This study presents the case of the Venice Lagoon Watershed (VLW) which was modeled by means of the physically-based SWAT (i.e. Soil and Water Assessment Tool). Due to the size of the watershed (about 2000 km2) and to the fact that VLW discharges in the Lagoon through several outlets, SWAT model was applied by dividing the VLW area into 8 main sub-basins that were modelled and simulated separately. Although reliable in its predictions, this modeling solution makes quite labour-costing implementing the scenario analysis. To overcome this limit we developed a meta-model that could be effectively used for speeding up the scenario analysis. The meta-model was build on the basis of the SWAT simulations through the training of a MultiLayer Perceptron (MLP) algorithm. The MLP meta-model was successfully trained and showed an overall accuracy higher than 70% both on the training and on the evaluation set. MLP and SWAT predictions were compared for two different scenarios: the “80-90s” scenario considering the VLW situation in the 80-90 decade and a BAT scenario considering the application of the Best Available Technologies applied to all the point sources in the VLW. MLP meta-model was highly reliable and absolutely comparable in its predictions to the SWAT model, allowing a significant simplification in conducting scenario analysis.

Meta-modeling of a physically-based model for predicting the nutrient load discharged to the Venice Lagoon

AZZELLINO, ARIANNA;SALVETTI, ROBERTA;VISMARA, RENATO FRANCESCO
2012

Abstract

This study presents the case of the Venice Lagoon Watershed (VLW) which was modeled by means of the physically-based SWAT (i.e. Soil and Water Assessment Tool). Due to the size of the watershed (about 2000 km2) and to the fact that VLW discharges in the Lagoon through several outlets, SWAT model was applied by dividing the VLW area into 8 main sub-basins that were modelled and simulated separately. Although reliable in its predictions, this modeling solution makes quite labour-costing implementing the scenario analysis. To overcome this limit we developed a meta-model that could be effectively used for speeding up the scenario analysis. The meta-model was build on the basis of the SWAT simulations through the training of a MultiLayer Perceptron (MLP) algorithm. The MLP meta-model was successfully trained and showed an overall accuracy higher than 70% both on the training and on the evaluation set. MLP and SWAT predictions were compared for two different scenarios: the “80-90s” scenario considering the VLW situation in the 80-90 decade and a BAT scenario considering the application of the Best Available Technologies applied to all the point sources in the VLW. MLP meta-model was highly reliable and absolutely comparable in its predictions to the SWAT model, allowing a significant simplification in conducting scenario analysis.
9788890355714
File in questo prodotto:
File Dimensione Formato  
1129 SIDISA Azz.pdf

Accesso riservato

: Post-Print (DRAFT o Author’s Accepted Manuscript-AAM)
Dimensione 284.21 kB
Formato Adobe PDF
284.21 kB Adobe PDF   Visualizza/Apri   Richiedi una copia

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11311/665749
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact