One obvious enhancement to LAS would be the use of an interactive tile-based (Google Maps, Bing Maps, Leaflet, Open Street Maps) map API to display data. This issue is a place to discuss the pros and cons. If we come up with specific ideas we want to try, I'll make separate tickets.
Some things to consider:
Most Map APIs create a separate DOM element for every glyph on the map (and for every info window) at the time the data are loaded. This makes for a huge load on the browser.
Most Maps use mercator projection which is nearly square in the mid-latitudes, but distorts strongly in latitude at the poles.
There is limited capability to color the glyphs based on the data values.
We might want to try making map tiles with data. PyFerret can easily make correctly projected plots which fit the underlying map tiling scheme. A tile can have any number of data types and data sets depicted without increasing the load on the browser beyond the cost of loading a single tile. A filled-contour, a line contour, vector arrows and a trajectory colored by data values could all share the same tile.
A major disadvantage of this approach is the cost of a round trip to the server in order to interrogate the data with a mouse click.
Example with two plot styles (one data parameter) below.
From an end user perspective, a few usefull features of a dynamic map would be:
interrogating a lat/lon location for non-plottable entites (e.g. station names, metadata of locations) and for local point values
depth-property profiles from clicking on a location (so breaking a trajectory-profile dsg into a map based trajectory plot which allows you to explore the profile component independently) , this is good for just profile dsg files as well
being able to toggle on/off certain groups of data within a larger collection... eg if all Alamo floats where in one dataset but each ID could be toggled
And a really big request
being able to overlay multiple datasets (eg model/reanalysis sst with cruise data or drifter data)
One obvious enhancement to LAS would be the use of an interactive tile-based (Google Maps, Bing Maps, Leaflet, Open Street Maps) map API to display data. This issue is a place to discuss the pros and cons. If we come up with specific ideas we want to try, I'll make separate tickets.
Some things to consider:
We might want to try making map tiles with data. PyFerret can easily make correctly projected plots which fit the underlying map tiling scheme. A tile can have any number of data types and data sets depicted without increasing the load on the browser beyond the cost of loading a single tile. A filled-contour, a line contour, vector arrows and a trajectory colored by data values could all share the same tile.
A major disadvantage of this approach is the cost of a round trip to the server in order to interrogate the data with a mouse click.
Example with two plot styles (one data parameter) below.