Sensor Notes


In pre-METCRAX-II calibrations, the best pressure reference available was the Paroscientific nanobarometers themselves.  The 3 purchased for SCP (N123996, N123997, N123998) agreed within 0.051 mb and the 1 purchased first (N122850) differed by only 0.2 mb.  We decided to use N123997 as the reference for all of the pressure calibrations -- including the Vaisala PTB220 barometers -- since it was closest to the mean of all 4 nanobarometers.  Our intent is to apply corrections in data processing to remove the small differences of the other 3 nanobarometers.  The Vaisala barometers will be sent to the field with calibrations to N123997 applied internally.

Serial Number Bias (mb) Slope (non-dim)
N122850 -0.2105 1.000218
N123996 0.0507 0.999948
N123997 0.0000 1.000000
N123998 0.0201 0.9999912


During field project operations, there were intermittent periods with missing or bad data from the barometers on the crater floor, but these were most likely caused by low battery voltage rather than failures of the barometers.

Post-project calibrations...


Temperature and Relative Humidity

Pre-project calibrations can be found in the logbook.

Only two of the TRH's were replaced during the project.  The 20m TRH at RIM was replaced on Oct 15, 08:20-08:40, because its temperature was reading high (low) during the day (night) compared to those above and below it, suggesting poor aspiration.  We later noted that the fan was wired backwards so that it blew air in the wrong direction.  The 30m TRH at NEAR was replaced on Oct 18, 11:40 - 16:00 (cable problems arose), because the fan current was intermittently higher than other TRH's and the temperature was often 0.1 deg higher than those above and below it.

Otherwise there were some relative humidity readings which were consistently, but to a minor degree, out of line with those above and below.  An offset of 2.9% was applied to the RH.25m.near data for all samples previous to Oct 8th, 15:30, to fit the profile measurements. Also, an offset of -3% was added to the RH.20m.rim for all samples after Oct 15th, 08:40.

Post-project calibrations can be found in the logbook.



A tipping bucket rain gauge was deployed at the ISFS BASE station.  The only significant precipitation occured in showers on October 10 from 01:50 through 13:20.  Some of this precipitation fell as snow in Flagstaff.


LEAF Wetness

A leaf wetness sensor was deployed on each of the three radiation stands.  The wetness sensors had a base level output of 0.25-0.26 V, and water on the sensor causes an output above the base level.    It was used to edit out data from the radiation and krypton hygrometer sensors, whose data would be erroneous when the radiometer domes or the krypton hygrometer windows were obscured by rain or dew.


3D Sonic Anemometers

During the period from July 25 to August 3, 2013, the wind component offsets for the CSAT3 sonic anemometers to be used in METCRAX-II were measured in the EOL calibration laboratory temperature chamber over a nominal range of -30 to 55 C (see ISFS METCRAX-II wiki logbook for details).  Within the temperature range expected during weather favorable to METCRAX-II research goals (0 to 20 C, based on data collected during METCRAX-2006), most of the sonics had offsets within EOL operating tolerances (+/- 4 cm/s for horizontal winds and +/- 2 cm/s for vertical winds).  Seven out of 25 sonics exceeded these tolerances for one or more wind components and were recalibrated by Campbell Scientific prior to METCRAX-II deployment.

The CSAT3 wind component offsets were measured again after the project.  Only two sonics had offsets below 35 C exceeding EOL operating tolerances.  The 20m sonic at NEAR had a v-component offset of less than -4 cm/s from 25 - 53 C with a maximum offset of -7 cm/s at 43 C.  The 20m sonic at RIM had a v-component offset less than -4 cm/s from 20 - 38 C with a maximum offset of -5.9 cm/s at 31 C.  See logbook for additional details.  Plots of the offsets are also available...

