TCCON > Sites > Reunion Island > About the Data

About the Data



The Reunion Island spectra are not affected by the laser sampling ghosts.

TCCON measurements at Reunion started in September 2011 with an upgraded laser board (ECL 04), which was optimised for a sample rate of 10 kHz.  At this sample rate, the ghost/parent signal-ratio is less than 1e-4.

2012-11-26 until 2012-12-13

The daily spread of the retrieved Xgas from November 26 2012 until December 13 2012 is considerably larger due to a degrading laser tube.  To obtain realistic errors for Xgas during this period of operation with a degrading laser tube, a new set of error scale factors was calculated for this subset of data, these error scale factors were applied to this subset during postprocessing.  This results in more realistic errors for Xgas during this period, the resulting errors for Xco2 are about three times larger compared to the errors obtained during optimal operation.

The error scale factors for the data recorded with the degrading laser tube are (from .vav.ada):
25  xair                    0.257
27  xhf                     0.760
29  xh2o                    0.530
31  xhdo                    0.486
33  xco                     0.173
35  xn2o                    0.217
37  xch4                    0.182
39  xco2                    0.174
43  xhcl                    0.379

The faulty laser tube was replaced on February 18 2013.

2013-10-14 until 2013-11-20

During this period a too large iris was used (0.8 mm instead of 0.5 mm) for the measurements, this resulted in saturation of the detector at low solar zenith angles.

The measurements which were affected by this saturation of the detector have been flaged based on the value of their corresponding SIA.  The value of "SIA" (solar intensity average) in the Reunion Island data is the mean detector voltage during a measurement, and has a range of 0 V to 0.7 V.  We consider measurements with an SIA > 0.5 V to be saturated.

This problem appeared following a site visit, when the complete automation software of the TCCON observatory was updated.  As a result of this software update the nightly data transfers did not take place for a few weeks, and thus the saturation problem went undetected for some time.

2016-02-07 until 2016-05-22

Here are some remarks to the data during this time.

1) Our prime pressure sensor 'Vaisala PTB210' stopped working on 6 February. We also record in parallel the pressure information using a 'Vaisala WXT520' sensor.  Both our sensors are calibrated on a regular basis to an independent source. The difference between the two pressure sensors is in their precision; the PTB201 has a precision of ± 0.15 hPa at 20 °C whereas the WXT520 has a precision of ± 0.5 hPa [0 °C;+30 °C].
For the currently submitted data, I have used the pressure information from our Vaisala WXT520 sensor.

2) We have a data gap between 8 February and 23 February due to problems with our suntracker motor including the break down of the hydraulic pump.

3) The suntracker mirrors degraded due to sea-salt conditions which resulted in loss of SNR. We increased the aperture to the next step = 0.8 mm (standard operation = 0.5 mm). The measurements with increased aperture are between 24 Feb to 24 April.

4) We had a site visit by Christian at end of April when he replaced the elevation mirror of the suntracker with a new one. This improved the signal to noise significantly. But the aperture was not changed back to 0.5 mm. This went unnoticed for some time. While analyzing the data now I realized that this caused a problem in the retrieval. The signal strength was too strong and the measured signal on the detector went to the non-linearity region of the detector. As a result, the retrieval result has to be handled with care. There are several measurements which show error in the sia (Solar intensity average). Although, most of the mid-day measurements are flagged, we could still use some measurements in the early morning and late afternoon (high SZA) which give reasonable result. However, due to high SZA and still high signal we have high error compared to our standard error. The affected time period is between 25 April to 22 May 2016. I think it is still good to provide the filtered data for our site as this will avoid data gap. We can provide information about handling the data with care.

2017-06-01 until 2017-08-31

The measurement computer at St. Denis had problems in July - August 2017 and finally it dies (after 6 years) towards the end of August. Before dying completely one of the hard disk was not recognized for a few days and some when it appeared again. So the computer was not stable and the log file of the NetTime programme (this programme is synchronizing the time) was overwritten with the new dates when I checked in the backup files of the computer. So I have no information of the old log data to verify if we had a timing error. We have a new computer now installed for the measurements. 

The timing error was identified by looking at the Xair as well as the individual column values of the gas retreival results. This was not so obvious in the Xgas values due to the cancellation of errors in the division of the two column values in the Xgas calculations.

The have written a short documentation "Spectra_time_correction_JulyAug2017.pdf" describing the error and the method used to correct the time error in the data. This pdf document can be downloaded from the link at the bottom of the page.



Tag page

Files 1

FileSizeDateAttached by 
Spectra time error correction for July - August 2017 RĂ©union data
1382.02 kB00:10, 25 Oct 2017m_shaActions