18
Algorithm implementation for a “Road Weather Information System”: the CLEAN-ROADS project case study ing. Gianluca Antonacci, CISMA Srl - Bolzano

SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Embed Size (px)

Citation preview

Page 1: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Algorithm implementation for a “Road Weather Information System”: the CLEAN-ROADS project case study

ing. Gianluca Antonacci, CISMA Srl - Bolzano

Page 2: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Description of the development of a set of algorithms within a “decision support system” by means of open source tools.

● Case study: the EU Life+ Project CLEAN-ROADS● Problem identification● Open source software tools● Achieved results

Summary

Page 3: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

The Life+ “CLEAN-ROADS” Project has been founded by EC and carried out by the following partners• Autonomous Province of Trento• IDM Südtirol• FAMAS System

AIM of the project● Forecast of road temperature in advance in order to support

road maintenance services in winter conditions in the Province of Trento● Improve the efficiency of the use of de-icing chemicals ● Reduce negative environmental impact and costs

by means of● a Maintenance Decision Support System (MDSS)

CISMA was contracted by IDM and responsible for part of the algorithm implementation, to be included in the decision support system toolchain

Case study

Page 4: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

• Ice forms on road pavements in high-humidity conditions when road surface temperatures are below 0°C → Slippery roads increase accident risk

• Spreading de-icers is common practice but excessive / uncontrolled use is both hazardous for the environment and cost-demanding for road maintenance system

(source: CLEAN-ROADS flyer – clean-roads.eu)

Problem identification (1)

Page 5: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

● Reliable temperature forecast needed to optimize maintenance costs and reduce the risk → weather forecast tools need to be included in the DSS

● A decision support system has to cope with differente “languages”, in this case not only informatics, but also engineering, statistics & physics

Problem identification (2)

decisionsupportsystem

Page 6: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Test Area

Implementation area

Page 7: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

• DSS should rely on already existing procedures and data, not everything can be re-engineered from scratch → customization of interfaces● Different competence needed● Calculus model needed ...● … with the requisite of being able to “talk” to different preexistent tools,

procedures, data

Problem analysis

statistics:environmentaldata analysis

engineering:on field de-icing

procedure +data acquisition

physics:modelization

of physicalprocesses

informatics:data elaboration,

transfer &visualization

DSS algorithms

Page 8: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Adopted software tools:- an open source temperature forecast model (named METRo)- high level statistical tools (R )- GIS software (GRASS, QGIS)- scripting tools (bash / python / php)- plotting (gnuplot and R)

Adopted software tools:- an open source temperature forecast model (named METRo)- high level statistical tools (R )- GIS software (GRASS, QGIS)- scripting tools (bash / python / php)- plotting (gnuplot and R)

Schematics (1)

Page 9: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Road temperature forecast scheme

EuropeanWeatherForecast

Local weather service

Extraction of airtemperatureforecast on

selected points

Database

Road-weatherstations (real time

data delivery)

Thermal mapping(scheduled delivery)

Single point temperature

forecast (METRo)

Spatialization & mapping

web service& graphics

Schematics (2)

Page 10: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

METRo model● One of the most widely used models for road weather predictions.● Created in 1999 by Canadian Environmental Agency, distributed under the GPL license and

actively mantained● Is a so called 1D model, i.e. performs forecasts at single points● Mainly written in Python but also use FORTRAN and C

Necessary input data: ● Data measured at a road weather station● Weather forecast (atmospheric data)● Road characteristics description

Advantages● Its code is freely available and well documented (http://home.gna.org/metro/.)● Performs well under a variety of weather conditions● Relatively easy to acquire, install and use.

Software tools

physics informatics

Page 11: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Data automatically gathered from installed road weather stations

access to DB through WSDL protocol → interface + data validity check (python)

Weather sensors: precipitationtemperature, relative humidity and global solar radiation

Road sensors: surface temperature, sub-surface temperature, surface condition

Famas System instrumentation(picture Final workshop presentation, Servizio Gestione StradeProvincia Autonoma di Trento http://clean-roads.eu)

Data gathering

physics informatics engineering

Page 12: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

● METRo software code was modified & adapted in the physical modelization in order to cope with our target (in fact is developed for Canadian area, not entirely suitable for our case), e.g. road emissivity, diffuse radiation, ground stratigraphy

● Multiple runs with different parameter settings → calibartion phase

● Tools: compilers, “R” statistical software and… a hardcopy meteorology manual!

Software setup

physics statisticsinformatics

Page 13: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

METRo is a 1D model, which performs forecasts at single points

To spazialize forecast, thermal mapping is performed: acquisition of temperature via an infrared thermometer mounted on a vehicle

Thermal mapping permits to identify the pattern and distribution of warm and cold sections along a road.

Thermal mapping permits to identify the pattern and distribution of warm and cold sections along a road.

(picture from: Thermal mapping as a valuable tool for road weather forecastand winter road maintenance - An example from the Italian Alps, RSCY 2016)

Spatialization

physics informatics engineering

Page 14: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Signal noise removal• Temperature noisy signal from termal mapping due to different road characteristic → low-pass Gaussian filter to remove this noise● Semi-automated process by means of “R” statistical packages (www.r-project.org)

Data analysis (1)

physics statistics

Page 15: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

characteristic thermal fingerprint and spatialized forecastsAlso this stage performed by means of “R” software

cold

warm

Data analysis (2)

physics statistics

Page 16: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

The combination of “single point” METRo forecast with the “thermal map” of the route allows to predict temperature over the entire route at selected times.

GIS based elaboration (SHP / KML files which can be overlayed on a map)

Mapping

informaticsengineering

Page 17: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

Web page accessible historyPHP and HTML coding, graphics generated on the fly using gnuplot program

Data publication

informatics

Page 18: SFScon16 - Gianluca Antonacci: "The CLEAN-ROADS project case study"

● The calculus system of a complex decision support system was successfully developed entirely adopting open source software

● Customization of well established calculus system (METRo) possible thanks to code accessibility

● Exploitation and scalability of the system is limited to the more general part; when integrating a tool in a already existent toolchain you're mostly forced to adapt the new part → the interfaces are really custom and not always reusable.

● The system has been running for nearly two years now from the first implementation

● Integration with existing procedures is very important; the system is modular and therefore maintainable in case of modification of other stages of the toolchain

Conclusion