Answer to Question #8910 Submitted to "Ask the Experts"

I am joining a survey monitoring radiation levels on the Columbia River near the Hanford Reservation in the state of Washington. I am new to this type of research. I currently have an Inspector Digital pocket Geiger counter. Its current default setting for alarm is 100 CPM or 0.1 mR/h. I don't foresee getting into any dangerous hot zones but would greatly appreciate your advice for where the alert should be set in the mR/hr - µSv/hr settings. I received a varied response to this question and would appreciate your input and advice (specifications at http://www.geigercounters.com/SpecInsp.htm).

Where the alarm point is set depends, at least in part, on what the intention is on the part of the surveyor for having the instrument alarm. For example, is the intent of the alarm to alert the surveyor that the reading is statistically above background so that careful or additional measurements for documentation and analysis purposes may be made, or is the alarm intended to warn the surveyor that the radiation level is sufficiently high to warrant concerns from a health and safety standpoint or that the level is approaching or exceeding some other administrative action level? The set points in these different cases may be dramatically different from one another. In order to judge the possible significance of the current alarm set point, we should first look at the level that you say has been set as the default value.

The specifications for the Inspector GM, as shown in the Web page that you referenced (http://www.geigercounters.com/SpecInsp.htm), show an expected response of 3,500 cpm/mR/hr, which is typical for a pancake GM of this sort. From this we would infer that a count rate of 100 cpm would correspond to an exposure rate of (100 cpm)(1 mR/hr/3,500 cpm) = 0.0286 mR/hr or 28.6 µR/hr. This does not jive with the 0.1 mR/hr that you include in your question; the 0.1 mR/h would correspond to a count rate of 350 cpm. If we assume that the 100 cpm set point is the correct one, we can do some simple assessments to judge its significance in terms of exposure rate and general relevance and applicability for monitoring.

External gamma radiation background levels may vary with location and with time at a fixed location, but values in the range from about 5 to 20 µR/hr would cover most locations in the United States. In the Columbia River region around Hanford, a typical value seems to be about 0.22 mR/day or 9.2 µR/hr. This is based on observation of data from a 2007 report of Washington State Department of Health, Division of Environmental Health. The setting of 100 cpm (28.6 µR/hr) then corresponds to about three times the expected normal background level. This value is certainly not of any health concern. For many instances, the setting of three times the normal background level represents a reasonable level at which to set the alarm point for routine monitoring.

If the intention is to have the detector alarm at any radiation level statistically distinct from the normal background, the alarm point could likely be reduced. The idea then is to identify a level that meets acceptance criteria without making the level so low that false alarms are unacceptably frequent. Different criteria may be used to establish such a level; one common technique is to set the alarm point at about three to five standard deviations above the mean normal background rate.

The instrument you are using is a digital meter and, according to the technical specifications, the display is updated every three seconds (i.e., the count integration time is three seconds). In addition, when the exposure rate is low, as for typical background, the meter displays the moving average exposure rate over the last 30 seconds.

Using this information, we can assess the uncertainty in the background. At an expected exposure rate of 9.2 µR/hr, the resultant average count rate would be (3,500 cpm/1,000 µR/hr)( 9.2 µR/hr) = 32.2 cpm = 0.537 cps. For a 30-second period (averaging time) the expected integral count would be (30 s)(0.537 cps) = 16 counts.

The Poisson standard deviation in this count is the square root of the count or 4 counts. In terms of rate, these numbers translate to 32.2 cpm with a standard deviation of 8.0 cpm. If we assume that background statistical variations are consistent with a normal distribution, the specification of an alarm point at three standard deviations above background would imply setting the alarm point at 32.2 cpm+ 3(8 cpm) = 56 cpm. For a constant background, with the meter exhibiting normal statistical variation, setting the alarm point at three standard deviations above the mean would result in an expected false alarm rate of 1 to 2 per thousand sequential readings. Since each reading represents a three-second update, the expected elapsed time to generate 1,000 consecutive displays is 3,000 seconds, or 50 minutes.

Many people might find such a false-alarm rate acceptable. In actuality, however, you may find that there are added variations that contribute to greater dispersion among readings. These include such factors as variations in how the surveyor operates (e.g., the orientation of the meter, its proximity to surfaces being monitored, how long the instrument is held in a fixed location, and how rapidly the surveyor moves with the instrument), possibly added variations associated with the electronic characteristics of the instrument, and real variations in the natural background rate.

Such factors may contribute to a greater false-alarm frequency. In such a case, increasing the set point to five standard deviations above the mean may be desirable. Such a set point would correspond to 72 cpm. This is only about 25 percent less than the value of 100 cpm that you refer to, thus providing some additional validation for the default value. Naturally, you must abide by operational performance requirements that may require that the instrument must be capable of measuring certain specified levels above background.

There are other considerations that may apply in performing extensive surveys and selecting and using specific instruments. If you are expecting to be involved in this area, you may want to review other materials and recommendations. In particular, I would recommend that you review the very popular and useful MARSSIM document, Multi-Agency Radiation Survey and Site Investigation Manual.  

If your major concern in setting the alarm is to alert you that the radiation levels may be increasing to a point where you should take special precautions, such as modifying your survey pattern or speed, or evacuating the area, it would be common that some administrative action levels would apply. These would depend on the specific requirements of the licensee or operator of the contaminated site. If such directives are not forthcoming then I would recommend setting the alarm point at about 10 times the expected normal background exposure rate, which, in this case, would correspond to 92 µR/hr, which you might round up to 100 µR/hr, one of the values you specified in your question (0.1 mR/hr). Such a value is high enough above typical background that you can be quite certain that it is indicative of the presence of significant contamination, especially if the survey meter is being held a few feet above the earth. Such areas will likely require additional attention to define the source of the elevated levels and possible remediation.

If a reading of 0.1 mR/hr at a nominal 100 cm above the earth is associated with a small dimension source near the surface of the earth, the exposure rate near the surface may approach 1,000 mR/h or possibly more. Such levels are not an immediate health hazard, but one would want to be aware of them so as to avoid spending more time than necessary in the immediate area. A reading of 0.1 mR/hr at the same height may not translate to a much higher exposure rate at the ground surface if the contamination is widely distributed over a significant area, although this could translate to a larger overall contamination problem.

In conclusion, we can say that there are various influencing factors that may affect what level one might select as an alarm set point, and there are various rationales that can be used in determining what value is appropriate. I hope the discussion here is helpful to you.

Good luck in your survey work.

George Chabot, PhD, CHP