theblitz Project

An attempt to map the bombs of the blitz over London. The idea is to learn to use d3.js and have a video-like effect as in the the wealth of nations example. (Or make a stop motion video out of it...)

The Data

The nice article and datablog entry of The Guardian -- 'London Blitz 1940: the first day's bomb attacks listed in full' provides access to the historical records of the first air raid over London in 1940. It is a spreadsheet listing for 843 drops, the time, the address, the type of bomb and the damage caused. The times are for 7th Sep 1940. I converted it to CSV - with ';' as separator - and stored the result in this project.

Obtaining lat/lon for the drops

I tried various approaches to geocode the addresses in the data set:

• using nominatim-python, a python library for Nominatim.

Geocoding via Nominatim

The lookup of the 73rd entry in the data file,

Royal Arsenal, Woolwich, SE18, London, UK

gives

>>> import nominatim
>>> import json
>>> address_lookup=nominatim.Geocoder().geocode('Royal Arsenal, Woolwich, SE18, London, UK')
>>> print json.dumps(address_lookup)
[
    {
        "display_name": "Royal Arsenal Woolwich Pier, Woolwich, London Borough of Greenwich, Greater London, London, England, SE18, United Kingdom", 
        "place_id": "2112469544", 
        "lon": "0.070824870400824", 
        "lat": "51.4956072184614", 
        "osm_type": "way", 
        "licence": "Data Copyright OpenStreetMap Contributors, Some Rights Reserved. CC-BY-SA 2.0.", 
        "osm_id": "103128361", 
        "boundingbox": [
            "51.4951553344727", 
            "51.495906829834", 
            "0.0705746933817863", 
            "0.0711841061711311"
        ], 
        "type": "pedestrian", 
        "class": "highway", 
        "address": {
            "city": "London Borough of Greenwich", 
            "state_district": "London", 
            "suburb": "Woolwich", 
            "country": "United Kingdom", 
            "county": "Greater London", 
            "pedestrian": "Royal Arsenal Woolwich Pier", 
            "state": "England", 
            "postcode": "SE18", 
            "country_code": "gb"
        }
    }
]

Geocoding via Google Maps v3

It turns out that a lot of addresses are not geocoded by Nominatim, for example entry 225

128 Aylward Street, Stepney E1, London, UK

So I tried Google Maps v3 using pygeocoder library:

>>> import pygeocoder as geo
>>> import json
>>> drop = '128 Aylward Street, Stepney E1, London, UK'
>>> gm_lookup = pygeocoder.Geocoder.geocode(drop)
>>> print json.dumps(gm_lookup.raw, indent=4)
[
    {
        "geometry": {
            "location": {
                "lat": 51.5159781, 
                "lng": -0.0462063
            }, 
            "viewport": {
                "northeast": {
                    "lat": 51.5173270802915, 
                    "lng": -0.04485731970849797
                }, 
                "southwest": {
                    "lat": 51.5146291197085, 
                    "lng": -0.04755528029150203
                }
            }, 
            "location_type": "ROOFTOP"
        }, 
        "address_components": [
            {
                "long_name": "128", 
                "types": [
                    "street_number"
                ], 
                "short_name": "128"
            }, 
            {
                "long_name": "Aylward St", 
                "types": [
                    "route"
                ], 
                "short_name": "Aylward St"
            }, 
            {
                "long_name": "London", 
                "types": [
                    "locality", 
                    "political"
                ], 
                "short_name": "London"
            }, 
            {
                "long_name": "Greater London", 
                "types": [
                    "administrative_area_level_2", 
                    "political"
                ], 
                "short_name": "Gt Lon"
            }, 
            {
                "long_name": "United Kingdom", 
                "types": [
                    "country", 
                    "political"
                ], 
                "short_name": "GB"
            }, 
            {
                "long_name": "E1 0QW", 
                "types": [
                    "postal_code"
                ], 
                "short_name": "E1 0QW"
            }, 
            {
                "long_name": "London", 
                "types": [
                    "postal_town"
                ], 
                "short_name": "London"
            }
        ], 
        "partial_match": true, 
        "formatted_address": "128 Aylward St, London, Greater London E1 0QW, UK", 
        "types": [
            "street_address"
        ]
    }
]

But even Google Maps fails, for example on entry 224

12 Jewel Street, Stepney E1, London, UK This street probably does not exist any more.

File addressesNotFound.txt lists the addresses I could not geocode automagically. I did spot some of them using scanned maps of London from 1908 and 1940 or guessed.

Timeline

======== I would like to use d3.js like in the wealth of nations example, i.e. with an animation. The raid started after 16:14 when the German bombers entered British airspace. So the times from in the Guardian spreadsheet need to be date stamped.