Bias of stormwater infiltration measurement methods evaluated using numerical experiments

Infiltration stormwater control measures (SCMs) have the potential to contribute towards mitigating the effects of urbanization on downstream receiving waters. Infiltration SCMs are most often successful when the in-situ saturated hydraulic conductivity (K_sat) is well characterized. In this paper numerical solutions of the Richards’ equation are used to quantify the bias of seven infiltration measurement methods, removing natural variability and random error from the analysis. The methods evaluated in this study include the double ring infiltrometer, Saturo infiltrometer, modified Philip–Dunne infiltrometer, Turf-Tec IN2-W infiltrometer, USBR 7300-89 well permeameter, Philip–Dunne permeameter, and the Guelph permeameter. Seven homogenous, isotropic soil textures were simulated at four initial soil moistures for the seven methods, resulting in a total of 196 simulations. The dimensionless bias is defined as the “measured” K_sat determined by a given method divided by the K_sat input to the numerical experiments. The “measured” K_sat is in quotations to identify the measurement occurs in a numeric experiment rather than in a physical experiment. In sand through silt loam soils that are typical of infiltration SCMs, the simulated methods have a bias in the range of 0.7–6.2. The Turf-Tec was the only infiltrometer that produced a bias >2.5 for these soils. Initial effective saturation had a minimal contribution to bias for most methods. Methods that rely on a one-dimensional (1D) flow assumption consistently overestimated the K_sat. Borehole methods produced results with bias similar to surface methods. Long duration methods did not consistently produce more accurate results than short duration methods.