Earlier this week I read a news article about a ghost house somewhere in The Netherlands. One of the neighbouring residents was interviewed about the issues they face lately. The next thing I read triggered my inner geospatial curiosity. Why??? Because the neighbour was told he wouldn’t tell the exact location of the house. Well, that’s a nice challenge for me while I’m stuck at home during the COVID-19 crisis.
Because I do not want to be the second person this month to steal something from @nixintel I mention that I’m using some techniques from his ‘Intelligence gap analysis’ write-ups mixed in a mind map to structure the investigation. Of course, I used Openstreetmap, google streetview and QGIS in this investigation.
News article
The article can be found at https://www.ad.nl/westland/je-zal-er-maar-naast-wonen-spookhuis-in-wateringen-staat-al-jaren-leeg~a863a0a2/. When you do not read this newspaper frequently, you will be able to read the complete article at once (you get 3 free premium articles per month). This was not the case for me so I could not read more than it is visible in figure 1.
Question: Where is the ghost house located?
For understandable reasons the neighbours rather don’t disclose the exact location of the ghost house. But is it realistic to assume that the location is not disclosed to the public? Of course in the direct vicinity of the town (Wateringen), everyone knows which house it is. And further away?
In this simple and short investigation, I will show you that geospatial privacy is not guaranteed.
Mindmap and gap analysis
I will use the gap analysis, explained by @nixintel, in a mindmap method. With this method, you will be able to systematically conduct your investigation bit by bit.
The main hub of the mind map is the title “Ghost house” and picture 1 (see figure 2). Next thing I add the 4 main questions:
1. What do we know?
2. What does this mean?
3. So what do I need to know?
4. How do I find out?
What do we know?
At this point we know a few things:
1. The context of the partial news article
2. Photo 1
3. Photo 2
The introduction tells us that the ghost house is in Wateringen and that it is untenanted for years, furthermore the house is decay what gives problems for the neighbours. And the last sentence of the introduction tells us that the exact location will not be disclosed in this article.
The first few lines of the first paragraph tell us the name of the neighbours and some information about the original resident, an old lady with a small white dog.
When we click on the photo the main photo will be enlarged. This photo shows the person in a backyard, likely his own (figure 1). There will also be some navigation buttons visible. The second photo shows us the main entrance of the ghost house and the house number of the ghost house but also the house number of a neighbouring house see figure 3.
What does this mean?
Based on the information extracted from the article We can narrow down the location to a small city in the province “Zuid-Holland” Wateringen. According to the wikipage (https://nl.wikipedia.org/wiki/Wateringen) of this city, it has approximately 16.000 inhabitants (according to the english version Wateringen has 13.000 inhabitants this page was created in 2009).
Since we saw the person at the first photo in a backyard on the right side of the ghost house. It is likely to say that the person lives at number 108.
It is somehow possible to read the whole article. If I want to investigate this story further I need to read the whole article.
The house with number 106 is most likely the ghost house because of the pile of mail, the decay of the garden and the fact that one of the comments on photo 2 says that the neighbours doing some maintenance at the front yard of the ghost house.
The ghost house is clearly a terraced house with on the left likely house number 104 and definitely 108 on the right side.
So what do I need to know?
First of all I want to read the whole article for more background information. So I need to know the methods to read a premium article on a news media site. There are plenty of methods to achieve this.
The background information can give me more leads or more questions to investigate, for example names, other locations etc..
In order to know where the ghost house is, I need to know where all houses with the house number 106/108 are located in the city of Wateringen.
How do I find out?
In order to read the full article I need to remove the blockade. In order to do this I want to start internet searching for the term “read content paywall newsarticle”. From there on I can investigate further.
The next thing I need to do is download some geospatial data that can be queried for residential information. Because of my experience with openstreetmap I know that the data for the Netherlands is present in the openstreetmap data. In a query I’m asking QGIS to give me all the houses with the number 106 in the city of Wateringen. Then export all the houses with number 106 to a new layer. After that I can check every house 1 by 1 in google streetview.
The investigation
Now lets start execute the investigation to the ghost house.
Internet research
When searching for the keywords “read blocked content paywall newspaper” google, bing or duck will provide us with a variety of suggestions to read. For example the site:
https://www.online-tech-tips.com/computer-tips/12-ways-to-get-past-a-paywall/ gives a tip to start with deleting cookies. I use ‘Cookie Quick Manager’ for this purpose but uBlock will work also for this site.
