Launch Watch · Analysis
Published
NASA’s sea-level zero point: a 1,093-row baseline audit
NASA’s classroom sea-level series adds 38.56 mm to every source value. Audit all 1,093 rows and see which changes that shift preserves.
The last sea-level value in one tab of NASA’s classroom workbook is 62.29 millimeters. In the other tab, the value at the same date is 100.85 millimeters. Neither number needs to be discarded: the classroom series has been shifted upward by 38.56 millimeters so that its first value is zero.
We matched all 1,093 rows between the two tabs in the NASA workbook, rather than inferring the transformation from the chart alone. The dates agree within workbook storage precision, and every classroom value equals the corresponding original column-12 value plus 38.56 millimeters, within floating-point rounding. The shift changes the vertical reference. It preserves the 100.85-millimeter rise between the first and last records.
The endpoints explain the two numbers
| Source decimal year | Original column 12 (mm) | Classroom tab (mm) | Added offset (mm) |
|---|---|---|---|
| 1993.011526 | −38.56 | 0.00 | 38.56 |
| 2022.684953 | 62.29 | 100.85 | 38.56 |
On the original reference, the endpoint change is 62.29 − (−38.56) = 100.85 millimeters. On the classroom reference, it is 100.85 − 0.00 = 100.85 millimeters. A negative starting value therefore does not describe a negative depth of ocean. It describes sea-level variation below the chosen reference level.
The original data header identifies a TOPEX/Jason 1996–2016 collinear mean reference. The classroom tab instead makes the first reading the zero point. Those references answer different display questions while retaining the same changes over time.
Method: match every date before subtracting
We read the workbook without modifying it. The “Derived Data Set” tab holds decimal-year dates and sea-level values; “Original Data Set” embeds the source header and its 12-column numerical records. We extracted both sets, verified that paired decimal-year values agree within storage precision, and subtracted original column 12 from the classroom value. The workbook contains stored numbers rather than cell formulas for this transformation.
Across all 1,093 pairs, the offset ranges from 38.559999999999995 to 38.56000000000001 millimeters in binary floating-point arithmetic. Both limits are 38.56 at the source’s hundredth-millimeter reporting precision. Checking all pairs matters: two matching endpoints alone would not rule out changes in the middle of the series.
We also fit an ordinary least-squares straight line to each series using the same decimal-year dates and an intercept. Both slopes round to 3.397022 millimeters per year. Adding a constant changes the fitted intercept, while leaving the slope and each adjacent change unchanged. This is an arithmetic check on this frozen series, not an uncertainty analysis or an estimate of today’s rate.
Reproduce the full 1,093-row reconciliation
Download the complete reconciliation CSV. It contains the matched decimal year, both sea-level values, their offset, and the original column-9 and column-11 values used in the processing comparison. The source dates run from 1993.011526 to 2022.684953. The check requires an absolute offset error below 0.0000000001 mm, far below the supplied 0.01 mm reporting precision.
For the slope check, fit each value series against the same decimal-year variable with a free intercept. Subtract the first decimal year from every date before fitting for numerical convenience. That shift in the time origin also leaves the slope unchanged.
A baseline shift is different from smoothing
The embedded source labels column 12 as a series with the GIA correction applied, a 60-day Gaussian-type filter applied, and annual and semiannual signals removed. The classroom’s extra 38.56-millimeter shift happens after those operations. It must not be described as the GIA correction or as seasonal adjustment.
The other columns show why. Original column 9 contains the GIA-adjusted, unsmoothed series. Column 11 contains its smoothed counterpart. Across the workbook, column 11 minus column 9 ranges from −9.01 to +18.55 millimeters. Column 12 minus column 11, associated with the seasonal-signal removal, ranges from −5.29 to +4.00 millimeters. Neither comparison is a constant offset. They can change the shape of the curve; changing its zero point cannot.
Keep the release and the reference with the result
The NASA classroom data sheet accompanies a historical teaching dataset. The workbook identifies its source date as 23 January 2023 and cites GSFC/PO.DAAC version 5.1. Its header marks estimates after 28 March 2022 as preliminary. Reusing the data today does not update that status or turn the fitted historical slope into a current rate.
Global mean sea-level variation also does not give the height of water at a particular coast. This audit supports a narrower statement: all 1,093 classroom values are the same released column-12 series shifted by 38.56 millimeters. When comparing two charts, identify the reference, processing and source release before interpreting a vertical separation as a physical disagreement.
Sources cited in this article
- NASA classroom sea-level workbook
- NASA classroom sea-level data sheet
- NASA sea-level height dataset page
- GSFC. 2021. Global Mean Sea Level Trend from Integrated Multi-Mission Ocean Altimeters TOPEX/Poseidon, Jason-1, OSTM/Jason-2, and Jason-3 Version 5.1. Version 5.1. PO.DAAC, California, USA. DOI: 10.5067/GMSLM-TJ151. Data accessed through NASA’s classroom workbook on 4 October 2026.
The dataset citation links to the official Version 5.1 release announcement, which records DOI 10.5067/GMSLM-TJ151. The DOI landing page returned an unavailable-page response when checked on 4 October 2026; the NASA workbook and its embedded source tab remain the inputs used for this audit.