CRASH

Inspiration:

weather has effects on transportation mobility. Especially the safety is at hazard when certain conditions arise (e.g., fog, ice, heavy rain). The goal is to see (quantitatively) the effects of weather on occurrence of traffic accidents.

What it does:

The aim is to do a statistical analysis of weather conditions when traffic accidents have occurred. What it actually NOW does is finding the weather conditions (temperature, dew point, total precipitation, snowfall, wind gusts, wind speed) at the location of all accidents in The Netherlands in the period 2006-2015.

How I built it:

1) Data fetching of ERA Interim from MARS. 2) Format accident data to make it accessible (e.g., time stamp converted to proper format) as Pandas dataframe for further analysis 3) Combining of weather and accident data 4) Initial exploratory data analysis via distribution plots

Challenges I ran into:

Accident data are continuous in time and space, that's hard to match with a discrete time and space reanalysis data when you have many (hundreds thousands) accidents. Especially if you want it to do in an efficient and fast way.

Accomplishments that I'm proud of:

reading in NetCDF and matching it with the accidents :-)

What I learned:

the demand of fast access to gather the meteorological conditions in points in space and time has to be addressed by the meteorological community. In order to allow more sectors/users/business to access fast and easily the meteorological conditions of points in time and space some effort has to be done.

What's next for CRASH:

We want to go further with the analysis. However the 3-hour resolution in time and 15km in space of ERA Interim might not be adequate for local events like accidents. We might look for higher resolution data from KNMI.

Built With

• python