Web client requirements

This page gives the functional and non-functional requirements for the web client as a simple list of what the software should be able to do!

This refers to the generic web client - we might also want to refine this for the EURDEP client. Some terms:

background map: some vector mapping that can define the content, e.g. roads, rivers, coast, ....
observations: the measurements obtained by a sensor at a given location, not always the thing we are interesting in.
state: the variable that the user is interested in plotting (note this is not always the same as the observations).
interpolation server: the implementation of the web service to support automatic interpolation, probably as a web processing service.

Functional requirements - things it can do

Allow the user to zoom / pan on the background map
Allow the user to select the features they want to display on the background map
List the state variables that can be interpolated on the map (?is this the only option - does this need to query SOS's?)
Allow a more detailed specification of the interpolation algorithm to be selected, depending on the level of interaction the user desires
Allow the user to specify a local file (?format?) that contains observations which are then sent to the interpolation service
Display a grid of values of the state as a (transparent) image overlayed onto a background map
Display a grid of variances (or other moments?) of the state as a (transparent) image overlayed onto a background map
Provide access to meta-data about the interpolation results which is returned by the interpolation server.
Show the locations, and using colour and size the values of the observations used to produce the map
Identify any observations defined as outliers in the interpolation process.
Allow the user to retrieve a histogram of the pdf of the state at a given location (from interpolation service - non-interactive, or derived locally from pdf - interactive)
Display the probability of the state exceeding a critical value specified by a user (from interpolation service - non-interactive, or derived locally from pdf - interactive)

Allow the user to load in a data set of observations in a range of formats (e.g. csv, tab delimited) and with the user define the necessary information to create observations conforming to our XML schema for submission to the interpolation server.
Allow the interpolation service to return the posterior distribution in parameterised form (e.g. from Project Process Kriging) and then compute the predictions client side.

reliability: The web client should always fail gracefully with exceptions caught and informative error messages.
usability: The web client should allow access by a variety of levels of user expertise, setting defaults sensibly where user do not specify these (?or should this be something the interpolation service does?)
performance: The client should be able to cope with N observations, and a grid of m by m interpolated results and should remain responsive.
portability: The client should run under Windows and Linux at a minimum.
extensibility: ?The client should be designed for easy extension to other web service based interpolation methods developed in the future?

The fact that much of the work could be done client or server side emphasises the importance of creating a well designed set of libraries that implement our methods - maybe in R, but AST will probably focus on C++, with an R interface.