The SIRS measures the amount of radiation that is upwelling (from terrestrial radiation and reflected solar radiation) and the downwelling radiation (mostly solar radiation, but also longwave emitted from the atmosphere). The SIRS are placed at almost every site at SGP, and are one of the cornerstone instruments of ARM. The SIRS datastream also includes measurements from the IRT (Infrared Thermometer), for both the surface radiating temperature and the atmospheric radiating temperature, which one may compare to the longwave emissions by using Stefan-Boltzmann relation.
Below is a table of the six Radiometer Types and Functions.
Field
Measurement
Code
Radiometer
Location
down_short_hemisp
Downwelling Shortwave
DS
Pyranometer Model PSP (Clear Domes
Metal Post (On Horizontal Plate Inside Ventilator)
down_short_diffuse_hemisp
Downwelling Diffuse Shortwave
DD
Pyranometer Model PSP (Clear Domes)
Solar Tracker (Under Shade Disc Inside Ventilator)
short_direct_normal
Direct Normal Shortwave
NIP
Pyrheliometer Model NIP
Solar Tracker (On Side Bracket)
up_short_hemisp
Upwelling Shortwave
US
Pyranometer Model PSP (Clear Domes)
Tower Boom (Inverted, No Ventilator)
down_long_hemisp_shaded#
Downwelling Longwave
DIR
Pyrgeometer Model PIR (Silver Dome)
Solar Tracker (Under Shade Disc Inside Ventilator)
up_long_hemisp
Upwelling Longwave
UIR
Pyrgeometer Model PIR (Silver Dome)
Tower Boom (Inverted, No Ventilator)
PSP = Precision Spectral Pyranometer (an Eppley Laboratory, Inc. designation, aka 8-48 Black and White)
NIP = Normal Incidence Pyrheliometer
PIR = Precision Infrared Radiometer
Most often the data will be of a reasonable level and withing expected range. To give you a general view of what to expect, here is a table of values for each measurement with four different clear sky levels. Without a TSI, it will be a guestimate as to what the actual amount of cloud cover is at any given moment.
Data Code
Clear Sky Low Sun
Clear Sky High Sun
Cloudy
Night
Comments [Physical Limits]
DS
300 - 500
600 - 1100
100 - 550
-10 to 5
Values can rapidly fluctuate under partly cloudy sky [0 to 1367 (solar constant)]
DD
50 - 200
50 - 200
100 - 550
-10 to 5
DD will increase slightly with cloud amount [0 to 500 under white clouds]
NIP
300 - 700
700 - 1100
-10 to 5
-10 to 5
Shadows indicate NIP is greater than 200 [0 to 1367 (solar constant)]
US
50 - 100
100 - 200
50 - 150
-10 to 5
Fresh snow cover can reflect as much as 90% of the DS [0 to 1230]
DIR
350 - 450
350 - 450
300 - 400
350 - 500
Clouds (day or night) will increase longwave [-40°C => 167 Wm-2]
UIR
350 - 450
450 - 650
350 - 450
400 - 500
UIR typically greater than DIR (the ground is warmer than the sky)
Low Sun = Within a few hours of sunrise or sunset High Sun = Near solar noon (± 1 hour from time when sun is due south) Cloudy = Opaque overcast (not thin cirrus!) Night = In the absence of light, these thermoelectric radiometers can produce negative signals due to temperature lags in the bronze instrument body (values to -15 Wm-2 are possible).
