Groundwater Level Dynamics across Temporal Scales for Urban and Rural Wetlands: A Wavelet - Multifractal Model

Wetlands are considered one of the most valuable terrestrial ecosystems because of their multiple functions, including being regulators of biogeochemical cycles. Previous research has demonstrated that New Jersey wetlands located in developed areas experience rapid and frequent wet/dry periods. The objective of this work was to elucidate the role of precipitation on the temporal dynamics of ground water level using a combination of wavelets and multifractal techniques. Four wetlands located in northeastern New Jersey, a densely urban part of the New York metropolitan region, ranging from 16 ha to more than 100 ha of narrow riparian corridor were instrumented with continuously recording wells which collected data at 6 hour intervals. Data from a period spanning 2 months during the period 2003-2005 were selected for this study. Daily rainfall data were obtained from the National Weather Service Cooperative Network from nearby locations. At each site, groundwater levels were characterized with a wavelet-multifractal technique. A multiresolution analysis with bases orthogonal and wavelet Haar was used to study the dynamics of precipitation and groundwater levels. Linear regression was used to find the scale (decomposition) and time lag at which precipitation and groundwater level were best correlated. A linear model was developed and used to predict groundwater level from precipitation data. Deviation between predicted and measured groundwater levels will be discussed in relation to the effect of landscape alteration and wetland size on groundwater dynamics.