Virginia Flood Information System (VFIS) - A Cloud-Based Real-time Decision support Modeling System for Flood Impact Warning

This repository contains code for a prototype cloud based flood warning system. This system provide warnings for flooded transportation infrastructure in coastal virginia, specifically, Hampton Road District. The system contains 3 components: A windows server to run the model, a linux webserver to visualize the results in a browser, and a storage bucket for data archival.

The Master branch represents the project phase I. The implementation of this phase can be accessed through this link http://vfis-aws.uvahydroinformatics.org/

Currently, we are working on phase II of this project, GitHub branch "phaseII." This phase can be accessed through this link http://vfis.uvahydroinformatics.org/

Deployment

There are three main parts to our cloud based workflow: A visualization instance, a model instance, and a storage bucket. Below are instructions for reproducing the model and visualization instances.

Model Instance Creation

The model instance is a windows machine with a powerful GPU to run the tuflow model. These instructions assume that the TUFLOW Model has been installed and is running. The source code required for the model instance is located in the Model Instance folder.

Along with the model the instance must also have a cloud CLI. If using AWS the instructions to install and configure it are located here.

Python and the dependencies in requirements.txt must be installed. The batch file must also be edited to point to the correct file paths and cloud resources.

Lastly the batch file is triggered over SSH which is not yet standard on Windows Server Machines. SSH can be used on windows machines by:

• Open Power shell and install chocolatey package manager

  Set-ExecutionPolicy Bypass
  iex ((New-Object System.Net.WebClient).DownloadString('https://chocolatey.org/install.ps1'))```

• Install the openSSH Server with this command

  choco install openssh -params '"/SSHServerFeature"'

Visualization Instance Creation

The visualization displays all of the model run results to the end user from a web browser. It runs on a linux webserver. Clone the server folder onto your webserver. Install the python requirements with

pip install -r requirements_Dec2017.txt

To run the python flask webserver program we use two other pieces of software: NGINX and pm2. NGINX handles all web requests from the users. PM2 monitors our pyhton process and keeps it running in the background with logs. To install NGINX:

•   sudo apt-get update 
    sudo apt-get install -y nginx

• sudo /etc/init.d/nginx start
  sudo rm /etc/nginx/sites-enabled/default
  sudo touch /etc/nginx/sites-available/app
  sudo ln -s /etc/nginx/sites-available/app /etc/nginx/sites-enabled/app
  sudo vim /etc/nginx/sites-available/app

  and the copy and paste the following configuration into the new file

  server {
      location / {
          proxy_pass http://localhost:8000;
          proxy_set_header Host $host;
          proxy_set_header X-Real-IP $remote_addr;
      }
      location /static {
          alias  /home/ubuntu/server/static/;
      }
  }

• Restart NGINX

  sudo /etc/init.d/nginx restart

To install pm2:

• First install nodejs using the correct instructions for your system.
• Run

   npm install pm2 -g

• Then run the following command to start the python server.

  pm2 start server/server.py

Powershell Script

Aws.ps1 is a proof of concept script to be run on a users local machine to either trigger the model run from the cloud or run the model locally. It will also install all necessary dependecies for both scenarios.

License

This project is licensed under the MIT License - see the LICENSE.md file for details