Calibration of SWAT models using the cloud

This paper evaluates a recently created Soil and Water Assessment Tool (SWAT) calibration tool built using the Windows Azure Cloud environment and a parallel version of the Dynamically Dimensioned Search (DDS) calibration method modified to run in Azure. The calibration tool was tested for six model...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Environmental modelling & software : with environment data news 2014-12, Vol.62, p.188-196
Hauptverfasser: Ercan, Mehmet B., Goodall, Jonathan L., Castronova, Anthony M., Humphrey, Marty, Beekwilder, Norm
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:This paper evaluates a recently created Soil and Water Assessment Tool (SWAT) calibration tool built using the Windows Azure Cloud environment and a parallel version of the Dynamically Dimensioned Search (DDS) calibration method modified to run in Azure. The calibration tool was tested for six model scenarios constructed for three watersheds of increasing size each for a 2 year and 10 year simulation duration. Results show significant speedup in calibration time and, for up to 64 cores, minimal losses in speedup for all watershed sizes and simulation durations. An empirical relationship is presented for estimating the time needed to calibration a SWAT model using the cloud calibration tool as a function of the number of Hydrologic Response Units (HRUs), time steps, and cores used for the calibration. •A cloud-based, parallel SWAT calibration algorithm is evaluated.•1000 runs are sufficient for flow calibration using up to 256 cores in parallel.•Speedup of the parallel SWAT calibration tool is linear for up to 64 cores.•Cost to calibrate a SWAT model can be estimated from HRU and time step counts.
ISSN:1364-8152
1873-6726
DOI:10.1016/j.envsoft.2014.09.002