• Home
• Products
  • Microbiological Air Sampling
  • Air Sampling
• Services
  • Food Processing Facility Filtration
  • Cleanroom Air Filtration
  • Case Study — Cebro Frozen Foods
• Technical Data
  • Microbiological Air Sampling Indicators
  • Sampling Science
  • Settling Plates (Petrifilm) vs. MB-2
  • Utility of Spatial Analysis in Air Sampling Results
• About Us
  • Contact Us
  • Legal Notice
  • Distributors
• Login

Utility of Spatial Analysis in Air Sampling Results

Before Treatment Air with >300 CFU/cubic meter is not acceptable and needs corrective action. These areas are colored red.

After Treatment Air with <100 CFU/cubic meter (m3) is considered clean and acceptable. These areas are colored blue.

Background

Recommendations have been made that food processing plants adopt air monitoring programs to detect possible contamination sources in their regular HACCP program. Spatial analysis by color contour mapping of air sample data is very useful to display these data and aid in the assessing the effectiveness of air treatment in food facilities.

Summary

The utility of colored spatial contour maps provide excellent visuals that show technical and plant management the extent and location of problematic air in their facility. Air quality color scale for food processing environments should be:

• Air with <100 CFU/cubic meter (colony forming units per cubic meter, or m3) is considered clean and acceptable, these areas would be colored blue.
• Air with 100 to 300 CFU/cubic meter is marginal, these areas would be colored a blend of blue to red.
• Air with >300 CFU/cubic meter is not acceptable and needs corrective action, these areas would be colored red.

See the attached spatial map.

Background

Spatial maps are a three-dimensional analysis in which two dimensions (X and Y) represent the positions of points on a horizontal plane, and the third (Z) represents data associated with the points. The data can be visualized as elevations above or below the plane.

Data may consist of:

1. Measured numerical values, colony forming units (CFU/m3) air sampling for mold, temperature, moisture content, etc. or derived numerical values (differences, ratios, probabilities, etc.) associated with fixed points on the horizontal plane and 
2. The X, Y-coordinates of the fixed points.

All variables are treated as continuous although CFU/m3 are discrete. The points are fitted to a three dimensional surface representing the variable (Z-axis) as a function of position on the X, Y-plane. 

The surface can be represented in two dimensions by a contour map, which shows the configuration of the surface by means of isolines (contours) drawn at regular intervals of Z, on the X, Y plane through the origin. Each contour represents the intersection of a horizontal plane with the surface, which means that all points on a contour have the same Z-value.

Observed Z-values are posted to coordinates on a map of a facility, (packaging room, warehouse, retail store, etc.), entered as a base map and then creates a denser grid of Z-values by interpolation, using one of several available algorithms. Normally radial basis functions with multiquadric algorithm is used, which is flexible and provides a good overall interpretation of most data sets. After interpolating, contours of Z are produced, based on the interpolated grid values. Goodness of fit is estimated by calculating residuals -differences between observed numbers and numbers predicted by the fitted surface.

Residuals and other derived values, such as differences and probabilities, can also be mapped by means of contour analysis. Spatial changes are best examined by grid subtraction,  which is done by subtracting the value of each node in a grid from the corresponding Spatial changes are best examined by grid subtraction, which is done by subtracting the value of each node in a grid from the corresponding node values of a second, identical grid; the differences are then assigned to corresponding nodes of a third grid and difference contours drawn. Areas of increase are indicated on difference plots by positive contours and areas of decline by negative contours.

In viewing areas for microbial levels, it is helpful to draw contours of indicator variables, rather than raw CFU/m3 counts, because indicator variables are affected less by unusually large (or small) counts. Indicator variables are obtained by converting observed CFU/m3 to probability. Sampling locations are first sorted in descending order by CFU/m3, then cumulative and normalized CFU/m3 are calculated for each location.

The normalized CFU/m3 (cumulative CFU/m3 divided by the total number of CFU/m3 in all locations) gives the probability of an equal or higher CFU/m3 count. An indicator value of 1 is assigned to all locations for which probability equals or exceeds some set level or action threshold (determined by experience and the needs of industry) and a value of 0 is assigned to the remainder.

References

1. Campbell, J. E., S. B. Franklin, D.J. Gibson, and J. A. Newman. 1998. Permutation of two-term local quadrat variance analysis: general concepts for interpretation of peaks. J. of Vegetation Science. 9:41-44.
2. Clark, I. 1979. Practical Geostatistics. Elsevier Applied Science Publishers, London, UK.
3. Cressie, N. 1985. Fitting variogram models by weighted least squares. Mathematical Geology 17: 563-586.
4. Cressie, N. A. C. 1991. Statistics for Spatial Data. John Wiley, New York, USA.
5. David, M. 1977. Geostatistical Ore Reserve Estimation. Elsevier, Scientific Publishing Co., Amsterdam, The Netherlands.
6. Deutsch, C.V. and A.G. Journel 1992. GSLIB Geostatistical Software Library. Oxford University Press, New York.
7. Goovaerts, P. 1997. Geostatistics for Natural Resources Evaluation. Oxford University Press, New York.
8. Griffith, D.A. 1987. Spatial Autocorrelation: A Primer. Association of American Geographers, Washington, D.C. 86 p.
9. Deutsch, C. V. and Journel A. G. 1992. GSLIB Geostatistical Software Library and User's Guide. Oxford University Press New York.
10. Dufrene, M. and P. Legendre. 1997. Species assemblages and indicator species: The need for a flexible asymmetrical approach. Ecological Monographs 67:345-366.
11. Issaks, E. H. and Srivastava R. M. An Introduction to Applied Geostatistics. Oxford University Press, Oxford, pg 561. 1989
12. Lecoustre, R., Fargette, D., Fauquet, C., and de Reffye, P. 1989. Analysis and mapping of the spatial spread of African cassava mosaic virus using geostatistics and the kriging technique. Phytopathology 79:913-920.
13. Mandelbrot, B.B. 1982. The Fractal Geometry of Nature. W.H. Free man, London
14. Myers, D. E. 1991. Interpolation and estimation with spatially located data. Chemometrics and Intelligent Laboratory Systems 11:209-228.
15. Myers, D. E. 1989. To be or not to be...stationary: That is the question. Math. Geology 21: 347-362
16. Rossi, R. E., Borth, P. W., Tollefson, J. J. 1993. Stochastic simulations for characterizing ecological spatial patterns and appraising risk. Ecol. Appl. 3:719-735.
17. Rossi, R. E.,  Mulla, D. J., Journel, A. G. and Franz, E. H. Geostatistical Tools for Modeling and Interpreting Ecological Spatial Dependence.  Ecology Monographs, 62, 277-314. 1992.
18. Wartenberg, D., F. Scott, and F.J. Rohlf. 1987. Putting things in order: a critique of detrended correspondence analysis. The American Naturalist 129:434-448.