The Inspire Helsinki 2019 event brought together around 170 people from 29 countries to foster discussion and new ideas on how to realise the full potential of spatial data. The three-day event featured data challenges, practical hands-on workshops and future-oriented keynote presentations. The event was summed up in a panel discussion, in which perspectives on tackling remaining challenges were brought up.
7. Route-network selection
The Dutch ‘Fietsknooppunten’ network, because:
Enough time between network nodes to perform ‘heavy server-side computing’, to allow for
dynamic rain-imposed route deviations
Still dense enough to avoid rain showers.
Recreational route network (detours less of a problem)
8. Datasources used
Route network data from:
https://www.gpstracks.nl/knooppunten-nholland.php.
Data needed to be corrected to be used as a topologically correct routing network.
Forecasts of precitipation data:
Originally from the KNMI in HDF5. Forecast data (upto 2 hours ahead) with a 5-minute refresh rate,
interpolated in between two weather stations in Den Helder and de Bilt.
We used the following image service* that ESRI created from the above mentioned KNMI-data:
https://meteo.arcgisonline.nl/arcgis/rest/services/KNMI/RAD_NL25_PCP_FM/ImageServer
*As the FME image service-reader does not suport arcgis image server services created with version 10.1
or above, we used in fme an ‘HTTPCaller’ on this service with the service’s export-image-capability.
9. Methodology I
The main challenge: matching for every ‘routesection’ in the network the cyclist’s expected arrival time with
the expected arrival time of rain.
In order to match these times:
– We perform 24 export-image requests of the whole province. A single image for every 5- minute interval for 2
hours of forecast
– Original network linkages are split up in routesections: ‘how far a cyclist can bike within 5 minutes (the
refresh rate of the raindata) going 18km/h’ = 1500m.
For all route sections the cyclist’s arrival time can now be calculated, and matched with the arrival time
of rain.
We classified ‘rain’ into 7 classes of intensity, as to be able to minimise how wet you get on a rainy day with
clouds everywhere. The higher the rain-intensity, the higher the cell’s weight, and the least likely it is to be
selected by the ‘shortest path-finding algorithm’
10. Methodology II
Rainfree routes cannot be planned ahead unlimited because:
– rain (and as a consequence raindata) may change its course of action continuously/(every 5 minutes).
– after the first ‘illogical’ rain-imposed detour the expected arrival time needs to be recalculated
– forecast data is only available for the first 2 hours.
To deal with these uncertainties a GPS-triggered function should be implemented to recalculate the route ahead at every
network node.
This brings along some uncertainties regarding the total cycling time / distance.
11. Bootstrap
Webclient:
HTML
Esri leaflet
FME Cloud
ArcGIS Server / ArcGIS.com
Origin and destination
Origin and destination
Dry route
Dry route
ArcGIS.com
Precipitation
forecast
Architecture
12. Architecture
Provincie
Provincie
ArcGIS.com
FME Cloud
ArcGIS.com
Choose origin and
destination
Precipitation
forecast
HDF5
Precipitation
forecast:
ArcGIS Image
Service
Bicycle network:
points
(hosted feature
service (REST))
Workbench
Bicycle network:
nodes and
network
Map layers Services
Base layers
Base layer
services
Which points are the origin and destination?
User input
Origin(ID) = 1
Destination(ID) = 2
Bicycle network:
proposed route
(REST service) ?
Bicycle network:
Origin,
destination and
proposed route
Architecture v02
BetterWetFromSweat Oct 2019
Precipitation
Dutch Meteorological
Institute (KNMI)
Bicycle network:
network
(hosted feature
service (REST))
13. Further possible improvements
GPS-triggered recalculation at every network node, after every break, or every significant change of
speed.
Inclusion of head/tail wind (INSPIRE)
Inclusion of cyclist’s willingness to increase speed (parameter), in order to make shorter-distance
routes possible.
Inclusion of a user parameter that lets the user balance acceptable intensities of rain (kg/m2/h)
with distance to be covered.
Inclusion of shelter places
Inclusion of messages for the user like expected remaining distance, expected arrival time, and the
required speed to stay dry!
Or with minor changes a totally different app to ‘Race against the rain’!