Application of Regression Models in Estimation of Urban Runoff PoIIution Load

A large amount of rain water is transferred to reception mainstreams (e.g. surface and ground water)
in urban areas due to the increased impermeable surfaces. Urban catchment's water that is produced
by precipitation or snow melting is considered as one of the most important non-point pollutants.
In this investigation, having discretely sampled the output drainage in 13 precipitation events in one
of Isfahan catchments during autumn/winter, 2002/2003, 10 qualitative/quantitative parameters were
measured to assess the general quality. Modulating the drainage pollution amount (using multivariate
statistics) nitrates, total suspended solids, chemical oxygen demand and average discharge factors
were considered as the main regression factors. Modulating input data through which a regression
model of the least error (0.009) and the highest correlation coefficient (0.998) was designated to
assess the amount of output drainage pollution. Input parameters of the model were total suspended
solids and chemical oxygen demand. To assess the model accuracy, error indices were estimated
using two data groups that was not which included in modulating which indicated the qualification of
the model for the next stage. Using this model, total pollution amount of output drainage can be
calculated for future events through measuring two inputparameters. Also, another regression model
was presented based on precipitation characteristics, which can be readily used, but of less accuracy,
These results can be useful for water resource managers, urban programmers and those responsible
for environment to control Zayande Rood River pollution, maintain environmental circumstances
and carry out reasonable urban catchment management.