STATISTICAL ANALYSIS OF RISK FACTORS CAUSING SCUM GENERATION IN URBAN ESTUARY

Abstract

 The generation of scum and foul odor is prominent in the tidal area of an urban small river. This study aims to clarify the contribution of the risk factors causing scum generation through statistical analysis of water quality and scum monitoring data in Shakujii estuary. There are several patterns of scum occurrence in temporal variation such as (1) when salinity is 0 and DO is high during spring tide, (2) when salinity is less than 2 and DO is low during neap tide and (3) when salinity is 0 and DO is 0 two days after rainfall. In univariate analysis of plots between scum coverage and water quality parameters, salinity, SSC, Chl-a and rainfall had clear relationship with scum coverage but not for water temperature, DO and water level.

 Based on the result of logistic regression analysis, the significant parameters to scum occurrence are DO, water temperature, water level, salinity, in the order of strongest contribution. When the values of these variables decrease, the risk of scum occurrence increases. Moreover, the best logistic regression model using these risk factors could properly estimate actual scum with accuracy of 80% during calibration while 65% during validation. Therefore, it was found that the aforementioned risk factors can express 70% of scum occurrence condition.