A guide to the Stratospheric ozone measurement program using Dobson Ozone Spectrophotometers.

The questions are based on the questions we have received from various sources. The guide is meant for observers, project leaders, and others making observations with a Dobson Ozone Spectrophotometer equipped with an encoder and computer in the US Weather Service, or as a cooperative program with the Climate Monitoring and Diagnostic Laboratory.

I've been assigned the Ozone Focal Point (or Lead Observer), what do you expect of me?

We expect you to:

I'm an observer type at a station, what do you expect from me?

We expect that, when you walk into the dome to make an observation set you:

How do I tell if the sensitivity is correct?

At the correct sensitivity, the needle on the microammeter will be slightly unstable, and you should feel as if the needle movement is the same as the R-dial movement.

How do I tell if an observation is a good one?

(Many of the following situations are now tested by the program to produce a pass or fail situation for the observation.)

First, each individual A, C, C' or D measurement has a standard deviation displayed with it. This number should be small -- a couple of tenths (C' are often larger, as much as a unit). If one of these standard deviations is large (couple of units) then the observer was not completely set up and making the measurement when the computer starting reading the position of the R-dial. Reject this observation and repeat it -- choose the 'operator prompted' method if you find the system prompting is too fast.

Next, if the difference between the highest and lowest A or C numbers is greater than 5 units -- reject it, and try again. If you are doing direct sun observations check the sunspot during the observation to make sure that it's not moving off the GQP during the observation. If you are doing zenith observations, the sensitivity may be too low.

Third, The ozone value should be realistic. Total ozone values in the Northern Hemisphere range from about 200 to 500 Dobson units. Negative values, or greatly lower or higher values mostly likely mean that a different kind of observation was performed than was selected, or the encoder/counter was not zeroed correctly.

Lastly, under most conditions, total ozone does not change greatly from observation to observation, hour to hour, or day to day, especially if the weather is stable. A hundred unit difference between observations is not likely due to natural causes, but to an error in the measurements.

What is expected from the stations in the form of observations?

We expect an observation set in the morning, near local noon, and in the afternoon. Traditionally, the times have been about 10:00, 12:00 and 14:00 Local Standard Time which fits into a schedule at a weather station. However, we actually prefer the morning sets earlier, and the afternoon sets later. Some stations, such as the South Pole station, have special times based on the local situation, or needs of researchers.

What is an observation set?

An observation set is several observations with various observation types. If the sky is clear, please make zenith observations as well as direct sun observations. If the mu value at the time of the set is greater than 2.0 and you are making direct sun observations, make observations with both the ADDSGQP and the CDDSGQP types.

What is "mu"?

Mu (Greek letter: m) is the relative distance through the ozone layer to the sun, e.g., if the sun were directly over head, the length (mu) is one. It is very similar to 'air mass' -- the calculation is almost the same, and the two values are almost identical for values less that 3.0. Mu is calculated from time, using the station's latitude and longitude. Not all observations can be reliably made at all mu values. Some limitations are instrumental; others are in the approximations used in the calculation of ozone from the observations.

What are we actually measuring?

The instrument measures the difference of intensity of wavelength pairs in the near ultraviolet spectrum. The greater the difference, the higher the reading on the R-dial. Two pairs are used in an observation so that the effects of water vapor, dust and other atmospheric gases cancel out of the reduction process. This information is used to calculate total ozone -- the total amount of ozone in the air above the observation point, from the ground to edge of the atmosphere.

Why do we make three measurements on the A (or C) and only two on the D (or C') wavelength setting?

The mean time of the observation is important, and this way of making an observation gives the same mean time for the A (or C) and D(C') measurement sets. The average of the three As (or Cs) and the average of the two Ds (C's) are used in the reduction of the observation to ozone.

I made an observation, and the screen displayed a total ozone amount of 275.6 ± 3.5 Dobson units -- what does this mean?

This number was calculated from the observation. The value 275.6 is the thickness of all the ozone from the ground to the edge of the atmosphere if it was converted to a layer of pure ozone at standard temperature and pressure -- 275.6 means a thickness of 0.2756 cm. The ±3.5 is a calculated uncertainty in that thickness based on the 'noise' in the five individual measurements in the observation.

How are the measurements turned into ozone values?

For the direct sun observations, an equation is used based on the physics of the measurement, using the readings, ozone absorption coefficients, the time and date. For the zenith observations, a statistical method is used, based on a history of comparing direct sun and zenith observations.

What happens to the measurement results?

The results -- the ozone values -- are published in many ways. The most common is in the 'Red Books: Ozone Data for the World. The most common format is a single representative number for a day. Monthly averages are made from these values.

If only a single value is reported for a day, why are we making three sets a day?

First, our organization maintains a database of all the observations for use by researchers. Second, a single number in a day is not useful unless there are several observations with similar results to give us confidence that the value we report is correct. Third, the values from one set give no information as to what happened to the ozone over the course of the day.

Why should I make two observations of the same type during one observation set?

We must evaluate each observation. One of the best ways you can help us to trust the results are to show that they were repeatable.

If we can get good direct sun observations on a particular day, in the morning, about noon, and in the afternoon, why also make zenith observations that day?

Direct sun observations do give the most reliable ozone numbers. However, we are discovering that we need zenith observations, even on the days with clear sun observing conditions. The reduction of the zenith observations is a statistical method based on a comparison of direct sun and zenith observations made close in time. A set of parameters is developed to relate zenith observation results to the ADDSGQP ozone values. We are finding that the parameters are changing as the years roll by. This may be an instrumental change or representative of changes in the ozone profile over the stations. It is especially important for stations that have extended periods of cloudiness during part of the year to continue to make zenith observations in the sunny periods so that we have more confidence in the zenith results. A statistical method is only useful if there are many sets of these 'quasi-simultaneous' observations.

