Probability Distributions Used in 100-Year Return Period of Air-Freezing Index

Abstract

The air-freezing index (AFI) is a measure of the combined magnitude and duration of air temperatures above and below freezing for a winter season. The 100-y return period of the AFI has been found to be an indicator of the amount of insulation required to protect a building foundation from frost heave and has been used to develop frost-protected shallow foundation (FPSF) design criteria. FPSF has recently been accepted as a change to U.S. building codes with resulting annual construction savings estimated at $300 million nationally for new residential homes. Previous work has found that different probability distributions can produce significantly different estimates of the 100-y return period of the AFI and thus the amount of insulation required in FPSF. To determine which of several probability distributions best fit the AFI sample data, a goodness-of-fit test and graphical analyses have been applied to locations which have long-term and high-quality climate records. Results indicate that the Weibull probability distribution is the best choice for estimating 100-y return periods of the AFI for all U.S. climate regimes.