Diva-on-web and OceanBrowser

class: center, middle

# Diva-on-web and OceanBrowser
Alexander Barth¹, Sylvain Watelet¹, Charles Troupin², Jean-Marie Beckers¹

¹GHER, University of Liège, Belgium

²SOCIB, Spain

![alt text](Fig/eos2.png "EOS")

---

# Diva characteristics

* Diva is written in Fortran (for the computation) and
* Shell Scripts (for managing files, ...)
* Requires a UNIX-like environment
* Uses text files to specify parameters and to activate some options
* Location of the observations and their values are also specified as text files
* Output in NetCDF
* Diva is distributed as source code and binaries (Linux and Cygwin/Windows)

---
# Command-line interface

* Similar to other high-performance code (such as ocean models) and familiar to those users
* Allows batch-processing
* Easily ported to clusters

→ but difficult for new users unfamiliar with the command line

---
# Software components of diva-on-web

* All software components are open source
* Server
 * Python (calling Diva and implementing a Web Map Server)
 * Matplotlib and basemap for graphics
 * nginx webserver as proxy

* Client
 * Openlayers (WMS client)
 * Custom Javascript code

<!--
 Exchanges between client and server

-->
---
# Data upload

* No registration required
* Data uploaded as text file
* For example (longitue, latitude, value):

```
29.76 45.15 16.14
29.76 45.15 16.34
```

* http://ec.oceanbrowser.net/emodnet/diva.html

---
# Analysis grid

* Sample data file: a week of ARGO temperature data
* The location of observations is shown
* Choice of the output grid

---
# Parameters

* Parameters of the analysis
 * correlation length
 * signal-to-noise ratio
* Velocity for advection constrain
* NetCDF file with the u- and v- velocity component
* Regular grid

---
# Correlation length

* Under some assumptions, correlation length can be derived from the data (divafit)

---
# Analysis

* Analysis is shown as a layer in the Openlayers window
* The allowed error variance can be
changed to hide areas with few
observations

---
# Layer selection

* Zooming and panning
* Available layers:
 * Location of obs.
 * Analysis
 * Error mask
 * Mesh
* The size of the mesh is related to the correlation length (very large here)

---
# Download the results

* Results can be downloaded as NetCDF files, Matlab/Octave files and KML files (for e.g. Google Earth)

---
# Link or embed results

* Direct link to the analysis
* Embed the results on your web pages
* Maps are still interactive

---
# Summary

* Diva-on-web is a web interface for Data-Interpolating
Variational Analysis (Diva)
* Should be easy to use
* Allow potential users to try Diva before installing
* Different options to download the analysis are available
* It is also possible to share the results
* http://ec.oceanbrowser.net/emodnet/diva.html

---
# What is OceanBrowser?

* Web-interface to __visualize gridded__ data sets in NetCDF
* Horizontal and vertical sections
* Scalar and vector fields
* Addresses:
 * __SeaDataNet__: http://sdn.oceanbrowser.net/web-vis/
 * __EMODNET Chemistry__: http://ec.oceanbrowser.net/emodnet/
In those projects it is used to visualize gridded data sets generated by the tool DIVA (Data-Interpolating Variational Analysis)

---
# Layer selection

* Simple directory __structure on the server__ is mapped into a hierarchical list of layers
* NetCDF files can be added on-the-fly (without a server restart)
* __Virtual sub-folder__ can be added to show some NetCDF variables more/less prominently

For DIVA field:
* 1st level: analysis masked by an error threshold
* 2nd level: Full field available under “Additional fields”

---
# Horizontal section

The right panel controls the current layer:

* __Select depth and time__
* Plot style
* Metadata
* Download of data product

---
# Vertical section

* Vertical section can be drawn with the mouse
* Data product will be extracted along this section (x-axis: distance from starting point and y-axis depth)
* Section coordinates can be saved (to visualize two parameters along exactly the same section)

---
# Vertical section

* The path of a vertical section can be generated automatically by:
 * fixed distance from coast
 * or fixed ocean depth

---
# Available styles

* Styles
 * Interpolated
 * Filled contours
 * Simple contours
* Colorbar options
 * Colormap
 * Range

---
# Download a data product

<img style="float: right" src="Fig/OceanBrowser_download.png" width="500">
* Download icon (green arrow)
* Data available as NetCDF or OPeNDAP
* Image / Animation export

---
# Export animations

* Winter distribution of phosphate (produced by SMHI)
* Centred 10-year average of all winter months
* OceanBrowser: export of animation by choosing MP4 or WebM animation.

---
# More information

* Info about the NetCDF file
* NetCDF Comments attribute
* Units
* Bounding box
* Time range
* Depth range
* Link to Sextant catalog from Ifremer

---
# Observations

* Web Feature/ Processing Service by Deltares
* Data location (within depth and time range)
* Color shows the number of observations
* Time series and profile plots

---
# Installation

* The hard way: install 13 packages, configure Apache and OceanBrowser and set file permissions
* The easy way:
```bash
docker run -p 8080:80 --name my-oceanbrowser-container \
  -v /some/netcdf/files/:/var/www/data:ro abarth/oceanbrowser
```
 * Docker autamatically downloads and runs OceanBrowser
 * OceanBrowser is available at port 8080 on localhost and servers files in the directory `/some/netcdf/files/`
 * Configuration using environement variables (public URL, name,...)
 * More info at the docker http://registry.hub.docker.com and search for OceanBrowser

---
# Summary

* OceanBrowser allows the visualization of gridded data sets:
 * along a __horizontal section__ (at given time and depth)
 * along a __vertical section__ (e.g. at a fixed distance from coast)
* Download options:
 * Full __NetCDF file__
 * Subset via __OPeNDAP__
 * __Image__ export (PNG, EPS, SVG, ...)
 * __Animation__ export (webm, mp4)
* Installation simplified using docker
* __Open source__ (AGPL) and based on python and matplotlib
* Using on OGC standards (WMS, WFS and NetCDF)