Figure 1 is a prime example of good data quality for the SIRS instrument from 04/04/2005. The short_direct_normal field (pink line) is the amount of radiation incident on the normal plane of the instrument. The down_short_diffuse_hemisp is the diffuse shortwave radiation measure over the whole hemispheric view of the radiometer. This is usually pretty low in magnitude, until clouds come into the picture, which will be discussed later. The down_short_hemispheric field is the total hemispheric radiation on the radiometer. This field is the most prone to small problems (such as dirt on the dome, drift, etc.). The orange line is an estimate of the down_short_hemisp field, using the short_direct_normal and the short_direct_diffuse. The derived estimate is usually a more accepted value than the down_short_hemisp measurement itself. This is due to the long time constant of the radiometer used to measure down_short_hemisperic. Thus in rapidly changing sky cover, large spikes in the measured differences can commonly be observed.Figure 2 shows the downwelling shaded longwave hemispheric radiation on the same day as above. The most important thing to note here is that the measured value is in the same ballpark range as the estimated measurements.Figure 3 shows the upwelling longwave and shortwave measurements for a clear day. The upwelling longwave field is the terrestrial radiation, which is directly linked to the radiative temperature of the ground. Thus there is usually not much change throughout the day, except one can see the diurnal pattern pretty well on clear days. The upwelling shortwave radiation is basically the reflected solar radiation. This is a perfect bell pattern on clear days, but is severely affected by cloud cover, thus a lot of noise can be expected in this field on most days.Figure 4 shows the comparison of calculated net radiation from the SIRS data with the co-located EBBR. The SIRS is usually a better estimate than the EBBR, and thus as long as the two measurements are within 50 Wm-2, it is considered okay. However, from inspection from figure 4, there also can be differences in the time of solar noon. As long as the two are in fairly good agreement, no action is necessary. However, when in doubt, it is probably the EBBR net radiometer that is the culprit.
Figure 5. Cloudy skies typically are difficult to spot problems in data quality. Figure 5 is a good example of a cloudy day at C1 on 04/06/2005. The downwelling shortwave hemispheric and the diffuse shortwave hemispheric radiation typically are about equal in magnitude, and variable throughout the day. Short breaks in the clouds can cause spikes upward in the normal incidence shortwave measurements. The difference field in cloudy conditions should be around zero, with possible spikes and noise in the data. The downwelling longwave radiation plots do not change significantly, so it will not be shown here.Figure 6 shows the upwelling shortwave and longwave components in cloudy sky conditions. Note that the longwave emission is about constant throughout the day. Cloudy skies prevent most of the surface heating, so no change is somewhat expected. The upwelling shortwave plot is very noisy, since most of the solar radiation is scattered and reflected back up from the ground. Net radiation plots are noisy, as one would expect, under cloudy conditions, but can exhibit a variety of different shapes, so one figure could not encompass all cloudy days. Thus, a figure of net radiation plots on a cloudy day are not presented.
Figure 7. The SIRS instruments are probably one of the most important instruments in ARM. However, they are only calibrated once a year in the fall. This means that throughout the year, the instruments will drift, and will become quite noticeable by the following summer (most notably the down_short_hemispheric). Figure 7 shows the plots associated with the drifting sensor. Since OPS cannot generally do much about the drifting, aside from cleaning the PSP (measures down_short_hemispheric) dome, it is good to note the differences when they approach 5% of the overall value of down_short_hemisp, or about 45 Wm-2, but not to issue a DQPR on the subject.Figure 8. The PSP dome can also frequently become dirty, or get a smudge on it. This can severely hamper data quality, but since it is easy to fix and relatively common, it is usually sufficient just to report. Sometimes a good soaking of heavy rain can clear off the PSP without cleaning by OPS. Figure 8 shows a typical clear day, but with a smudge on the PSP dome. A DQPR should be issued if preventive maintenance (PM) has been to the site and not fixed the problem, or the problem is persistent more than 2 – 3 weeks. However, it is left to the DQO to decide whether a DQPR is warranted for such a problem.
Figure 9. The first type of solar tracker failure is complete meltdown. The normal incidence values read zero and the diffuse measurements are reading the same value as the down_short_hemispheric,, which indicates the solar tracker is shading the completely wrong sensor. Figure 9 shows a clear day with this problem. Note how both values are still reporting and still give fairly reasonable estimates for down_short_hemispheric, as well as the perfect curve for a sunny, clear day from the down_short_hemispheric data.Figure 10. The second type of solar tracker failure is a partial tracker failure, which is well documented in DQPR #275. There appeared to be some drift in the north-south direction, which caused the tracker to partially shade the NIP (measures short_direct_normal). This is evidenced in figure 10, where there is a decrease in normal incidence values and a simultaneous increase in diffuse measurements. This particular problem got much worse over time and was not resolved until 9/15, and why it was resolved then, but not other attempts is unknown.
