Simple Moving Average Applied to "ISO Method" CPAM Concentration Estimates: An Unexpected Result.

ABSTRACT: A surprisingly large amount of variance reduction has been observed when filtering International Organization for Standardization (ISO) "ISO Method" continuous particulate air monitor (CPAM) airborne radioactivity concentration estimates with a simple three-point moving average. This processing has relatively little lag relative to the amount of variance reduction obtained. The key factor producing this effect is the specific autocorrelation structure of the estimated concentrations, which are based on taking first differences of integrated-count data; this scheme results in successive count differences that contain a common count value between them. The observed variance reduction factor has also been derived analytically.

Some Analysis of Integrated-count Processing for Fixed-filter Continuous Particulate Air Monitors.

A calculation for estimating concentrations of long-lived airborne particulate radioactivity using fixed-filter continuous air monitors is given in an ISO standard. The method uses counts integrated over relatively long time intervals, rather than the 'instantaneous' count rates that in digital systems are evaluated using much shorter time intervals and some form of variance-reduction filtering. This article presents three ways of deriving and interpreting this calculation, based on previously published mathematical models that were derived from first principles. The method is also extended here to apply for short-lived activity. Some statistical properties of the estimator are discussed, including its time-dependent variance and the presence of strong autocorrelation in the concentration estimates. An interactive simulation was used to examine the performance of the concentration estimation, using physically plausible concentration time-dependence profiles; example plots are provided. The conclusion of these studies is that the method, as modified herein, can perform remarkably well in providing periodic average-concentration estimates for both long- and short-lived activity, and it should be considered an appropriate method in those situations where the tracking of a time-dependent concentration is deemed necessary.

Incorrect interpretation of moving-filter continuous particulate air monitor responses.

The graphs supplied by the vendors of moving-filter continuous particulate air monitors (CPAMs) in their sales literature show linear curves on a log-log scale, with net count rate on one axis and concentration on the other. The implication is that the monitor user is to read the concentration from the graph, given an observed net count rate, at any time. For the nominal filter speeds commonly used for these monitors, using the graph in this way is incorrect. The graphs do not state the limitations of the calculation: (1) the nuclide measured must be long-lived; (2) the concentration of that nuclide in the sampled air must remain constant; and (3) the reading of the net count rate must be obtained after a specific time, called the "transit time." This time is typically on the order of several hours. Reading the net count rate at any time earlier than this will result in an incorrect concentration estimate. Given that a major purpose of a CPAM is to alert plant personnel to a change in airborne radioactivity concentrations, by definition when this happens the concentration is not constant. Thus, using the supplied curves will result in an incorrect estimate of that concentration. The solution is to use instead a fixed-filter CPAM and a previously-published quantitative method. With this approach, there is no need to attempt to estimate a concentration, much less to assume that it is constant over long periods of time or that it can only change in a stair-step manner. With this alternative to a moving-filter CPAM, a signal proportional to the time-integrated worker intake can be generated continuously for any time-varying air concentration, including the sums-of-exponentials shapes expected during transient events in compartmental systems.

Estimation of high-level, rapidly-changing concentrations using moving-filter continuous particulate air monitors.

A previously published mathematical model for the dynamic response of moving-filter continuous particulate air monitors has been enhanced to extend that model to include decay chains. During this work, it was observed that a quantitative relationship appeared to exist between the monitor count rate and the time-dependent particulate airborne radioactive material concentration if, and only if, the filter (tape) speed was much faster than the nominal 2.54 cm h(-1) (1 in h(-1)). The extended model demonstrated that operating moving-filter monitors at this nominal filter speed does not provide a quantitative measurement of a changing airborne particulate concentration of a fission product or other contaminant. By contrast, at faster filter speeds [e.g., 76.2 or 152.4 cm h(-1) (30 or 60 in h(-1))], numerical experimentation with this model showed that the count rate trace has essentially the same shape as the concentration profile. It was then found that a quantitative relationship applies, but only when the filter speed is sufficiently fast so that a Taylor series expansion of the monitor count rate can be reasonably well truncated at the first-order term. This mode of operation, which does not require any new monitor hardware, is capable of tracking rapidly changing concentrations. Since the fast filter speed also reduces the monitor's count rate, all else being equal, the approach will best be used for relatively high-level concentrations, such as may occur in abnormal or "accident" conditions. The count rate suppression may also be useful for reducing the detector saturation that can occur with higher levels of airborne particulate radioactivity in post-accident situations.

Degassing Lakes Nyos and Monoun: defusing certain disaster.

Since the catastrophic releases of CO(2) in the 1980s, Lakes Nyos and Monoun in Cameroon experienced CO(2) recharge at alarming rates of up to 80 mol/m(2) per yr. Total gas pressures reached 8.3 and 15.6 bar in Monoun (2003) and Nyos (2001), respectively, resulting in gas saturation levels up to 97%. These natural hazards are distinguished by the potential for mitigation to prevent future disasters. Controlled degassing was initiated at Nyos (2001) and Monoun (2003) amid speculation it could inadvertently destabilize the lakes and trigger another gas burst. Our measurements indicate that water column structure has not been compromised by the degassing and local stability is increasing in the zones of degassing. Furthermore, gas content has been reduced in the lakes approximately 12-14%. However, as gas is removed, the pressure at pipe inlets is reduced, and the removal rate will decrease over time. Based on 12 years of limnological measurements we developed a model of future removal rates and gas inventory, which predicts that in Monoun the current pipe will remove approximately 30% of the gas remaining before the natural gas recharge balances the removal rate. In Nyos the single pipe will remove approximately 25% of the gas remaining by 2015; this slow removal extends the present risk to local populations. More pipes and continued vigilance are required to reduce the risk of repeat disasters. Our model indicates that 75-99% of the gas remaining would be removed by 2010 with two pipes in Monoun and five pipes in Nyos, substantially reducing the risks.

Bias in concentration estimates due to nonconstant flow rate during particulate air sampling.

Correction for a nonconstant flow rate during particulate air sampling is often done by taking the arithmetic average of the initial and final flow rates. This average is then used in concentration calculations as if the flow rate had been constant at that value during the entire sampling period. For long-lived activity this approach is reasonable, but for shorter-lived activity and longer sampling times the estimated concentrations can be biased low. This note examines the magnitude of this bias, and also provides expressions for estimating the concentration, given an observed count, assuming an exponential flow rate time-dependence. One expression uses an estimate of the exponential time-dependence, while another expression uses a linear approximation for the flow rate time-dependence. The parameters used in these expressions are estimated from the same initial and final flow rates used in the arithmetic average.