Relative humidity at 47 m
Half-hourly relative humidity at 47 m on the CH-LAE tower, 2004 to September 2026, complete. The file is 04_METEO_RH_CORRECTED_2004-2026.09 (parquet and CSV): 398,832 records on a continuous 30-minute index TIMESTAMP_MIDDLE (middle of the half-hour, UTC+01:00, no daylight saving), from 2004-01-01 00:15 to 2026-09-30 23:45. 2026 is a partial year (January to September).
The interactive dashboard summarises the product on one page.
Method, evidence and checks:
04_METEO_RH: builds the product: merge of the two screenings, 2012 corrections, theRH > 100 %correction, the identification of the probes and their states, the 2016 and 2026 transfers, and the two reconstruction passes.02_METEO_TA: the same probe deliversTA, and the dates of every probe change and probe state are derived there. Notebook04reads its product, so02runs first.- Meteo Product Chain:
VPDis computed from this product.
The 47 m record comes from three probes, and the second of them changed state twice before it was replaced (see Sensor history). The measured column steps up by 2.1 to 3.3 percentage points of RH at the 21 January 2016 probe change (3.9 to 4.1 on the half-hours away from saturation). At the 5 June 2026 probe change it becomes drier, increasingly so with the humidity.
- Inside one probe state, use
RH_T1_47_1, which is what the probe reported. - Across 21 January 2016 or 5 June 2026 (period means, trends, year rankings, anomalies) use
RH_T1_47_1_HOMOGENIZED, which puts the other probes on the level of the CS215. It removes the steps at the probe changes, not the drift within each era (see Known limitations). - Where provenance has to be certain, select
FLAG_RH_T1_47_1_SOURCEin{0, 1, 4, 6}. Codes6and7carry the probe’s heating in sun in both columns; exclude code7wherever the sun-dependent part ofRHmatters.
Columns
Every record carries a value, and both flags are defined at every record and describe both value columns.
04_METEO_RH_CORRECTED_2004-2026.09.
| column | unit | description |
|---|---|---|
RH_T1_47_1 |
% | Relative humidity, corrected and gap-free, every probe as measured. Bounded to [0, 100]. Not homogeneous across 2016-01-21 or 2026-06-05. |
FLAG_RH_T1_47_1_MISSING |
- | Whether the value was measured at 47 m, and if not, which sensor it was reconstructed from. |
RH_T1_47_1_HOMOGENIZED |
% | The same series on the level of the CS215. Bounded to [0, 100]. |
FLAG_RH_T1_47_1_SOURCE |
- | Which probe, and which state of it, stands behind the record. |
FLAG_RH_T1_47_1_MISSING
Filter on == 0 for measurements at 47 m. Every other code marks a value transferred from another instrument. Nothing in this product is modelled: a reconstructed value is another measurement of the same quantity, corrected by a monthly offset fitted within the probe state it belongs to.
| code | meaning | records | share |
|---|---|---|---|
| 0 | measured at 47 m | 382,181 | 95.83 % |
| 3 | reconstructed from NABEL at 49 m on the same tower | 15,729 | 3.94 % |
| 4 | reconstructed from MeteoSwiss Lägern, 2.5 km away | 922 | 0.23 % |
Codes 1 (never measured) and 2 (removed as faulty) are defined in the notebook but do not occur: the two references cover every gap.
Code 3 carries about 3 % RH of uncertainty, from leave-one-year-out validation against the years both sensors measured. Code 4 is weaker, since Lägern is 2.5 km away on the ridge at 845 m, about 100 m above the tower sensor (the tower stands at about 690 m). It occurs almost only after the NABEL sensor ends in 2018: 877 of its 922 records fall in 2019 (759), 2023 (9), 2024 (107), 2025 (1) and 2026 (1), the last two being the two probe half-hours. The other 45 fill short NABEL gaps in 2004 (6), 2005 (24), 2011 (10) and 2016 (5).
FLAG_RH_T1_47_1_SOURCE
The codes are those of notebook 02, and notebook 04 asserts that the two flags agree on every record.
