Desired features and required classes

[heistermann]

This issue starts a first discussion about desired features and required classes. Based on this discussion, new issues can be created in order to discuss specific aspects in more detail, and priorities can be set. It is clear that we should start small, but it could help to see where it could go in the mid and long term. For example, the motivation is to view RADOLAN composites. But knowing that we'd like to extent wradvis to other Cartesian and also polar grids will help us to design more generic classes in the very beginning.

Desired features

• visualise cartesian grids of radar-related variables
• visualise polar grids of radar-related variables (see here)
• provide a high-level and platform-independent GUI that does not require scripting
• interactive zooming, panning etc. of figures
• consider radar data as time series in order to create e.g. accumulations or animations
• visual overlay of point data (e.g. rain gauge observations)
• visual overlay of line features
• visual overlay of polygon features
• interactive inspection of radar and point data (e.g. click on a bin, get a time series)
• interactive control of time windows and intervals
• interactive control of spatial bounding boxes
• zonal statistics for points and polygons, and their visualisation as time series
• unit testing and Travis-CI
• web-based documentation and tutorials based on Sphinx/ReST
• context help
• save and/or preset your configuration in a configuration file (i.e. project file)
• logging and output in several levels (user, warning, devel, debug etc)
• control thresholds in listed colorbars
• export time series and images (formats?)
• ability to browse through nested zip/tar archives without the need to extract before
• ...

Which data models/formats should be supported

Radar data

Define data models for PolarGrid and CartesianGrid, and start with providing interfaces from RADOLAN and DX to these classes. Then successively add interfaces to other formats.

Rain gauge and other point data

As a kind of PointSeries, imported e.g. from

• shapefiles,
• csv,
• hd5 files
• pandas hdf5 files
• etc.

Auxiliary polygon data

E.g. catchment or administrative boundaries - imported from shapefiles

Auxiliary line data

E.g. river reaches (for visual overlay, only) - imported from shapefiles

Auxiliary grid data

E.g. DEMs or flood hazard maps (for visual overlay, only) - imported from

• GeoTIFF
• ArcINFO ASCII
• etc.

How data should be imported to the application

I think there are two main strategies:

1. Let the user specify windows in space and time as well as target variables/products, and then let the wradvis search for available data.
2. Let the user specify a directory. Then wradvis can screen supported data and offer the user filter mechanisms which data show.

Additional thoughts

• We should seriously consider to organise and visualise time series (e.g. time of rain gauges, zonal statistics, radar data) using pandas, in particular by using pandas' Series object.
• ...

Required classes

Maybe also be inspired by the Py-ART core classes

• Point
• Grid
• PolarGrid
• CartesianGrid
• PolarGridSeries
• CartesianGridSeries
• PointSeries
• GUI
• GlCanvas
• ...

TO BE CONTINUED