Earth Data School
About
Questions/q10
q10·intermediate

Is Arctic sea ice breaking up earlier each year?

cryosphereoceanclimatesea-ice Datasets: 6 15–30 min
▶ Find the data for your area

Draw a rectangle to pick your area of interest, then see what NASA data covers it (live, here in your browser) or download a ready-to-run notebook with your AOI pre-filled. The notebook runs in any Python environment — it needs a free Earthdata Login to fetch the data.

Current AOI: -160, 70 → -120, 80 (Beaufort Sea)
On this page

Is Arctic sea ice breaking up earlier each year?

What you can answer

  • Sea ice extent time series since 1979 (passive microwave continuity)
  • Sea ice thickness (ICESat-2 ATL10 freeboard converted to thickness via snow depth assumption, 2018+)
  • Breakup-date trend by region (Beaufort, Chukchi, Kara, Laptev seas)
  • Minimum-extent date (typically late Sept, with declining trend)
  • Multi-year ice fraction decline (one of the canonical climate-change signals)

What you can NOT answer with these alone

  • Day-to-day floe-by-floe dynamics without Sentinel-1 + ICESat-2 fusion
  • Below-ice ocean conditions without combining with PO.DAAC ECCO output
  • Pre-1979 — that’s the start of the continuous passive-microwave record

Code template

import earthaccess
import xarray as xr
import pandas as pd

earthaccess.login(strategy="netrc")

# 1. Sea Ice Index daily extent (the operational continuity record)
# Easiest path: download CSV from NSIDC directly
extent_url = ("https://noaadata.apps.nsidc.org/NOAA/G02135/north/daily/"
              "data/N_seaice_extent_daily_v3.0.csv")
extent = pd.read_csv(extent_url, skiprows=2, parse_dates=["Date"])
# Filter to last 30 years; compute minimum-date per year

# 2. ICESat-2 ATL10 thickness for the modern era
arctic_aoi = (-180, 65, 180, 88)
window = ("2018-10-01", "2025-12-31")
atl10 = earthaccess.search_data(short_name="ATL10", bounding_box=arctic_aoi,
                                temporal=window)
# Use icepyx or h5py to read freeboard along strong beams
# Convert freeboard → thickness via assumed snow depth (NESOSIM or W99 climatology)

# 3. Sub-Arctic region breakup date — e.g., Beaufort Sea
beaufort = (-160, 70, -120, 80)
# From extent time-series, find date when SIE drops below 50% per year
# Plot trend of breakup-date vs year

# 4. AMSR-2 daily SIC for high-resolution monitoring
amsr2 = earthaccess.search_data(short_name="AU_SI12", bounding_box=beaufort,
                                temporal=("2024-01-01", "2025-12-31"))

Expected output

  • Time-series: September minimum extent 1979–present (the canonical chart)
  • Regional breakup-date trend: e.g., Beaufort Sea median sea-ice retreat date by year
  • Thickness map: 2024 mean sea-ice thickness from ICESat-2
  • Multi-year ice fraction: 1984 vs 2024 comparison

Caveats

  • Sea Ice Index extent is the consensus reference — use it for any “extent” claim
  • Freeboard → thickness conversion has ~30 cm uncertainty depending on snow-depth product
  • The September minimum trend (~12.7% per decade decline) is the most-cited single signal in Arctic climate — reproduce it as a sanity check
  • Antarctic sea ice is more complex (no clear monotonic trend until 2016 collapse) — different scientific story than Arctic

Cross-DAAC composition

NSIDC DAAC (ICESat-2, AMSR-2, Sea Ice Index) — single DAAC.

Sources

⚲ How a scientist answers this
Parameters
Daily sea-ice extent (million km2) from the NSIDC Sea Ice Index (G02135, passive-microwave continuity since 1979) and AMSR-2 ice concentration; sea-ice thickness from ICESat-2 ATL10 freeboard (2018+, converted via a snow-depth assumption); MODIS/VIIRS for cloud-free imagery context. Breakup/minimum dates are derived per region (Beaufort, Chukchi, Kara, Laptev) against a regional concentration threshold (commonly 15%).
Method
Define breakup as the day concentration/extent in a region falls below the threshold, extract that date each year, and fit a Theil–Sen + Mann–Kendall trend (days/decade) on the breakup-date and minimum-extent series; complement with the multi-year-ice-fraction decline. Robust trend statistics handle the year-to-year variability.
Validation
Use the long passive-microwave record but flag the sensor/algorithm changes spliced into it and the melt-season concentration biases from surface melt ponds; cross-check thickness assumptions (snow depth dominates ATL10 freeboard-to-thickness error) and confirm regional trends against the NSIDC Sea Ice Index.
In plain EnglishFor each region, find the day every year when the ice cover drops below a set level, then see whether that breakup day is creeping earlier over the decades. Robust statistics separate a real trend from the natural noisy ups and downs.

Make it yours → Set the region/sector, the concentration threshold, and the year range in the notebook to compute breakup-date trends for your area.

⌨ Run the core method · no login

The robust trend (Theil–Sen + Mann–Kendall) at the heart of this question — runnable on synthetic data, right here. The full earthaccess code template further down does it on real NASA data (needs an Earthdata login).

editable · runs in your browser