ICTCT workshop, Olomouc, Czech Republic, October 26-27th, 2017 “Traffic safety and the City" - How to enhance traffic safety and liveability in urban areas? ________________________________________________________________________ In Incle lement weather effect on road casualties in in urban areas: : the case of the metropolitan area Gdansk, , Gdynia ia, Sopot Ruth Bergel-Hayat(*) Wojciech Kustra, Joanna Zukowska RBH Conseil & Formation, France Highway & Transportation Engineering Department Nathalie Focant Faculty of Civil & Environmental Engineering Gansk University of Technology, Poland VIAS Institute, Belgium __________________________________________ (*)University Paris-Est, IFSTTAR GRETTIA for a part of the work
Plan of the presentation 1. Context of the research 2. Tricity metropolitan area 3. Casualty data 4. Inclement weather data 5. Modelling 6. Model’s outputs 7. Discussion 8. Conclusion 9. References Inclement weather and Road Safety in urban areas - ICTCT 26th _27th October 2017 2 Workshop in Olomouc, Czech Republic
The context • The relationship between weather conditions and road safety outputs has been questionned in 2010 (specially cold spells over Europe) • Researchers invested this topic intensively (SafetyNet, 2004-2008) • The IRTAD group of OCDE initiated a formal reflexion for analysis and short-term forecasting purposes (from 2014 onwards) • The emergence of the reflexion about road safety and cities leads to raise this question for urban areas. Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 3 Workshop in Olomouc, Czech Republic
Tricity Gdansk – 464.000 inhabitants 262 km 2 Sopot – 37.500 inhabitants 17,3 km 2 Gdynia – 245.000 inhabitants 135 km 2 Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 4 Workshop in Olomouc, Czech Republic
Casualty data • Road safety data for Gdansk-Gdynia-Sopot Metropolitan Area has been extracted from the Polish Police Database • The casualty numbers, for drivers+passengers, are disaggregated according to : - the severity ( killed and serious injuried , slight injured ); - the user type ( VRU among which pedestrians mainly, vehicle users among which car users mainly) • For the period 2000-2005, the total of victims in cars was uncomplete but this could be successfully corrected. Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 5 Workshop in Olomouc, Czech Republic
Casualty data Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 6 Workshop in Olomouc, Czech Republic
Casualty data Total number of victims: Disaggregation by severity type: 96,4% are injured,:slight injured (85,7%), seriously injured (10,69%), 3,6% are fatalities Disaggregation by road user type 48,4% are VRU: Pedestrians (34,4%), bicyclists (9,3%) and motorised 2 wheelers (4,6%) 51,6% are Vehicles: Car + lvc users (45,9%), hvl (1,5%), Bus + Trolleys (2,8%), Trams (1,2%) Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 7 Workshop in Olomouc, Czech Republic
Inclement weather data • The weather data have been gathered from the Agri4cast website • The weather data are measured by grids • The center of grid nr 118132 with Latitude 54,30 and Longitude 18,70) is the nearest grid center to the Tricity (Gdansk itself having Latitude 54,37 and Longitude 18,63) • The weather data, averaged over all grid centers in Poland, are used for measuring the average weather for the whole of Poland Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 8 Workshop in Olomouc, Czech Republic
Weather data: 3 types of days with inclement weather rainy and cold days (13%) , dry and cold days (13%), rainy and warm days (33%), and normal days (41%) 100% 90% 37% 80% 41% 70% 60% normal 14% 13% 50% dry_cold 16% 40% 13% rain_cold 30% rain_only 20% 33% 33% 10% 0% Gdansk/Tricity Poland Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 9 Workshop in Olomouc, Czech Republic
Modelling • State space methods • A locally linear trend (stochastic level and slope) and seasonal • Dependant variable: Number of victims in the month; • Independant (explanatory) variables: numbers of 3 different types of days with inclement weather in the month; • With interventions • On a monthly basis (January 2000 - December 2016) • Main output • Independant variables coefficients’ estimates (3 per model) • Using Stamp (in Oxmetrics) • or other softwares (such as the SAS system, proc UCM) Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 10 Workshop in Olomouc, Czech Republic
Model’s output: Tricity 2% 1% 0% -1% -2% -3% -4% rain_cold dry_cold -5% -6% rain_only -7% -8% Lvictims LvictimsCARS LvictimsVRU LvictimsPed LvictimsBic LvictimsMot Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 11 Workshop in Olomouc, Czech Republic
Model’s outputs: All cities in Poland 2% 1% 0% -1% -2% -3% -4% rain_cold dry_cold -5% -6% rain_only -7% -8% Lvictims LvictimsCARS LvictimsVRU LvictimsPed LvictimsBic LvictimsMot Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 12 Workshop in Olomouc, Czech Republic
Model’s outputs: estimates ’ interpretation - Tricity • A rainy and cold day is linked with: a decrease in the numbers of victims in the month, regarding the total of VRU • A dry and cold is linked with: a decrease in the numbers of victims in the month, regarding the total of VRU • A rainy day without cold is linked with: a decrease in the numbers of victims in the month, regarding the total of VRU a increase in the numbers of victims in the month, regarding car users Unfortunately, no estimates could be computed regarding each type of VRU. Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 13 Workshop in Olomouc, Czech Republic
Model’s outputs: estimates ’ interpretation - all cities in Poland • A rainy and cold day is linked with: a decrease in the numbers of victims in the month, regarding all VRU: pedestrians, bicyclists and motorized 2W (-1%, -7% to -8% resp.) • A dry and cold day is linked with: a decrease in the numbers of victims in the month, regarding all VRU: pedestrians, bicyclists and motorized 2W ( -1,5%, -4% to -6% resp.) • A rainy day without cold is linked with: a decrease in the numbers of victims in the month, regarding 2 Wheelers: bicycles and motorized 2W ( around -2% each) a increase in the numbers of victims in the month, regarding pedestrians (+0,8%) and also, but less significantly regarding car users (+0,4%). Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 14 Workshop in Olomouc, Czech Republic
Discussion • Choose between estimates ’ significancy and data disaggregation • Significant coefficients’ estimates can only be provided by aggregate (Gaussian) models when the victim numbers are large enough • Other models are more appropriate for disagreggated victim numbers • Choose according to the final aim • Analysis of inclement weather impact on specific victim types data • Analysis of (aggregated) weather impact on aggregated victim data • Analysis or correction for the weather impact ? Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 15 Workshop in Olomouc, Czech Republic
Conclusions • The significant results highlight the relevance of taking 3 inclement weather configurations into account in road safety trend analysis. • Information related to rain, snow and frost, appropriately aggregated at urban area-level and at the scale of the month, enrich the short-term trend analysis performed at the monthly level. • A relevant output of the analysis is that it provides correlations between weather variables and types of road user's casualty variables aggregated at urban area-level. • The work done on the Tricity area provides comparisons with the total of casualty variables in urban areas in Poland. Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 16 Workshop in Olomouc, Czech Republic
Outlook • Comparisons not only on other (the total of ) casualty datasets in urban areas in Poland but also on the same type of road user's casualty datasets in urban areas in other countries. • These results could serve to improve road safety management (alert levels by inclement weather are to be set), and to improve road safety trend assessment as well (short-term changes in the trend, due to atypical inclement meteorology are to be corrected for, which highlights the changes imputable to other factors such as road safety policies). Inclement weather and Road Safety in urbain areas - ICTCT 26th October 2017 17 Workshop in Olomouc, Czech Republic
Recommend
More recommend