Two of the sonic anemometers, those deployed inside the crater at sites SSW2 and SSW4, transmitted their data via bluetooth wireless connections to the DSM (Data System Module) serving the Flux-PAM station located at the center of the crater floor.   The DSM applies a time tag to each sonic data sample when it is received but, because the latency (delay) time for bluetooth transmission is variable, the DSM time tags for those two sonics will not accurately reflect the times when the data were measured.   (The variation of the transmission latency is on the order of a fraction of a second.)  However the sonic output message for each data sample includes a sequence number that increments from 0 to 63, and these sequence numbers may be used after the project to re-create a time series with evenly-spaced data samples corresponding to the known measurement frequency (20 samples per second). 

The sonic orientations were measured with the DataScope with respect to magnetic north for the sonics at a height of either 3m or 5m at each site.  These azimuths were then adjusted to true north using a declination of 10.5 deg E.  The orientations for the sonics above the lowest level on the NEAR and RIM towers were first assumed to be identical to those measured for the lowest sonic, but then project data with winds at 10m exceeding 10 m/s were used to adjust the sonic azimuths with the assumption that there was no turning of the wind with height during those high-wind cases (see logbook for details).

The sonic data have been rotated with the planar fit technique into coordinates normal to the time-averaged flow field.  Since none of the sonics were replaced or moved during project operations, data from the entire project were used to calculate each of the planar fits.  Files of 5-minute-averaged means and covariances are available in both instrument (or boom-normal) coordinates and flow-normal coordinates.  At the RIM site, the fit was calculated using winds from +/- 60 deg from the assumed axis of the saddle, SSW or 202.5 degrees.  In sonic coordinates this was from -10 to -110 degrees from the sonic u_x axis.  While a mean vertical offset was calculated and applied at most of the sites, a mean vertical velocity of zero was imposed on the data at the RIM site because this gave more believable planar fits to the data (see logbook for details).

Individual sonic data samples have been deleted in post-processing when any of the bits in the diagnostic flag are set.


2D Sonic Anemometers

2D sonic anemometers were used to measure winds at the height of 10m at the FAR, BASE, and FLR sites.  At BASE, the orientation of the 10m sonic was measured with the DataScope by sighting along the side of the tower where the 10m boom was mounted.  At FAR and FLR, the orientation of the 10m sonic was determined from the project data by assuming that the wind direction was identical to that for the 3m sonic during periods of high winds (over 10 m/s at FAR and 3 m/s at FLR).  See logbook for details.


Krypton Hygrometers

The Campbell Scientific krypton hygrometer is a fast-response, ultraviolet-absorption hygrometer used for the measurement of water vapor fluctuations, which are used in conjuction with sonic anemometer vertical velocity measurement for the calculation of turbulent fluxes of water-vapor.   These worked well during the project.  The krypton hygrometer at FLR was replaced on October 11, 10:35 - 10:55, because of intermittently very noisy data.   However, the questionable data continued with the new hygrometer, leading to the conclusion that the problem was rather with the 'serializer' in the sonic electronics, an NCAR data processor that digitizes an analog signal and integrates it into the sonic serial output message.  The serializer in the FLR sonic was replaced on October 14, 12:05-12:10, which cleaned up the krypton signal for several days; however, noise returned on October 24.



There was a signicant outage of the DSM (NCAR local data system) at FLR on October 28 from 10:45 through 14:15.  This was remedied with a simple recyling of power to reboot the system.  The cause is currently unknown.  Data was lost for all sensors on the ISFS 40' FLR tower, radiometers and soil sensors, as well as the barometers and sonics in the array on the SSW (and NNE) slopes.


Mote transmission of Radiation and Soil data

Radiation and soil data was transmitted by Xbee radios at the sensors to the DSM (NCAR data system) mounted on the nearby tower.  During METCRAXII, the radio connection between the Xbee and the DSM was frequently dropped, likely due partly to the current version of the firmware.   The problem was worst at FLR, perhaps because of the heavy bluetooth radio transmission from the barometer sites on the SW slope.  The outages during the first part of the project were often on the order of 3 hours, suggesting that a reset command in the firmware may have both initiated and then later re-established the connection.  Several firmware fixes were tried and gradually mitigated the problem outside the crater.  On October 16 around 14:00, the problem was eliminated at FLR  by hardwiring a direct serial connection from the radiation and soil motes to the DSM. Since the missing data were not saved elsewhere, no actions can be taken to restore these observations.


