Closed jpcurbelo closed 6 months ago
@jameshalgren
Hello @jpcurbelo,
Once you have UA credentials to log into VPN, please email me for Pantarhei account creation. Could you ensure the email body includes your full name and UA email address (if you have one)?
[!NOTE] Please make sure if multiple users need access, then each user will have to send an email from their account.
@jpcurbelo we will need your phone number to put in VPN request. Please Slack us your number.
@jpcurbelo please share your UA email ID when you get chance on slack. thank you.
VPN Access Request Form is submitted today. Thank you.
Hello @jpcurbelo , Your account is created on Pantarhei. Please send an acknowledgment here to confirm you can access the cluster.
Closing this ticket as Jesus is provided access to Pantarhei.
1. Requester Information:
Name: Raymond Spiteri
Institution: University of Saskatchewan
Position: Professor - Computer Science
Email: spiteri@cs.usask.ca
2. Project Information:
This project focuses on benchmarking Long Short-Term Memory (LSTM) models across Canadian and US basins (from CAMELS_spat, under development). Specifically, we concentrate on hydrological modeling tasks related to extreme precipitation and flood events. Leveraging the shared infrastructure, this initiative serves as a platform for executing models efficiently across diverse basins, thereby contributing to the establishment of a benchmarking standard for LSTM models.
Jesus Perez Curbelo, Postdoc at the University of Saskatchewan, will be accessing and using the resources and will work under the supervision of Raymond Spiteri and Martyn Clark. We have collaborative efforts with Chaopeng Shen in this research.
Utilizing the NeuralHydrology library, we plan to replicate published findings from the CAMELS-US dataset. Additionally, we will employ CAMELS_spat for modeling Canadian basins. The NeuralHydrology library, hosted on GitHub, provides the necessary framework for our modeling endeavors.
3. Project Description:
The project involves using and developing LSTM models using virtual environments where we would install the dependencies for the NeuralHydrology library. Also, the software and scripts will be developed in Python including packages such as TensorFlow, PyTorch, and Keras. The data will be stored in
netcdf
andcsv
files.4. Resource Requirements:
5. Timeline:
Starting as soon as possible and continuing until December 31, 2024
6. Security and Compliance Requirements:
No apply
7. Approval:
No apply