Extracting the Diurnal Component from Wet Weather Flows Using Time Series Analysis

摘要

The diurnal component of wet weather flow arises from the day-to-day use of a sanitary system. Often times during rain events flow data has additional flow components present in the time series data, similar to that of combined sanitary systems. The additional wet weather flow components present in sanitary systems are the result of inflow and infiltration. The ability to accurately model and extract the diurnal component of a flow signal is important to accurately study these systems. Most notably, when attempting to analyze and model inflow and infiltration effects within a sanitary system, nuances in long-term infiltration dynamics and rapid peaking during large rain events can be lost or obscured if the diurnal is extracted improperly. Typically, the diurnal component of wet weather flows is estimated using spatial and temporal physical parameters such as housing distributions and their connections to the infrastructure, distribution of populations and water billing records. Frequently, average daily patterns (or possibly average weekly patterns) are used directly as a model of the diurnal. This requires one parameter per flow sample over the duration of interest. For example, in the case of hourly data with a weekly pattern, this leads to 168 parameters in order to specify the diurnal pattern. This methodology unduly incorporates specific errors present in the flow measurements, as well as any baseflow fluctuations, into the diurnal model. This over-fitting of data can potentially lead to an inaccurate estimate of the diurnal pattern that in turn obscures important information in the inflow and infiltration components of the flow signal.