You are here:
Spatial and temporal structure within moisture measurements of a stormwater control system
Citation:
Kertesz, R., L. Rhea, AND Dan Murray. Spatial and temporal structure within moisture measurements of a stormwater control system. JOURNAL OF HYDROLOGY. Elsevier Science Ltd, New York, NY, 516:222-230, (2014).
Impact/Purpose:
Green infrastructure, in the field of stormwater control, is a grouping of technologies that when constructed and installed, either in series or parallel, can alleviate the environmental impact of our built environment on the natural surrounding within which urban or even ultraurban areas operate. Major issues lay in determining how well the different green infrastructure control methods perform, either individually, or in concert. This research investigation is the first of a series to explore truly novel, low cost, and straightforward in-situ monitoring devices and techniques. These methods, including proper spatial distribution and number of sensor nodes provides sorely needed information for researchers and land owners alike. This will result not only in better quantification of stormwater control efficiency in the future but also has implications on related fields such as agriculture.
Description:
Moisture sensing is a mature soil research technology commonly applied to agriculture. Such sensors may be appropriated for use in novel stormwater research applications. Knowledge of moisture (with respect to space and time) in infiltration based stormwater control measures (SCM) such as permeable pavement (PP) aids in the understanding of continuous performance parameters such as rainfall-infiltration response, clogging, and storage hydroperiod. Questions remain as to the necessary spatial distribution of sensors to statistically represent a pavement surface area. This study compared transmission line oscillation (TLO) sensor performance in aggregate media (beneath concrete PP) to that in soil (bioretention area) at a 0.64 acre academic parking lot in Cincinnati, OH, USA. Soil sensor readings were found to deviate only slightly from a normal Gaussian distribution while concrete sensors showed extreme non-normality in the measurements. Short (τ2) and long range (σ2) semivariance components were a better estimate of the random process variance for the soil variograms than beneath PP, suggesting poorer statistical confidence in the aggregate data. This was supported by aggregate sensor variograms that showed an estimated practical range (representative distance of a sensor) well beyond the maximum length of the parking lot. Efforts to reduce the influence of sensor noise using a low frequency signal extraction or calculating the product of spatial covariance and temporal auto-correlation were unsuccessful, indicating a complex relationship between spatial and temporal variance. At least sixteen locations (rather than the 9 observed) would be needed to maintain a median signal to noise ratio above 10, which is the minimum value required to estimate the mean to within ±20%. It is apparent that due to the lack of typical soil properties, large aggregate media typical of stormwater retention/detention are not served well by TLO technology in its current form. It may be more appropriate to consider using sensors in higher number to identify PP performance.
