Skip to main content
ARS Home » Plains Area » Fort Collins, Colorado » Center for Agricultural Resources Research » Water Management and Systems Research » Research » Research Project #443277

Research Project: A Tool to Facilitate Model Calibration Using Cloud Enabled Multi-group Particle Swarm Optimization

Location: Water Management and Systems Research

Project Number: 3012-13660-010-004-S
Project Type: Non-Assistance Cooperative Agreement

Start Date: Sep 1, 2022
End Date: Dec 31, 2023

The Cooperator will design and test new methods of deploying the MultiGroup Particle Swarm Optimization (MG-PSO) method for calibration of the ARS Ages watershed model. Linking broadly parallel model calibration with a new interface will promote adoption of these cloud computing services and enhance wide-spread applications of Ages to address food and water security. All ARS software developed in this project will be open source and freely available for use within and outside of ARS.

The Agricultural Ecosystem Services (Ages) model is designed to simulate biophysical interactions from farm field to watershed scales. Calibration of Ages is essential for its application to food and water security, because once fully calibrated, Ages can extend experimental research regionally and address “what-if” scenarios. Models like Ages currently take weeks to calibrate, but could be calibrated in hours using cloud computing to expedite research projects and reduce total development costs. Wide-spread deployment of Ages requires improved tools for model calibration, and the proposed model calibration interface is a crucial end-product to support efficient, cloud-based model calibration. Rapid calibration, testing, and deployment of model results will extend core ARS research to other researchers and action agencies. ARS and university collaborators have developed a cloud-service calibration method that integrates Particle Swarm Optimization (PSO) to find optimal values of model parameters. The new approach classifies model parameters by process into multiple groups that run sequentially (stepwise) or in parallel. The MultiGroup-PSO (MG-PSO) method has been implemented as a Python library and uses the Cloud Services Integration Platform (CSIP) developed. Further work is needed to develop the services into a readily accessible tool that is deployed for a wide range of model users. The MG-PSO web services will be demonstrated using Jupyter Notebooks, which may be hosted on JupyterHub and an Application Lifecyle Management (ALM) repositories, for example. Additionally, notebooks and links to the new user interface will be made available on platforms, such as the NSF-sponsored Hydroshare or Community Surface Dynamics Modeling System (CSDMS), to provide broad dissemination of the findings and the tools to a wider modeling community.