10.6084/m9.figshare.3382093.v1 Rita S. Brito Rita S. Brito M. Céu Almeida M. Céu Almeida José S. Matos José S. Matos Estimating flow data in urban drainage using partial least squares regression Taylor & Francis Group 2016 Flow measurement data gaps partial least squares wastewater 2016-05-16 10:38:09 Dataset https://tandf.figshare.com/articles/dataset/Estimating_flow_data_in_urban_drainage_using_partial_least_squares_regression/3382093 <p>Flow monitoring in wastewater systems is used for system operation or for billing purposes, among others. Given the difficult measurement conditions, gaps in measurement series occur frequently and stakeholders need an appropriate method to estimate missing data. In data scarcity situations, mathematical modelling of the underlying physical processes may not be feasible and other methods are required. Partial least squares (PLS) regression is a multivariate statistical method suited to correlated data and has been frequently used for water quality estimates. PLS suitability for hourly and daily flow estimations was tested, based on previous flow and precipitation data, which urban water utilities currently monitor. Results were evaluated using proposed performance criteria and classes. The estimation errors were comparable to the ones obtained in physical process modelling. The application of the proposed method for flow estimation in sewers, in two common scenarios of wet and dry weather flows, is presented and discussed.</p>