Loose/Damaged wires connecting the sensor to the datalogger can cause frequent, small data gaps. Since data gaps are usually easy to spot, a figure is not shown.
Figure 11. Damaged wires can frequently cause readings to go offscale, or missing. Figure 11 shows a situation where the tracker had spun past 360 degrees and damaged the NIP cable. The diffuse and down_short_hemispheric values look as if nothing had gone wrong, which is the major tell on when it is a tracker failure and when it is a cable problem. Since the diffuse measurements were unaffected, the obvious culprit was a damaged NIP cable. Often times, the actual data values in the file on the DQ server will report -9999, which is the code for missing/offscale, however, the flags may not be set for that variable to be reported as missing. Thus sometimes the flags will report as “incorrect” instead of missing.
Figure 12. This is a very tough problem to catch most of the time. However, it is much easier to spot when the datalogger program is changed, or the instrument is compared to another co-located instrument. One such instance is shown in figure 12. The downwelling longwave measurements developed a significant bias after the datalogger program change, where there was no problem beforehand. Thus, the only thing that could have caused the change was the program that had been installed.Figure 13. A more subtle problem with calibration coefficients is shown in figure 13. The way to tell that something is slightly wrong is how consistent the differences are very large in cloudy conditions. Comparing with the normal cloudy day, the mean difference is lower than the zero line. This is not the only behavior that one can notice for incorrect calibration coefficients, by any means. Whenever a new program is loaded or something is changed out, the serial numbers on the instrument and the associated calibration coefficients within the data logger should match. There is no way for the DQO to check this, so this will only be caught by the mentors or by noticing different behavior before and after the change (different for the worse). The mentors have access to the serial numbers and such, so when issuing the DQPR, suggest that the calibration coefficients be checked (if you think the calibration coefficients are to blame).
¶ SKYRAD down_short_hemisp udner no day light conditions
The down_short_hemisp (dsh) is consistently measuring lower than the other variables and is negative the entire day. At first glance this looks like a problem, but after contacting the mentor it was determined this is OK.
Instrument Mentor Comment:
This is normal nighttime variability for these passive thermopile based radiometers. They are both temperature and wind affected due to heat fluxes and the location of the hot and cold thermopile junctions. Anytime these radiometers are not in steady state they will output a non-zero signal. The instruments themselves are typically accepted to be accurate to about 20 W/m2 at full scale. The larger negative DSH values you see in the PSP data is due to its IR response to a cold sky. This is a well known problem and varies by radiometer type. There is currently (Dec. 2006) an IOP at Barrow dealing with the IR effects on NSA radiometers.
The UIR values from SGP E15, E18, and E19 currently exhibit some seemingly questionable oscillations. If the values are in a "ballpark" acceptable range, you should disregard these oscillations and dub these data OK.
Instrument Mentor Comment:
We sort of just accept it until all the cables are remade some day on the future. It is noise on the excite/delay/measure UIR thermistor measurements for both case and dome. Perhaps due to bad shielding, extended cable lengths or something else there is noise, and not always persistent, on the UIR case and dome temperature measurement. The noise seems to average to zero and can appear and disappear randomly but seems to persist more in cooler months than summer.
¶ Approximate 1 hour drop outs in July - September
Approximate 1 hour drop outs in shortwave measurements are typical for the SIRS network from late July into Septemeber. These drop outs occur due to instrument change-outs being performed by Ops. When these drop outs occur their time and cause (instrument change-out) should be referenced in the DQAs. Two images of these drop outs are below for reference. If you are unsure if an instrument change-out is the cause of a particular drop out you can contact the mentor. They are notified of instrument change-outs and can verify the cause.
Here is an example of one of the drop outs caused by an instrument change-out. Notice that there is at first a very large drop out (14:10 - 14:26 UTC) and then there's a remaining slight drop in some of the data that follows (till ~15:00 UTC).Here is another example. Just like the previous example, there is a large drop out followed by a slight bias in the data (even though the bias is slightly different in this example). Also notice the hours of the drop outs. If they're caused by an instrument change-out they should be occurring at a reasonable hour that Ops may actually be there (aka not at night or late evening).
