This small example illustrates a simple uncertainty-enabled workflow that corrects a station pressure measurement to sea level pressure. The correction is made (for the UK only) using elevation data from the Shuttle Radar Topography Mission. The main aim is to illustrate the uncertainty enabled service workflow we have created in UncertWeb, which was created to prototype different mechanisms for orchestrating services in a workflow. All of the services are exposed as processes that conform to the Open Geospatial Consortium (OGC) Web Processing Service (WPS) standard.
Input
Specify the station location by clicking on the map (somewhere in the UK - we haven't stored the elevation data for the rest of the world!) or entering the coordinates in the form below. To refine the station location, either drag the marker or click the map again to reposition. Enter a pressure measurement - this should be the station level pressure, not the sea level pressure - and click 'Start' to begin execution of the workflow.
Details
This process retrieves elevation samples from the Shuttle Radar Topography Mission. For the station location, given as a GML point, a grid measuring 9x9 cells is created around the location and the elevation and associated uncertainty at each cell is returned.
The data is returned as an O&M observation collection with embedded UncertML distributions. In this particular example, all elevations have the same estimated uncertainty and correlations are ignored, but the UncertML schema allows a different estimate and characterisation of uncertainty for every data point in a collection, or a joint specification of the uncertainty. This process is hosted on the UncertWPS service.
Output
Profile of x-axis nearest to location
Profile of y-axis nearest to location
Request document
Click to showResponse document
Click to showDetails
This process interpolates the elevation samples using the INTAMAP automatic interpolation service to obtain a predicted elevation at the station location. INTAMAP exposes a range of interpolation algorithms via a WPS, including interpolation methods that enable the propagation of uncertainty from the individual samples to the interpolated output.
A prediction of the elevation at the station location is returned, complete with an uncertainty estimate, again represented as an UncertML distribution. This process is hosted on the INTAMAP service.
Output
Distribution of prediction elevation
Request document
Click to showResponse document
Click to showDetails
This process creates a set of realisations for the predicted elevation distribution, a step required as the pressure correction process only accepts a set of realisations as an input.
A set of predicted elevation realisations are returned, encoded as UncertML. This process is hosted on the UncertML Translator service which at the moment only offers very simple translations but will be extended in the future.
Output
Request document
Click to showResponse document
Click to showDetails
This process converts the pressure measurement to sea level using the predicted elevation realisations. The correction algorithm has additional sources of uncertainty. Based on expert elicitation, the standard temperature is assumed to have a Gaussian distribution with a mean of 288.15 and a standard deviation of 8°C. A Gaussian distribution is also assumed for the temperature lapse rate, with a mean of 0.00649 and standard deviation of 0.00175°Cm-1. Ideally, of course, these inputs would also come from data sources or models and thus have reduced uncertainty. For each of the elevation realisations, the pressure correction algorithm is performed with a realisation from the standard temperature and lapse rate distributions to propagate their uncertainty.
A set of resulting UncertML realisations is returned by the service. This process is hosted on the UncertWPS service.
Output
Request document
Click to showResponse document
Click to showDetails
This prototype workflow can be orchestrated using the Business Process Execution Language (BPEL). If you wish to orchestrate the workflow yourself, download the process bundle or view the documents below. The bundle contains additional configuration files for Apache ODE, but it is possible to deploy the BPEL document on any engine. A slightly more edible version is also available.