Why are we making observations like this -- surely there's a more modern way?

Don't call me Shirley... And there are other ways, but the resultant ozone values often differ from the Dobson's. There is a Brewer Ozone Spectrophotometer, also named for the principal developer. The Brewer can be fully automated, can make many observations on all types, and can sit outside on a concrete pad. The instrument is expensive ($100K), complicated, and still requires someone to watch over it. There are also small hand-held sun photometers that can measure UV radiation and thus estimate ozone, but the long-term stability of these instruments is questionable. They are also only able to measure ozone using high sun (Mu < 1.5).

Why haven't the satellite measurements replaced these measurements?

The satellite instruments do give wide coverage. Not all of the satellite instruments can see all the way to the ground, and clouds can deceive them. Some instruments have had serious problems with calibration drift, and are not easily repaired. Some have had short useful life spans; others have lasted years past their expected usefulness. All the satellite instruments have had calibration problems.

How does the encoder work?

Inside the encoder is a glass disk with 1000 radial lines scribed on it. The lines are wide enough so that the disk has equal areas of light and dark. Two sets of light emitting diodes and photo-detectors are positioned to look through the disk. As the R-dial is turned, the detectors send out electrical pulses. The counter is set up to count the transitions from dark to light and light to dark. As there are two detectors, there are 4000 transitions. The detectors are positioned so that direction is determined from the phase difference between the pulses from the detectors. The counter counts up or down depending on the phase difference. This system does not know the actual R-dial position at the start; it knows the change from the startup point. The operator has to visually set the R-dial to zero, and then reset the counter by pressing the 'RESET' button and then the 'COUNTER ONE' button while the display is flashing.

Why do ADDSGQP and CDDSGQP give different results?

The results differ because we do not know every thing about the physics of the measurement. The reduction of the observation to ozone uses information obtained in the laboratory, and uses some generalizations about the atmosphere. The actual stratospheric conditions vary station to station and through the year. This causes the calculated results from the ADDSGQP and CDDSGQP observations to differ. At the Mauna Loa Observatory in the tropics, the difference is the smallest. Higher latitude stations tend to have the larger differences.

How is this difference handled?

ADDS and CDDSGQP observations can both be made reliably when mu is between 2.0 and 3.0. Observations made with both types close in time in this region are used to find a multiplying factor for the CDDSGQP (and CDDSFI) observations so that the average from the CD observation types is the same as the AD types. This can only be useful if there are a lot of comparative observations. CDDSGQP and CDDSFI observations are used most often at higher (greater than 40 degrees) latitude stations. Note that we also use the AD -- CD difference as an indicator of instrument 'health' -- a sudden change in the difference says that we should check it out.

I get different results from the direct sun and from the zenith observations in an observation set -- why is this?

The algorithm in the instrument computer must be much simpler than the one we use for the final processing of the data. It was developed at Wallops Island Flight Center and works best there. We are also investigating a different type of algorithm that will be more specific to each station.

What does the Mercury Test test?

The mercury test shows how well the 'Table of Settings of Q' matches the actual wavelength calibration. The table tells you where to set the Q-levers based on the temperature of the instrument. The mercury lamp injects a known signal into the instrument, and the instrument is 'tuned' to receive the signal. If the table matches the position were the signal is detected, then the table is correct. The table is a set of parameters in the instrument's computer, as well as the printed table.

I did a mercury test, and got a difference from the table -- what do I do?

A small difference (less than ±0.3) is within specified limits. If the difference is greater, then repeat the test a day later. If the difference is still greater, then wait another day and repeat the test again. If still greater, change the table -- if the difference was greater than a unit: Call Bob -- 303 497 6679.

What does a standard lamp test test?

The standard lamps have reference values assigned at the last calibration of the Dobson instrument. The test results are compared to the reference values, and the difference is defined as a change in the calibration. It is important that the test be performed carefully, as the results affect the observation results. We do monitor the test results, which are used for the data processing. The lamp results are another indication of the health of the instrument -- a sudden change should be checked out.

If we are sending the daily ozone values to you, why do we have to send you all the files at the end of the month?

We process and publish data in monthly blocks. The main processing program has more information and processes the observations more completely that the program in the instrument's computer -- which can only make ozone estimates from some observations types. We are required to have published ozone data for researchers so many days after the end of the month.

Email changes the data by adding headers, changing some characters, changing the line length or adding blank lines. The main program is very picky about such things.

Why are the monthly standard lamp tests required to be with the data?

A standard lamp test is required on or after the last day of the month for the main processing program. The program uses this test as a calibration check and calculates an adjustment based on the results of the test.

I'm a technician here at the station, what do you expect of me.

We expect you to:

You're asking a lot from us, what should we expect from you guys in Boulder?

You are right, we are asking a lot. There are Dobson Spectrophotometers being operated at five NWS sites, one NASA site, four baseline monitoring sites, and six cooperative (university and other agency) sites. We are also responsible for the calibrations of all the other Dobson instruments in the world. There are three of us -- none is full time permanent. The budget is unknown. We have lost much of our computer services support (data entry, programming assistance, processing, etc) in the last couple of years. We are attempting to 'pick up the slack' by moving some of the work into the computer on the instruments. We are still learning what we can and cannot do with these machines. We are working on a new computer program that will help you with the end of the month tests, and help determine if the observations are acceptable.

You should expect that we: