COMPARISON OF SAMPLING AND ANALYSIS PROCEDURES FOR CLASSIFYING THE SOIL THERMAL REGIME IN CANADA¹

Abstract

Daily soil temperature records at the 0.5-m depth (a.m. reading) under short grass cover from six locations across Canada representing soil temperature classes from cold Cryoboreal to mild to moderately warm Mesic were used to assess the effect of sampling frequency as well as interpolation and analysis procedures on estimated soil climate parameters. Sampling frequencies ranging from four equally spaced to 13 “random” samples per year for two 3-yr periods at each location were compared by correlating estimated parameters with “normal” values and by examining means and standard deviations of residual values and coefficients of variation.

Primary soil climate classifiers (mean annual soil temperature (MAST) and mean summer soil temperature (MSST)) were accurately calculated by using both a monthly sine wave interpolation procedure and a Fourier series function with a second harmonic fit to “normals” data. However, the monthly sine wave interpolation procedure provided more reliable estimates of secondary soil climate classifiers (length of season and accumulated degree-days above 5°C and 15°C) due to lack of fit of the Fourier equation at locations where winter snow cover dampens out the annual soil temperature curve.

MAST was estimated to within ±0.5°C of normal values by simply averaging four equally spaced samples per year for a 3-yr period. MSST was estimated within ± 0.8°C by sampling on the 15th day of June, July, and August only. Both estimates were considered sufficiently accurate for the primary classifiers of the soil climate classification system being used in Canada. Sampling 12 times per yr at midmonths or 13 random samples per year for 3 yr resulted in larger residuals in some secondary soil climate classifiers than desirable for accurate classification of soil climate subclasses, indicating that more frequent and/or longer sampling may be required.