Air pressure at 47 m
Half-hourly atmospheric pressure at 47 m on the CH-LAE tower, September 2005 to September 2026, exported as measured. The file is 05_METEO_PA_2005-2026.09 (parquet and CSV): 369,552 records on a continuous 30-minute index TIMESTAMP_MIDDLE (middle of the half-hour, UTC+01:00, no daylight saving), from 2005-09-02 00:15 to 2026-09-30 23:45.
The interactive dashboard summarises the product on one page.
Method, evidence and checks:
05_METEO_PA: builds the product: unit harmonisation of the two screenings, 2012 corrections, identification of the barometer from the logger programs, the check of the absolute level against the barometric formula, and the two hardware boundaries.- Meteo Product Chain: no other meteo product is computed from this one.
This product is not gap-filled. There is one value column and no flag column. About 1 % of the record is missing: a non-null value is a measurement and a NaN is a real gap.
There is no _HOMOGENIZED column, although the record changes level by about 0.09 kPa at the January 2016 acquisition change. See Known limitations for what to do about it.
Columns
05_METEO_PA_2005-2026.09.
| column | unit | description |
|---|---|---|
PA_T1_47_1 |
kPa | Atmospheric pressure at 47 m, measured and corrected. Gaps are NaN. |
The unit is kPa, the FLUXNET convention. The database stores the two screenings in different units, kPa before 2022 and Pa after, because the two loggers wrote different units. The notebook harmonises them from the magnitude of the data and stops if that result disagrees with the database unit tag. Measured values range from 89.45 to 95.54 kPa, with a median of 93.31 kPa.
Coverage
365,542 of the 369,552 records carry a value, 98.9 %. The record begins on 2005-09-02, when the archive of this logger begins; the barometer was already in service, but no earlier data are held. There is no PA for 2004 and the first eight months of 2005, and none is reconstructed, so this product starts later than SW_IN, TA and PPFD_IN.
Of 415 gaps, 399 are one hour or shorter. The longest is the 20.6-day outage of January 2016, during which the tower logger was replaced.
| year | measured | why |
|---|---|---|
| 2005 | 99.4 % of 2 Sep to 31 Dec | the record begins on 2 September |
| 2009 | 97.0 % | a 9.8-day outage in August |
| 2010 | 98.9 % | outages |
| 2012 | 94.1 % | power-supply failure and storm damage |
| 2016 | 94.3 % | the January outage during which the logger was replaced |
| 2019 | 95.7 % | outages |
In 2012 a logger clock error shifted one block of August by 15.5 hours and was corrected. A tower power-supply failure in late July and August and storm damage in late October and November left biased values, and those periods were removed. The removal windows come from this variable’s own disagreement with the independent hut barometer, so they do not coincide exactly with the windows of SW_IN, TA and PPFD_IN. Use each product’s own missing values rather than assuming the products exclude identical periods.
The sensor and its absolute level
The instrument is a Vaisala PTB101B barometer. Every logger program that has measured it, the six CR10X programs of 2004-2006 and the CR1000 program installed in January 2016, converts its signal with the same multiplier and offset, spanning 600 to 1060 hPa, the range of that barometer. The maintenance record does not name the instrument in twenty-one years: as far as the record goes it was never replaced, calibrated or cleaned, and it is not in the device inventory.
Pressure is the one meteo variable here whose absolute level can be checked against physics, and it passes. The barometric formula, with nothing fitted to the tower data, predicts the difference between this barometer and each reference from that reference’s own pressure, temperature and humidity. It places the barometer within 5 m of its nominal elevation using MeteoSwiss Lägern (2.5 km away at 845 m, about 110 m above the barometer) and within 9 m using the NABEL barometer at the hut; the two references agree with each other to 4 m. It also reproduces the annual wave in the difference between two barometers at different heights (89 Pa measured, 79 Pa predicted, correlation 0.98). A unit error, a wrong offset or a lost factor of ten would fail this check by orders of magnitude.
The January 2016 acquisition change
It moved the level by about 0.09 kPa, the only hardware change in this record with a measurable effect.
The barometer and its conversion did not change. The logger did: the CR10X measured the sensor single-ended, the CR1000 measures it differentially. A single-ended measurement includes any offset between the sensor’s ground and the logger’s; a differential one does not.
Three barometers place the change on the tower. Both differences involving the tower move across 2016, by -0.0884 and -0.0888 kPa, while the difference between the two references moves by -0.0001 kPa. The barometric prediction shows which era is correct: averaged over 2016 to 2025, the later era agrees with the prediction against both references and the earlier era lies about 0.1 kPa above it. Year by year, the departure of the later era from the prediction against Lägern goes from about +0.02 kPa in 2016 to about -0.06 kPa in 2025 (the drift under Known limitations). The record before 21 January 2016 therefore reads approximately 0.09 kPa high.
The 20.6-day outage that ends on 21 January 2016 brackets the step, so it cannot be dated more precisely. The logger program on disk is dated 18 January 2016, and the maintenance record names the same program file.
The December 2021 change of screening software has no effect here: the difference against MeteoSwiss changes less across that boundary than in an ordinary year.
Known limitations
- The record is not homogeneous across 21 January 2016, and is not corrected. Values before that date read approximately 0.09 kPa high. No homogenised column is provided: the step is one part in a thousand of the value (under ten metres of air), so air density and the flux quantities computed from it change by the same one part in a thousand; a rescaled column would change the set of columns this product and the merged product ship; and a residual drift of the same size (next point) would remain after any homogenisation. For one absolute level across the whole record, subtract 0.09 kPa from the records before 21 January 2016. Analyses within a year, or of pressure changes rather than levels, are unaffected.
- A slow drift of about 0.1 kPa over twenty years cannot be attributed. Apart from the two dated events, the difference between this barometer and MeteoSwiss Lägern falls by approximately 0.1 kPa across the record. Slightly more than half of that falls after mid-2018, when the second barometer at the site stops, so no third instrument can show which of the two is moving. It develops over years, so there is no boundary for a correction, and the barometric formula cannot settle it because it is computed from Lägern’s own pressure. January to September 2026 continues the drift: the difference against Lägern lies 0.023 kPa below the same months of 2025, the largest year-on-year change since 2016, but below the 0.04 kPa at which the notebook counts an event.
- The MeteoSwiss Lägern pressure changed level in October 2010. The rebuild of that station’s instrumentation that moved its radiation record also moved
PA_LAE_MS, by approximately 0.06 to 0.07 kPa relative to both barometers at this site. This product never uses Lägern to correct a value, but do not read a difference between this product and MeteoSwiss Lägern as evidence about the tower across that date. - The NABEL hut barometer is not independent after July 2018. The database series
PA_NABEL_H1_2_1runs nominally to the end of 2018, but from 3 July 2018 its values repeat the tower series to five decimal places. The notebook truncates the reference there. Any other use of that series should do the same. - The record begins in September 2005. When joining this product to
SW_IN,TAorPPFD_IN, whose index starts on 2004-01-01, expect a shorter series.