Radiation Measurements

ISFS measured incoming and outgoing radiation at three locations, FAR, NEAR, and FLR, in concert with measurement of soil heat flux and sensible and latent heat flux. Incoming and outgoing shortwave and longwave broadband radiation will be measured at all three sites, while global and diffuse radiation will only be measured at the near and floor sites.

The radiometers at FAR and NEAR were cleaned fairly frequently throughout the project, while the radiometers at FLR were first cleaned on the afternoon of October 4, but neglected from October 11 at 10:40 until October 19 15:45 - 15:55.  Bird droppings were found on the upward-facing FLR pyranometer on October 19.  It appears that the deposits may have occurred on October 14.  FLR radiometers were cleaned again on October 21, 15:35, again with some bird droppings on the north side of the upfacing pyranometer dome.  Finally, the FLR radiometers were cleaned again on October 26 about 14:54. 



Incoming and outgoing broadband shortwave radiation will be measured with Kipp & Zonen CM21 pyranometers which have been calibrated at NOAA with respect to  their reference active cavity radiometer which has been compared to the World Radiometric Reference (WRR) at the World Radiation Center in Davos, Switzerland (see Don Nelson email 10/06/2011).

See above for periods when the upfacing CM21 on FLR was obscured by bird droppings and below for an analysis that supports using the SPN1 global radiation as a surrogate for the obscured CM21. data were deleted in post-processing from October 14-20 when they departed from a linear fit of vs by more than 20 W/m^2.



Incoming and outgoing broadband longwave radiation will be measured with Kipp & Zonen CG4 pyrgeometers.  These have been calibrated at EOL with outdoor, nighttime intercomparisons with respect to two EOL CG4s calibrated at NREL (July-August 2012) with respect to their transfer-standard Eppley PIR-30557F3 which has been calibrated at Davos, Switzerland by intercomparison to the World Infrared Standards Group (WISG).

The down-facing pyrgeometer thermopile at FAR failed intermittently during the project.  It was essentially flat-lined from 09:10, October 7, until 17:25, October 12.  It appeared to be fixed by opening it up and disconnecting and reconnecting to wires from the radiometer to the embedded microprocessor board.  However it flat-lined again from 17:00, October 13, through 15:10, October 15.  This time we noticed that a bit of solder may have unintentionally grounded one side of the thermopile.  After cleaning this up, the pyrgeometer worked fine through the remainder of the project. The data during these two periods have been deleted in post-processing.  The periods were determined from a time-series plot of Rpile.out.far, as well as an xy-plot of Rlw.out.far vs Rlw.out.near.  The periods estimated from the time-series plot were nearly identical those determined by selecting data from the residuals of a linear fit of Rlw.out.far vs Rlw.out.near.  The bad data were selected when the absolute value of the residuals exceeded 17 W/m^2.


Global and Diffuse Radiation

Global and diffuse shortwave radiation were measured with factory-calibrated Delta-T SPN1 Sunshine Pyranometers at near and flr.  The SPN1 calibration can be evaluated by comparing its global radiation to the incoming broadband shortwave radiation measured with the co-located, up-facing Kipp & Zonen CM21 pyranometer.

An artifact of being in the crater was seen in the flr data.  While still shadowed by the crater wall, both global and diffuse values agreed well, as expected.  Immediately after being exposed to the sun, global radiation went up to nearly the same value as at near, also as expected.  However, dfs dropped to nearly zero for as much as 30 minutes after this "local sunrise".  This would indicate that each of the internal photocells were getting the same amount of radiation (none shaded by the internal shadowband).  Photographs taken by Sebastian Hoch during the dawn of 22 Oct do not confirm this -- some photocells appear to have been completely shaded and others completely illuminated.  During an email exchange, the manufacturer was unable to explain this behavior.  (They did suggest a series of tests which were impossible to perform due to time and logistics.)  We did try rotating the sensor by 120 degrees in azimuth to expose a different part of the instrument.  In this configuration (from 21--26 Oct), the agreement of with near improved and the drop in Rsw.dfs was less, but still significant.  I note that, in Sebastian's photo below, the flr SPN1 sees the crater walls (from all sides).  My guess is that, at low sun angles at dawn, the west crater wall is brightly illuminated, causing the SPN1 to see a lot of sunlight from a wide range of view angles.  In this case, diffuse radiation would be even higher, though this effect would be a smooth transition through sunrise -- not a jump.

