = compute_mld(
        profile["TEMP"].values,
        profile["PSAL"].values,
        profile["PRES"].values,
        float(profile["LATITUDE"].values),
    )
    mld_values.append(mld)

print(f"Mean MLD (N=50 profiles): {np.nanmean(mld_values):.1f} dbar")

Step 3 – Compute Eddy Kinetic Energy from Altimetry

import copernicusmarine as cm
import numpy as np
import xarray as xr
import matplotlib.pyplot as plt
import cartopy.crs as ccrs
import cartopy.feature as cfeature

# Download gridded surface current anomalies (AVISO altimetry)
alt = cm.open_dataset(
    dataset_id="cmems_obs-sl_glo_phy-ssh_my_allsat-l4-duacs-0.25deg_P1D",
    variables=["ugosa", "vgosa"],        # geostrophic current anomalies
    minimum_longitude=-80.0,
    maximum_longitude=-40.0,
    minimum_latitude=25.0,
    maximum_latitude=55.0,
    start_datetime="2023-01-01T00:00:00",
    end_datetime="2023-12-31T23:59:59",
)

# Compute time-mean EKE (m²/s²)
eke = 0.5 * (alt["ugosa"] ** 2 + alt["vgosa"] ** 2)
eke_mean = eke.mean(dim="time")

# Plot
fig, ax = plt.subplots(
    subplot_kw={"projection": ccrs.PlateCarree()}, figsize=(10, 7)
)
ax.add_feature(cfeature.LAND, facecolor="lightgray")
ax.add_feature(cfeature.COASTLINE, linewidth=0.5)
ax.gridlines(draw_labels=True, linewidth=0.3, linestyle="--")

pcm = ax.pcolormesh(
    eke_mean.longitude,
    eke_mean.latitude,
    eke_mean.values * 1e4,               # convert to cm²/s²
    cmap="plasma",
    transform=ccrs.PlateCarree(),
    vmin=0,
    vmax=500,
)
plt.colorbar(pcm, ax=ax, label="EKE (cm²/s²)", shrink=0.8)
ax.set_title("Annual Mean Eddy Kinetic Energy – North Atlantic 2023")
plt.tight_layout()
plt.savefig("eke_north_atlantic_2023.png", dpi=150)
plt.show()

---

Advanced Usage

T-S Diagram with Density Contours

import gsw
import numpy as np
import matplotlib.pyplot as plt

def ts_diagram(temp_array, psal_array, pres_array, lat, color_by=None,
               title="T-S Diagram"):
    """
    Plot a Temperature-Salinity diagram with TEOS-10 potential density contours.

    Parameters
    ----------
    temp_array : 2-D array (N_PROF, N_LEVELS)
    psal_array : 2-D array (N_PROF, N_LEVELS)
    pres_array : 2-D array (N_PROF, N_LEVELS)
    lat        : float, representative latitude
    color_by   : 1-D array (N_PROF,) used to color-code profiles, optional
    title      : str
    """
    SA_all = gsw.SA_from_SP(psal_array, pres_array, 0.0, lat)
    CT_all = gsw.CT_from_t(SA_all, temp_array, pres_array)

    # Density contour grid
    sa_grid = np.linspace(SA_all[np.isfinite(SA_all)].min() - 0.5,
                          SA_all[np.isfinite(SA_all)].max() + 0.5, 100)
    ct_grid = np.linspace(CT_all[np.isfinite(CT_all)].min() - 1,
                          CT_all[np.isfinite(CT_all)].max() + 1, 100)
    SA_g, CT_g = np.meshgrid(sa_grid, ct_grid)
    sigma0_grid = gsw.sigma0(SA_g, CT_g)

    fig, ax = plt.subplots(figsize=(8, 7))
    cs = ax.contour(sa_grid, ct_grid, sigma0_grid,
                    levels=np.arange(22, 30, 0.5), colors="gray",
                    linewidths=0.6, linestyles="--")
    ax.clabel(cs, fmt="%.1f", fontsize=8)

    sc = ax.scatter(SA_all.ravel(), CT_all.ravel(),
                    c=color_by if color_by is not None else "steelblue",
                    s=1, alpha=0.4, cmap="viridis")
    if color_by is not None:
        plt.colorbar(sc, ax=ax, label="Depth (m)")

    ax.set_xlabel("Absolute Salinity SA (g/kg)")
    ax.set_ylabel("Conservative Temperature CT (°C)")
    ax.set_title(title)
    plt.tight_layout()
    plt.savefig("ts_diagram.png", dpi=150)
    plt.show()