TIMESTAMP_MIDDLE).
| code | probe and state | period | records |
|---|---|---|---|
| 0 | Campbell CS215 on SDI-12, normal state | 2016-01-21 14:15 to 2024-10-21 23:45 | 153,428 |
| 1 | Rotronic MP101A, single-ended analog | 2004-09-20 10:45 to 2015-12-31 23:45 | 197,739 |
| 2 | acquisition changeover, probe undetermined | 2016-01-01 00:15 to 2016-01-21 13:45 | 988 |
| 3 | before the tower record begins | 2004-01-01 00:15 to 2004-09-20 10:15 | 12,645 |
| 4 | Campbell HygroVUE10 on SDI-12 | 2026-06-05 11:45 to 2026-09-30 23:45 | 5,641 |
| 5 | probe handling or probe swap half-hour | 2025-05-16 09:45 and 2026-06-05 11:15 | 2 |
| 6 | Campbell CS215 after the handling of 16 May 2025, reads high in sun | 2025-05-16 10:15 to 2026-06-05 10:45 | 18,482 |
| 7 | Campbell CS215 with excess daytime heating, reads low in sun | 2024-10-22 00:15 to 2025-05-16 09:15 | 9,907 |
No probe at this height stands behind codes 2, 3 and 5. During the changeover (code 2) the tower recorded nothing for 20 days, and those records are reconstructed from NABEL. The two code-5 half-hours are removed and reconstructed from Lägern with the offset of the state before each. To stay inside one probe state, select a single code.
Coverage
The tower measurement begins on 2004-09-20 10:45. Everything before that is reconstructed from NABEL.
| year | measured | why |
|---|---|---|
| 2004 | 28.0 % | the record begins on 20 September |
| 2009 | 97.2 % | scattered outages |
| 2012 | 92.8 % | logger clock error, power-supply failure and storm damage |
| 2016 | 94.3 % | the January outage during which the probe and logger were replaced |
| 2019 | 95.7 % | an outage after the NABEL reference ends, filled from MeteoSwiss |
In 2012 a logger clock error shifted one block of August by 15.5 hours, and a tower power-supply failure in late July and August and storm damage in late October and November left records that could not be repaired. The clock error was corrected, the other two periods were removed, and all of it was reconstructed from NABEL.
Sensor history
TA_T1_47_1 and RH_T1_47_1 arrive from one probe in one sensor call, so every date below is the same as in notebook 02, where most of them are derived. The field name and its database tags did not change at any of them.
- Rotronic MP101A, to 31 December 2015 (code
1). Six CR10X programs of 2004-2006 name it and its analog conversion. - Campbell CS215, from 21 January 2016 (code
0). Named by the CR1000 program installed at the changeover, read over SDI-12. - The CS215 with excess daytime heating, 22 October 2024 to 16 May 2025 (code
7). Dated in notebook02from the daytime excess of the 47 m temperature over the 35 m sensor of the same tower, which grows in the weeks after intake tubing was attached at the 47 m sonic anemometer on 19 September 2024. - The CS215 after 16 May 2025 (code
6). The probe was handled during a site visit that day; the fieldbook records the heating connection of the tower inlet found melted and disconnected. See The May 2025 change. - Campbell HygroVUE10, from 5 June 2026 (code
4). The fieldbook adds it at this location that day. See The June 2026 probe change.
The fieldbook registers no temperature or humidity device at the 47 m location before the HygroVUE10, so the 2016 identification rests on the logger programs and on the data.
RH_T1_47_1_HOMOGENIZED applies to each source code.
| source code | correction |
|---|---|
| 1, 2, 3 | the 2016 transfer, which depends on the humidity, added to every record |
| 0, 5, 6, 7 | none: the CS215 is the reference level |
| 4 | the HygroVUE10 transfer, made in vapour pressure and converted back with the homogenised TA of notebook 02 |
What happened in 2016
The probe and its acquisition system were replaced together: a Rotronic MP101A on a single-ended analog channel gave way to a Campbell CS215 over SDI-12. Air temperature and relative humidity come from that one probe, so the same date splits both variables.
The step is on the tower. Three humidity series span the changeover: the tower probe, a NABEL probe at 49 m on the same mast, and MeteoSwiss Lägern. Both comparisons that contain the tower probe step up at January 2016 by about ten times their ordinary year-to-year change, and the comparison between the two references does not move.
| comparison | change across 2016 | ordinary year | away from saturation |
|---|---|---|---|
| tower minus NABEL 49 m | +3.32 | 0.29 | +4.14 |
| tower minus MeteoSwiss Lägern | +2.12 | 0.22 | +3.85 |
| NABEL 49 m minus MeteoSwiss Lägern | -0.35 | 0.61 | -0.24 |
The 100 % ceiling hides part of the step. The measured series has a median of 82.8 % RH, and 11.1 % of its half-hours are at or above 99.9 %. Where both probes are at that bound the step cannot show, so the whole-record figure understates it: away from the ceiling the step is 1.2 to 1.8 times as large. Below 85 % RH it is 3.2 to 6.3 percentage points, above 95 % RH at most 1.1.
