Skate Web

This web application is the interactive component of the Skate project. For the analysis component, see https://github.com/retrievertech/seismogram-pipeline.

The Skate web app allows browsing a large collection of analog WWSSN seismograms, overlaid with digital trace data that was extracted using our analysis component.

Skate is currently deployed at https://seismo.redfish.com

NOTE: The master branch is currently deployed to https://seismo.redfish.com, running Ubuntu 22.04, node 14.16.1, mongod 5.0.9. See https://github.com/retrievertech/seismogram-app/tree/node-16-mongo-8 for a version of the code that works with node v16.20.0 and mongod v8.0.1.

Local Development

Start mongo:

mongod --dbpath <some directory>

data-tools

Tools to sanitize/import the station and file data. The data is in the repo in data-tools/data. From data-tools/, run:

npm install
node jsonize.js

This will produce files.json and stations.json. Then you can import data into mongo:

node mongo-import.js dev

If you have AWS credentials for the seismogram metadata bucket, you can import the latest edited metadata into your local mongo.

Set up an aws cli profile called seismo with credentials to read the s3 bucket:

aws configure --profile seismo

And run the same mongo-import script without the dev flag:

node mongo-import.js

server

A node server atop mongo.

cd server
npm install
cp config.example.js config.js
node server.js # starts on port 3000

client

An angular frontend.

bower install

Serve the client directory using any simple web server (e.g. https://simplewebserver.org/) and load it in the browser, assuming the above steps are done.

image processing

The image processing scripts are tracked in a separate repo: https://github.com/retrievertech/seismogram-pipeline. The web app will run and allow you to browse seismograms without setting up these image processing scripts. Only if you want to use the web app to analyze and trace seismograms do you need to set them up.

Clone https://github.com/retrievertech/seismogram-pipeline and put the directory next to seismogram-app/ on your computer. The server expects seismogram-pipeline/ and seismogram-app/ to live as siblings in the same directory on your filesystem.

For the rest of the image processing setup (e.g. conda and python dependencies), follow the "setup processing pipeline" steps at the bottom of the Cloud Development section.

Cloud Development

EC2

On our EC2 instance, the mongodb config file is located at /etc/mongod.conf.

The server listens on port 3010 and sits behind an nginx proxy that listens on port 3000. nginx implements SSL and serves the static web app content on ports 80 and 443. The nginx config is located at /etc/nginx/sites-enabled/seismogram.

Restart the server with pm2 restart all, or have a look at deploy.sh.

Spinning up a new deployment

nginx

sudo apt install nginx

nvm

nvm install 14.16

pm2 + bower

npm install -g pm2 bower

conda

mongo

aws

aws configure --profile seismo

clone seismo repos

git clone https://github.com/retrievertech/seismogram-app.git
git clone https://github.com/retrievertech/seismogram-pipeline.git

setup db

cd ~/seismogram-app/data-tools
npm install
node jsonize.js
node mongo-import.js

setup server

cd ~/seismogram-app/server
npm install
cp config.server.example.js config.js
pm2 start pm2.ecosystem.config.js
pm2 startup
pm2 save

setup client

cd ~/seismogram-app/client
bower install
cd ~/seismogram-app
sudo cp nginx.conf /etc/nginx/sites-available/seismogram
cd /etc/nginx/sites-enabled/
sudo ln -s ../sites-available/seismogram .
sudo chmod +x $HOME
sudo chown -R :www-data ~/seismogram-app/client
sudo systemctl restart nginx

setup processing pipeline

cd ~/seismogram-pipeline
conda create --name seismo python=2.7 numpy scipy
conda activate seismo
# had some issues installing geojson and docopt w/ conda
python -m pip install --upgrade pip
pip install geojson docopt matplotlib

setup HTTPS w/ certbot