# Example usage with synthetic data
rng = np.random.default_rng(42)
n_prof, n_lev = 100, 50
pres_syn = np.tile(np.linspace(5, 500, n_lev), (n_prof, 1))
psal_syn = 35.0 + rng.normal(0, 0.3, (n_prof, n_lev))
temp_syn = 15.0 - pres_syn / 50 + rng.normal(0, 0.5, (n_prof, n_lev))

ts_diagram(temp_syn, psal_syn, pres_syn, lat=45.0,
           color_by=np.repeat(np.arange(n_prof), n_lev),
           title="Synthetic T-S Diagram – North Atlantic")

Seasonal SST Cycle from Reanalysis

import copernicusmarine as cm
import xarray as xr
import matplotlib.pyplot as plt
import numpy as np

ds = cm.open_dataset(
    dataset_id="cmems_mod_glo_phy_my_0.083deg_P1D-m",
    variables=["thetao"],
    minimum_longitude=-10.0,
    maximum_longitude=36.0,
    minimum_latitude=30.0,
    maximum_latitude=46.0,
    minimum_depth=0.0,
    maximum_depth=1.0,
    start_datetime="2010-01-01T00:00:00",
    end_datetime="2020-12-31T23:59:59",
)

sst = ds["thetao"].isel(depth=0)

# Compute climatological monthly mean
sst_clim = sst.groupby("time.month").mean(dim="time")

# Spatial mean over the basin
sst_basin = sst_clim.mean(dim=["latitude", "longitude"])

fig, ax = plt.subplots(figsize=(9, 4))
ax.plot(np.arange(1, 13), sst_basin.values, "o-", color="firebrick", lw=2)
ax.set_xticks(np.arange(1, 13))
ax.set_xticklabels(["J","F","M","A","M","J","J","A","S","O","N","D"])
ax.set_xlabel("Month")
ax.set_ylabel("SST (°C)")
ax.set_title("Mediterranean Sea – 2010–2020 SST Climatological Cycle")
ax.grid(True, linestyle="--", alpha=0.5)
plt.tight_layout()
plt.savefig("mediterranean_sst_climatology.png", dpi=150)
plt.show()

---

Troubleshooting

Symptom	Likely Cause	Fix
Missing credentials error	COPERNICUSMARINE_USERNAME / PASSWORD not set	Run copernicusmarine login or export env vars
TimeoutError on large downloads	Requesting too large a spatial/temporal domain	Split into smaller chunks; use cm.subset() to download files locally
KeyError: 'N_PROF'	Argo dataset dimension name differs by product	Inspect ds.dims and adapt index variable
gsw returns NaN arrays	Pressure must be in dbar and non-negative; salinity must be > 0	Filter out fill values (typically 99999) before calling gsw functions
Cartopy install fails on Windows	Binary wheel not available	Use conda install -c conda-forge cartopy instead of pip
404 Not Found for dataset_id	Product retired or renamed in CMEMS catalog	Run cm.describe(contains=["keyword"]) to find current product ID

---

External Resources

• Copernicus Marine Service catalog: https://data.marine.copernicus.eu/products
• copernicusmarine Python package docs: https://toolbox-docs.marine.copernicus.eu/
• GSW-Python documentation: https://teos-10.github.io/GSW-Python/
• Argo data management: https://www.argodatamgt.org/
• de Boyer Montégut et al. (2004) MLD climatology: https://doi.org/10.1029/2004JC002378
• AVISO altimetry products: https://www.aviso.altimetry.fr/
• xarray documentation: https://docs.xarray.dev/

---

Examples

Example 1 – Black Sea Surface Salinity Trend (2015–2023)

import copernicusmarine as cm
import numpy as np
import xarray as xr
import matplotlib.pyplot as plt
from scipy import stats

# Download monthly mean surface salinity
ds = cm.open_dataset(
    dataset_id="cmems_mod_blk_phy_my_2.5km_P1M-m",
    variables=["so"],
    minimum_depth=0.0,
    maximum_depth=1.0,
    start_datetime="2015-01-01T00:00:00",
    end_datetime="2023-12-31T23:59:59",
)

sal = ds["so"].isel(depth=0)
basin_mean = sal.mean(dim=["latitude", "longitude"])