The replaced probe had drifted. 1,931 half-hours of the analog era (0.99 % of its 195,606 measured records) read above 100 % RH, the typical sign of an aged capacitive humidity element. Their share grows through that era to 3.6 % of 2015. The SDI-12 era has none.
The step is present at every hour, and is larger at night (+4.3 % RH against NABEL) than in daylight (+2.2 % RH). The radiation-shield difference that explains part of the TA step acts only in sunshine, so it does not explain this one.
The 2016 transfer
Because the step depends on the humidity, a constant cannot remove it. The homogenised column raises the MP101A era by a transfer that varies with the humidity: per humidity bin, the difference between the eras is added to the earlier one.
Fitted on the era means, such a transfer is not stable. Both eras drift against Lägern, in opposite directions and by more over their own length than the step between them. In the 60 to 80 % RH band, over January to September of each year, the MP101A era drifts by -0.34 percentage points of RH per year (-3.44 over 2005-2015) and the CS215 era by +0.44 per year (+3.07 over 2017-2024): 6.51 in total, against a step of 5.46 between the era means. Estimates from the first and second half of each era differ by up to 4.93 % RH, the size of the correction, and summer and winter by up to 4.72 % RH.
Fitted at the break, it is stable. The transfer is fitted on tower minus NABEL, binned on the NABEL humidity: the MP101A over 2013-2015 against the CS215 from February 2016 to July 2018, the end of NABEL’s independent record. Close to the break the drift has had little time to act. Shorter windows change the result by less than the year-to-year noise of each bin plus a year of drift: from 40 % RH upwards by at most 1.27 % RH, never more than 0.95 of that bin’s tolerance. Each bin’s correction is attached to the MP101A’s own reading in that bin, interpolated between bins, and the result is capped at 100 %, so a saturated MP101A record stays saturated.
The transfer raises the MP101A era by 0 to 5.51 % RH, by 2.98 % RH on average over its measured records, most in the middle of the humidity range and close to zero near saturation. Scored against Lägern, which entered no fit, the mean step across the break falls from +4.21 to +0.83 % RH, within a tolerance of 1.81 % RH (Lägern’s own year-to-year change, 0.71, plus the drift contained in the fitting windows, 1.10). Every record before the changeover (codes 1, 2 and 3) receives the transfer, reconstructed records included, because their reconstruction offsets put them on the MP101A level.
The May 2025 change
The CS215 reports RH relative to its own temperature, so a probe that is too warm in sun reads a lower RH for the same amount of water vapour. Whether the change of 16 May 2025 is of that kind or a change of the humidity element is decided in vapour pressure, computed from each sensor’s own RH and TA: a temperature error moves RH and leaves vapour pressure unchanged.
Over the 30 days either side of the handling, against the same calendar dates of the normal CS215 summers 2022-2024, RH in sun rises by +4.43 % RH against Lägern, outside the range of the normal years (-3.07 to +1.13 % RH). The vapour pressure changes by +0.13 hPa, within theirs (+0.13 to +0.45 hPa), and the tower temperature in sun falls by 1.57 °C against Lägern. Against the normal years RH moves by +5.17 % RH, which at 17.9 °C corresponds to 1.06 hPa; the vapour pressure moves by -0.20 hPa, 19 % of that. The change is in the probe’s temperature, not in its humidity element.
RH_T1_47_1_HOMOGENIZED therefore leaves codes 6 and 7 as measured, like code 0. A correction of these records would be a correction of the probe temperature, which is the task of notebook 02. Against Lägern, RH in sun is +0.18 % RH under code 7 and +4.62 % RH under code 6, where the normal CS215 summers on the same dates give +0.18 to +3.37 and -0.20 to +3.12 % RH.
The June 2026 probe change
On 5 June 2026 a Campbell HygroVUE10 (SDI-12) replaced the CS215. A gap in the raw 1-minute record from 11:10 to 11:31 dates the swap; the half-hour 11:00 to 11:30 is removed and reconstructed (source code 5). In vapour pressure against Lägern the HygroVUE10 reads drier than the normal CS215, by 1.27 hPa over 12 June to 30 September, against a year-to-year scatter of 0.19 hPa among the CS215 summers. In RH as measured it lies 2.3 to 4.6 % RH below the CS215 summers between 30 and 90 % RH.