The SPN1 manual does state: "The SPN1 is sensitive to fast changes in temperature, and these will create a positive error on cooling, or a negative error on warming." During the 22 Oct dawn event, changed from -2 to +8C in 30 minutes. The sense of this effect is as we observe, though we have not yet checked the magnitude. Thus, this response is still a mystery.

An event (with enhanced radiation in both the global and diffuse outputs) near dawn on 15 Oct probably is caused by dew/frost on the SPN1 dome.  (Air temperatures were near freezing, so we cannot distinguish between dew or frost.)  RH had been over 80% during this night and Tcase on the upward pyrgeometer (as a surrogate for the SPN1 Tdome) was even colder than the air temperature.  Thus, water could have condensed on the dome surfaces. lost data through the mote shortly after dawn, but did appear to have somewhat higher radiation values as well.  The Wetness sensor did not detect this condensation. 


Soil Sensors

During tear-down, it was noted that the soil sensors were installed somewhat differently than normal.  See a logbook table for a list of the soil sensor depths.  At flr, the sensors were nearer to the surface, which may (in part) have been due to some soil blowing away during the project.  Soil heat storage will be derived using the actual depths.


Soil Temperature

A preliminary, cursory look at the soil temperature data suggests that there were several periods of out-of-range data at NEAR, at least one for each depth of the 4-level probe.  These data have been deleted in post-processing.  No out-of-range data were obvious from either FAR or FLR.

Both 4-prong and single soil temperature sensors were installed at each site, in case there were failures.  As noted in the logbook table referenced above, each single probe was installed at a different depth.  Taking these measurement depths into account, the comparison with the 4-prong probes was quite good.


Soil Moisture

The soil moisture sensors at FAR and FLR measured negative soil moisture content with the factory calibration, with the exception of a few days after the rain on October 10. An email exchange with the factory confirmed that odd calibrations would be expected in soils with a high iron content, as was the case here.  (A magnet picked up dirt.) Gravimetric soil moisture measurements on October 22 indicated that these reading should have been changed by -2.4 to 9.6 %volume.  Unfortunately, no other field soil samples were taken.  To compensate for this lack of field measurements, a test plot was set up at the base site and watered.  Data from this plot indicated that a change in slope of a factor of 3 was justified.  However, applying this slope appeared to make Qsoil values worse (see logbook comment).  Thus, we have simply applied biases based on the one gravimetric sample to the final data.  We note that Qsoil at flr appears to be about 2% higher than at the other sites after this adjustment.


Soil Heat Flux

The heat flux sensors all appear to have functioned normally.  Data are noisy following the 11 Oct rain, as expected.  The magnitude of heat fluxes at flr are the largest, near are the lowest, and far is in between.  This behaviour is explained by the different depths at which the sensors were buried (1cm, 5cm, 3cm, respectively).  No data have been edited. 


Soil Heat Capacity

The TP01 data for METCRAXII are the cleanest we've ever seen.  There is a clear signature of the 11 Oct rain event, followed by soil drying.  Most of the data are from dry soil conditions.  A change is not evident after this rain, indicating that the probes had settled immediately after installation.  This could be due either to wet conditions after installation (but before data collection) and/or the ability of sandy soil to reform into a nearly undisturbed state.  There is only one "spike" in Cvsoil.far on 10 Oct 12:00, which presumably was due to rain.  The median values of Cvsoil over the entire project are 1.2, 1.3, and 0.8 MJ/m^3 K for far, flr, and near, respectively.