# Convert time to decimal year for linear regression
times = basin_mean["time"].values
decimal_years = [
    float(str(t)[:4]) + float(str(t)[5:7]) / 12
    for t in basin_mean["time"].dt.strftime("%Y-%m").values
]

slope, intercept, r, p, se = stats.linregress(decimal_years, basin_mean.values)
trend_line = slope * np.array(decimal_years) + intercept

fig, ax = plt.subplots(figsize=(10, 4))
ax.plot(decimal_years, basin_mean.values, color="steelblue", lw=1.5,
        label="Monthly mean salinity")
ax.plot(decimal_years, trend_line, "r--", lw=2,
        label=f"Trend: {slope*10:.3f} PSU/decade (p={p:.3f})")
ax.set_xlabel("Year")
ax.set_ylabel("Surface Salinity (PSU)")
ax.set_title("Black Sea Basin-Mean Surface Salinity 2015–2023")
ax.legend()
ax.grid(True, linestyle="--", alpha=0.4)
plt.tight_layout()
plt.savefig("black_sea_salinity_trend.png", dpi=150)
plt.show()

print(f"Trend: {slope*10:.4f} PSU/decade, R²={r**2:.3f}, p={p:.4f}")

Example 2 – Argo Float Profile Visualization with MLD Overlay

import gsw
import numpy as np
import matplotlib.pyplot as plt
import copernicusmarine as cm

# Download a single float's profiles
argo = cm.open_dataset(
    dataset_id="cmems_obs-ins_glo_phy-temp-sal_nrt_argo_P1D-m",
    variables=["TEMP", "PSAL", "PRES", "TIME", "LATITUDE", "LONGITUDE"],
    minimum_longitude=-35.0,
    maximum_longitude=-30.0,
    minimum_latitude=45.0,
    maximum_latitude=50.0,
    start_datetime="2023-07-01T00:00:00",
    end_datetime="2023-09-30T23:59:59",
)

# Select first valid profile
i_prof = 0
pres = argo["PRES"].isel(N_PROF=i_prof).values
temp = argo["TEMP"].isel(N_PROF=i_prof).values
psal = argo["PSAL"].isel(N_PROF=i_prof).values
lat  = float(argo["LATITUDE"].isel(N_PROF=i_prof).values)

# Mask fill values
valid = (pres > 0) & (psal > 0) & np.isfinite(temp)
pres, temp, psal = pres[valid], temp[valid], psal[valid]

# Compute derived quantities
SA = gsw.SA_from_SP(psal, pres, 0.0, lat)
CT = gsw.CT_from_t(SA, temp, pres)
sigma0 = gsw.sigma0(SA, CT)

# MLD (density threshold 0.03 kg/m³)
ref = sigma0[0]
mld_idx = np.where(sigma0 - ref > 0.03)[0]
mld = pres[mld_idx[0]] if len(mld_idx) > 0 else pres[-1]

# Plot
fig, axes = plt.subplots(1, 3, figsize=(12, 6), sharey=True)
ax_t, ax_s, ax_d = axes

ax_t.plot(CT, pres, "b-", lw=1.5)
ax_t.axhline(mld, color="red", ls="--", label=f"MLD={mld:.0f} m")
ax_t.set_xlabel("Conservative Temp CT (°C)")
ax_t.set_ylabel("Pressure (dbar)")
ax_t.invert_yaxis()
ax_t.legend(fontsize=9)
ax_t.grid(True, linestyle="--", alpha=0.4)
ax_t.set_title("Temperature")

ax_s.plot(SA, pres, "g-", lw=1.5)
ax_s.axhline(mld, color="red", ls="--")
ax_s.set_xlabel("Absolute Salinity SA (g/kg)")
ax_s.grid(True, linestyle="--", alpha=0.4)
ax_s.set_title("Salinity")

ax_d.plot(sigma0, pres, "k-", lw=1.5)
ax_d.axhline(mld, color="red", ls="--")
ax_d.set_xlabel("Potential Density σ₀ (kg/m³)")
ax_d.grid(True, linestyle="--", alpha=0.4)
ax_d.set_title("Potential Density")

plt.suptitle(f"Argo Profile – Lat {lat:.2f}°N", fontsize=12)
plt.tight_layout()
plt.savefig("argo_profile_mld.png", dpi=150)
plt.show()