摘要
Analysis of hourly underground temperature measurements at a medium-size (by population) US city as a function of depth and extending over 5+ years revealed a positive trend exceeding the rate of regional and global warming by an order of magnitude. Measurements at depths greater than ~2 m are unaffected by daily fluctuations and sense only seasonal variability. A comparable trend also emerged from the surface temperature record of the largest US city (New York). Power spectral analysis of deep and shallow subsurface temperature records showed respectively two kinds of power-law behavior: 1) a quasi-continuum of power amplitudes indicative of Brownian noise, superposed (in the shallow record) by 2) a discrete spectrum of diurnal harmonics attributable to the unequal heat flux between daylight and darkness. Spectral amplitudes of the deepest temperature time series (2.4 m) conformed to a log-hyperbolic distribution. Upon removal of seasonal variability from the temperature record, the resulting spectral amplitudes followed a log-exponential distribution. Dynamical analysis showed that relative amplitudes and phases of temperature records at different depths were in excellent accord with a 1-dimensional heat diffusion model.
Analysis of hourly underground temperature measurements at a medium-size (by population) US city as a function of depth and extending over 5+ years revealed a positive trend exceeding the rate of regional and global warming by an order of magnitude. Measurements at depths greater than ~2 m are unaffected by daily fluctuations and sense only seasonal variability. A comparable trend also emerged from the surface temperature record of the largest US city (New York). Power spectral analysis of deep and shallow subsurface temperature records showed respectively two kinds of power-law behavior: 1) a quasi-continuum of power amplitudes indicative of Brownian noise, superposed (in the shallow record) by 2) a discrete spectrum of diurnal harmonics attributable to the unequal heat flux between daylight and darkness. Spectral amplitudes of the deepest temperature time series (2.4 m) conformed to a log-hyperbolic distribution. Upon removal of seasonal variability from the temperature record, the resulting spectral amplitudes followed a log-exponential distribution. Dynamical analysis showed that relative amplitudes and phases of temperature records at different depths were in excellent accord with a 1-dimensional heat diffusion model.
