Simulating 21st century boreal forests and fire with a state-of-the-art process-based model

Submission Number: 109
Submission ID: 201
Submission UUID: b9cb6032-a202-4c2d-a723-9ea1fa4aedeb
Submission URI: /form/project

Created: Mon, 09/06/2021 - 12:58
Completed: Mon, 09/06/2021 - 12:58
Changed: Tue, 01/10/2023 - 13:41

Remote IP address: 67.250.115.253
Submitted by: Winslow Hansen
Language: English

Is draft: No
Webform: Project
Simulating 21st century boreal forests and fire with a state-of-the-art process-based model
CAREERS
image.png
biology (515), cluster-support (244), dependencies (218), deployment (451), netcdf (370), performance-tuning (17), r (32)
Finishing Up

Project Leader

Winslow Hansen
406-580-8890
845 677-7600 x138

Project Personnel

Neil McGlohon
Xuejing Wang
{Empty}

Project Information

Background
Temperature in the North American boreal forest is rising 2.5 times faster than the global average. Warming has caused boreal wildfires to become more frequent, larger, and severe than at any point in the last 10,000 years, which is eroding the resilience of forests, causing abrupt ecological change. Fires also increasingly threaten people, including Alaska Native and Canadian First-Nations communities. With climate change, the risk of fire and potential for forest degradation will only accelerate in the future. Advanced predictive modeling to evaluate and prioritize conservation and adaptation strategies hold great potential for mitigating the impact of increased fire by reducing risk, cost, and damage. My team develops advanced simulation models of boreal forests and fire. We use these models to identify leverage points that could alter trajectories toward sustainable social, economic, and ecological outcomes.

Projection description
The purpose of this project is to deploy a state-of-the-art process-based simulation model of forests and the necessary dependencies in an HPC environment at Rensselaer Polytechnic Institute. Tasks will include compiling the model on "bare metal", optimizing and benchmarking performance, and developing a workflow for managing 100s of replicate runs. The key to project success will be replicable workflows that are well documented so a non CS expert can repeat the processes.

Project Information Subsection

1. step by step documentation of the project so a non CS expert can repeat and adapt the processes
1. simulation model and dependencies deployed in RPI HPC environment
2. Input datasets uploaded
3. workflow established for running 100s to 1000s of replication model runs
4. workflow established for analyzing model output in R statistical environment
{Empty}
Undergrad/ grad student with interest in facilitating scientific HPC.
Student must have exceptional written and oral communication skills, especially with scientists lacking CS expertise.
Experience with geospatial analysis in R on an HPC a plus.
{Empty}
Practical applications
{Empty}
Cary Institute
{Empty}
CR-Rensselaer Polytechnic Institute
{Empty}
Yes
Already behind2Start date is flexible
3-6 months
{Empty}
06/08/2022
{Empty}
{Empty}
  • Milestone Title: Unpack iLand from the Enroot container and compile directly on AIMOS.
    Milestone Description: The student will compile iLand and its QT library dependency directly on AIMOS. The student will evaluate whether there are methods for optimizing model runs, and will conduct benchmarks to quantify computing performance.
  • Milestone Title: Optimize performance and benchmark gains
    Milestone Description: The student will evaluate whether there are methods for optimizing model runs, and will conduct benchmarks to quantify computing performance with and without optimization.
  • Milestone Title: Develop a workflow for managing 100s of replicate model runs.
    Milestone Description: Once optimized, the student will develop a bash workflow for managing 100s of replicate runs. This will include submissions and labeling and moving output databases into organized folders for later analysis.
{Empty}
{Empty}
This project will support an NSF funded project to PI Hansen to simulate boreal forests and fire in western North America. We anticipate 4-6 peer reviewed scientific publications from the project.
The student will learn how to deploy bespoke software and dependencies, how to manage large datasets that input to simulation models, and how to setup pipelines for the analysis of big data as outputs from the models. Such skillsets will be highly sought after at National Labs and other research centers that run global climate models and Earth System Models.
{Empty}
{Empty}
HPC resources have already been secured from Rensselaer Polytechnic Institute.
This project submission was encouraged by RPI CCI Director Dr. Chris Carothers.

Final Report

{Empty}
{Empty}
{Empty}
{Empty}
{Empty}
{Empty}
{Empty}
{Empty}
{Empty}
{Empty}