BEGIN:VCALENDAR VERSION:2.0 X-WR-CALNAME:EventsCalendar PRODID:-//hacksw/handcal//NONSGML v1.0//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York LAST-MODIFIED:20240422T053451Z TZURL:https://www.tzurl.org/zoneinfo-outlook/America/New_York X-LIC-LOCATION:America/New_York BEGIN:DAYLIGHT TZNAME:EDT TZOFFSETFROM:-0500 TZOFFSETTO:-0400 DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZNAME:EST TZOFFSETFROM:-0400 TZOFFSETTO:-0500 DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT CATEGORIES:College of Engineering,Graduate Studies,Lectures and Seminars,Th esis/Dissertations DESCRIPTION:Faculty Supervisor: Dr. Gokhan Kul, Computer & Information Scie nce/Data Science听 Committee Members: Dr. Gavin Fay, SMAST / Fisheries Oce anographyDr. Ashokkumar Patel, Computer & Information Science/Data Science 听Dr. Asif Turzo, Computer & Information Science/Data Science Abstract: Pr ocess based marine ecosystem models such as Atlantis Provide high fidelity simulations of species interactions and environmental dynamics, but their computational cost limits their use in real time forecasting and large sc ale scenario analysis. This study proposes a deep learning based surrogate modeling framework to emulate Atlantis simulations and enable efficient p rediction of marine framework to emulate Atlantis simulations and enable e fficient prediction of marine biomass dynamics in the Northeast U.S. Large Marine Ecosystem. Unlike traditional approximation approaches, the propos ed method focuses on learning spatio temporal dependencies directly from s imulation outputs. Atlantis data spanning 1964 - 2020 across multiple guil ds and spatial polygons are structured into temporal sequences incorporati ng environmental drivers such as temperature and salinity. A Bidirectional Long Short Term Memory (Bi-LSTM) architecture is employed to capture both forward and backward temporal relationships and model nonlinear ecosystem behavior. The surrogate achieves strong predictive performance, with an R 2 score of 0.90 on held out test data, while maintaining consistency acros s ecological groupings. In addition to accuracy, the model significantly r educes computational overhead compared to full Atlantis simulations, enabl ing rapid multi step forecasting and scalable exploration of management sc enarios. These results demonstrate that deep sequence models can serve as effective surrogates for complex ecological simulators, providing a practi cal pathway toward real time, data driven decision support in ecosystem ba sed fisheries management. 听 For further information please contact Dr. Go khan Kul at gkul@umassd.edu.\nEvent page: /events/cm s/surrogate-modeling-of-atlantis-deep-learning-approaches-for-predicting-a nd-forecasting-biomass.php\nEvent link: https://teams.microsoft.com/meet/2 7173518080619?p=89Z8ItRNZFdiQC8x2x X-ALT-DESC;FMTTYPE=text/html:

糖心logo入口

Faculty Supervisor:
Dr. G okhan Kul\, Computer & Information Science/Data Science听

Committe e Members:
Dr. Gavin Fay\, SMAST / Fisheries Oceanography
Dr. A shokkumar Patel\, Computer & Information Science/Data Science听
Dr. A sif Turzo\, Computer & Information Science/Data Science

Abstract:< /p>\n

Process based marine ecosystem models such as Atlantis Provide hig h fidelity simulations of species interactions and environmental dynamics\ , but their computational cost limits their use in real time forecasting a nd large scale scenario analysis. This study proposes a deep learning base d surrogate modeling framework to emulate Atlantis simulations and enable efficient prediction of marine framework to emulate Atlantis simulations a nd enable efficient prediction of marine biomass dynamics in the Northeast U.S. Large Marine Ecosystem. Unlike traditional approximation approaches\ , the proposed method focuses on learning spatio temporal dependencies dir ectly from simulation outputs. Atlantis data spanning 1964 - 2020 across m ultiple guilds and spatial polygons are structured into temporal sequences incorporating environmental drivers such as temperature and salinity. A B idirectional Long Short Term Memory (Bi-LSTM) architecture is employed to capture both forward and backward temporal relationships and model nonline ar ecosystem behavior. The surrogate achieves strong predictive performanc e\, with an R2 score of 0.90 on held out test data\, while maintaining con sistency across ecological groupings. In addition to accuracy\, the model significantly reduces computational overhead compared to full Atlantis sim ulations\, enabling rapid multi step forecasting and scalable exploration of management scenarios.

These results demonstrate that deep seq uence models can serve as effective surrogates for complex ecological simu lators\, providing a practical pathway toward real time\, data driven deci sion support in ecosystem based fisheries management.

听

For further information please contact Dr. Gokhan Kul at gkul@umass d.edu.

Event page: /events/cms/surrogate-modeling -of-atlantis-deep-learning-approaches-for-predicting-and-forecasting-bioma ss.php
Event link:

DTSTAMP:20260425T131933 DTSTART;TZID=America/New_York:20260508T133000 DTEND;TZID=America/New_York:20260508T143000 LOCATION:Online SUMMARY;LANGUAGE=en-us:Surrogate Modeling of Atlantis: Deep Learning Approa ches for Predicting and Forecasting Biomass UID:443e83fb4a1ac633ee74dd5e42032388@www.umassd.edu END:VEVENT END:VCALENDAR