The HygroVUE10 also reads cooler than the CS215 in sun (notebook 02), so a transfer in % RH would mix the humidity elements with the temperature sensors. The transfer is therefore made in vapour pressure:
- Vapour pressure
eis computed from each sensor’s ownRHandTA, on two season-matched samples: the HygroVUE10 from 12 June to 30 September 2026, after its configuration change of 11 June, and the normal CS215 (code0) on the same calendar dates of 2022 to 2024. - The HygroVUE10-minus-CS215 difference
Delta_eis fitted as a line in the Lägern vapour pressure (5,328 HygroVUE10 and 15,885 CS215 records). It runs from -0.90 hPa at 8 hPa to -1.66 hPa at 20 hPa. After the transfer the residual over all hours of the sample is -0.08 hPa, against -1.27 hPa before and a tolerance of 0.19 hPa, and no vapour pressure bin exceeds 0.09 hPa. In a month hold-out (each month predicted by fits without it) the residuals are -0.17 hPa in June, -0.32 hPa in July, -0.01 hPa in August and +0.19 hPa in September, all within a tolerance of 0.37 hPa pooled over the months. RH_HOMOGENIZED = 100 * (e - Delta_e) / e_s(TA_HOMOGENIZED), wheree_sis the saturation vapour pressure at the homogenisedTAof notebook02for the same record, capped to[0, 100].
From June 2026 the homogenised RH refers to the homogenised temperature of notebook 02, not to the probe’s own temperature. That temperature carries no radiative excess, so in sun the homogenised RH lies above what a CS215 would read, and at night close to it. Against the CS215 summers it lies 0.74 % RH higher in sun, 0.30 % RH higher at night and 0.60 % RH higher over all hours; per Lägern humidity bin, 0.03 to 1.25 % RH higher between 30 and 90 % RH and 2.52 % RH higher above 90 %. On the HygroVUE10 records the homogenised column adds 3.92 % RH on average, at most 8.21 % RH.
Known limitations
- The homogenised column removes the steps, not the drift. Against MeteoSwiss Lägern the MP101A era drifts down by 0.34 and the CS215 era up by 0.44 percentage points of RH per year, so each era moves further over its own length than the 2016 step (see The 2016 transfer). A trend from either column, or a comparison of years far from the break, still contains sensor movement. The drift is not corrected: it cannot be attributed to either station, and after the NABEL sensor stops in 2018 there is no third humidity series at the site to attribute it with. Correcting it towards MeteoSwiss would replace the measurement with the reference.
- The 2016 transfer depends on the reference. Fitted against Lägern over the same windows instead of NABEL, it would raise the measured MP101A records of 2013-2015 by +3.36 instead of +2.82 % RH on average, 0.54 % RH more, and by up to 2.44 % RH more just below saturation (at a reading of 95 %), where NABEL finds almost no step and Lägern a large one. That difference is the uncertainty of the homogenised column across 2016, and it is largest in humid conditions. Shorter fitting windows give a correction smaller by 0.61 to 0.69 % RH on average between 40 and 85 % RH, so part of what the fitting windows attribute to the probes may be drift.
- Codes
6and7carry the probe’s heating in both value columns: in sun code7reads low and code6high against the normal CS215. - The HygroVUE10 transfer rests on one summer and one reference. It is fitted on June to September 2026 against Lägern and applied to the whole HygroVUE10 era, including the transitional week of 5 to 11 June. Its month hold-out residuals follow the season, from -0.32 hPa in July to +0.19 hPa in September, so the transfer leaves the early summer months slightly too dry and the late ones slightly too wet. Its behaviour outside the summer is untested, and it is to be refitted once the HygroVUE10 record includes a winter.
- Which probe is correct is not settled. The homogenised column is on the level of the CS215; the evidence does not establish an absolute accuracy for any of the probes.
VPDinherits these properties. Notebook07computesVPD_T1_47_1from the measuredTAandRHandVPD_T1_47_1_HOMOGENIZEDfrom the two homogenised columns. Use the homogenisedVPDacross 2016 or June 2026. See Vapour pressure deficit at 47 m.- Reconstructed records are estimates, not measurements. Code
3carries about 3 % RH of uncertainty, against a sensor-to-sensor agreement of about 2 % RH in summer; code4more. January to September 2004 is reconstructed throughout and is the least constrained part of the record. - Values near 100 % are corrected, not restored. A drifting probe loses resolution close to saturation. Where the analog era read above 100 %, a daily offset was subtracted and the remainder clipped, which removes the impossible values but does not recover the information they stood for. Saturation frequency differs between the eras for this reason as well as because of the step.
- The NABEL reference is not independent after July 2018. From August 2018 the ingested
RH_NABEL_T1_49_1column reproduces the tower column exactly. Nothing in this file is derived from those months, but a comparison of this product with that NABEL column over them compares the product with itself. - 2026 is a partial year. A 2026 annual mean covers January to September only.