Tool for fast assessment of stormwater flood volumes for urban catchment: A machine learning approach.

Specific flood volume is an important criterion for evaluating the performance of sewer networks. Currently, mechanistic models - MCMs (e.g., SWMM) are usually used for its prediction, but they require the collection of detailed information about the characteristics of the catchment and sewer network, which can be difficult to obtain, and the process of model calibration is a complex task. This paper presents a methodology for developing simulators to predict specific flood volume using machine learning methods (DNN - Deep Neural Network, GAM - Generalized Additive Model). The results of Sobol index calculations using the GSA method were used to select the ML model as an alternative to the MCM model. It was shown that the DNN model can be used for flood prediction, for which high agreement was obtained between the results of GSA calculations for rainfall data, catchment and sewer network characteristics, and calibrated SWMM parameters describing land use and sewer retention. Regression relationships (polynomials and exponential functions) were determined between Sobol indices (retention depth of impervious area, correction factor of impervious area, Manning's roughness coefficient of sewers) and sewer network characteristics (unit density of sewers, retention factor - the downstream and upstream of retention ratio) obtaining R2 = 0. 55-0.78. The feasibility of predicting sewer network flooding and modernization with the DNN model using a limited range of input data compared to the SWMM was shown. The developed model can be applied to the management of urban catchments with limited access to data and at the stage of urban planning.

Modeling nutrient removal and energy consumption in an advanced activated sludge system under uncertainty.

Activated sludge models are widely used to simulate, optimize and control performance of wastewater treatment plants (WWTP). For simulation of nutrient removal and energy consumption, kinetic parameters would need to be estimated, which requires an extensive measurement campaign. In this study, a novel methodology is proposed for modeling the performance and energy consumption of a biological nutrient removal activated sludge system under sensitivity and uncertainty. The actual data from the wastewater treatment plant in Slupsk (northern Poland) were used for the analysis. Global sensitivity analysis methods accounting for interactions between kinetic parameters were compared with the local sensitivity approach. An extensive procedure for estimation of kinetic parameters allowed to reduce the computational effort in the uncertainty analysis and improve the reliability of the computational results. Due to high costs of measurement campaigns for model calibration, a modification of the Generalized Likelihood Uncertainty method was applied considering the location of measurement points. The inclusion of nutrient measurements in the aerobic compartment in the uncertainty analysis resulted in percentages of ammonium, nitrate, ortho-phosphate measurements of 81%, 90%, 78%, respectively, in the 95% confidence interval. The additional inclusion of measurements in the anaerobic compartment resulted in an increase in the percentage of ortho-phosphate measurements in the aerobic compartment by 5% in the confidence interval. The developed procedure reduces computational and measurement efforts, while maintaining a high compatibility of the observed data and model predictions. This enables to implement activated sludge models also for the facilities with a limited availability of data.

In Search of a Soil Moisture Content Simulation Model: Mechanistic and Data Mining Approach Based on TDR Method Results.

Soil moisture content simulation models have continuously been an important research objective. In particular, the comparisons of the performance of different model types deserve proper attention. Therefore, the quality of selected physically-based and statistical models was analyzed utilizing the data from the Time Domain Reflectometry technique. An E-Test measurement system was applied with the reflectogram interpreted into soil volumetric moisture content by proper calibration equations. The gathered data facilitated to calibrate the physical model of Deardorff and establish parameters of: support vector machines, multivariate adaptive regression spline, and boosted trees model. The general likelihood uncertainty estimation revealed the sensitivity of individual model parameters. As it was assumed, a simple structure of statistical models was achieved but no direct physical interpretation of their parameters, contrary to a physically-based method. The TDR technique proved useful for the calibration of different soil moisture models and a satisfactory quality for their future exploitation.