After installing the addon and clicking the cookie icon, click ‘Delete current Site Cookies (x)’. After refreshing the current page I’m able to read the whole page and the whole article.
This article gives me more information about the original owner of the house but also her son. The story provides me with some leads to a Facebook account, but nothing more about the ghost house itself.
QGIS
So I need to download, convert and load the OpenStreetMap data of the province of Zuid-Holland into QGIS. When the data is loaded into QGIS I need to query the data. For that, I’m going to create a small test query in a controlled environment. I’m going to extract:
1. All the houses in a small village for example “Kwintsheul”
2. Search and select a random house
3. Inspect the feature data and use this knowledge to build a query for the randomly selected house
4. Run the test query
If it returns the expected results we can build the query for the ghost house.
Download, convert and load data into QGIS
Go to geofabrik.de and download the osm.pbf file for Zuid-holland.
http://download.geofabrik.de/ → Europe → Netherlands → select ‘.osm.pbf’ next to Zuid-Holland or download http://download.geofabrik.de/europe/netherlands/zuid-holland-latest.osm.pbf
When the download is finished we need to convert it so that QGIS understands the data and can present it on a screen.
At this point, I expect you already installed QGIS with all the extra dependencies, further from this point I explain it with a windows 10 installation. Search (in start menu type: OSGeo4W) for the OSGeo4W shell and open it. When the command prompt from figure 5 shows up type in the following command ogr2ogr.exe
When you see the message from figure 6, you’re in business.
Now in order to convert the ‘zuid-holland.osm.pbf’ to a QGIS database file navigate (1) to the downloaded file (use dir to know the correct name of the file (2)) in the shell and execute the following command (3):
ogr2ogr.exe -f “SQLite” zuid-holland.db Zuid-Holland.osm.pbf
When the process is finished you see a message. We can now close the shell and start QGIS. At the moment of writing I use QGIS 3.12. When QGIS is ready click on ‘New empty project’ and click and drag the ‘zuid-holland.db’ file into the screen of QGIS. At this point, I’m only interested in points, so select points to import.
Before we can go further we need to install a few plugins if not done already:
1. Streetview
2. QuickMapServices
The streetview plugin gives you the possibility to go to a streetview scene and QMS gives us a load of maps such as google maps or openstreetmap. So load OSM maps from the QMS plugin:
Start querying (in a controlled area)
1. Uncheck the point layer Zuid-Holland
2. Zoom into the Kwintsheul area
1. Check the point layer zuid-holland back on
2. Click on the information icon
3. And select a random house point
When selected a house point the following information screen will pop-up:
1.The fieldname ‘other_tags’ is the field with the juicy information
2.The value of the field ‘other_tags’ contains the city, streetname and the housenumber as well (Kwintsheul, De Raaphorst (3) & 22 (4)).
Copy and paste this information to a notepad and close this screen. Now be sure that the point layer is selected (1) and click on the ‘Object selections with expression’ (2) (3).
Enter the following at the left screen:
“other_tags” ilike ‘%”addr:city”=>”Kwintsheul”%’ and “other_tags” ilike ‘%addr:housenumber”=>”22"%’”
And execute the selection query. After a while 28 points will be selected (yellow). Nice!!! That means this method works.
Now lets create the same query but for Wateringen and house numbers 106.
“other_tags” ilike ‘%”addr:city”=>”Wateringen”%’ and “other_tags” ilike ‘%addr:housenumber”=>”106"%’
When clicking (3) you zoom to the area with selected points. In our case there will be 10 houses selected!!! Lets check them out in streetview. But first save the selected houses by rightclicking on the layer and select Export → export selectected.
Use the ESRI shape format (1) and a save location (2) and press ok (3).
By now we have a new layer named ‘wateringen’. Uncheck the zuid-holland point layer and check the new wateringen layer (and select it) (1). Open the attribute table (this is the table with all the points in it) by pressing F6 or clicking the table icon (2).
1. Select the first row
2. Click on the magnify icon (zoom to selected row)
3. QGIS will bring you to the selected house
4. Click on the streetview icon
5. Select the street
6. After releasing the mouse button a browser will be opened and a streetview of the house will be present
Repeat the steps above until you found the ghost house
Comparison between streetview and article photo
1. Tree still standing (5)
2. Red/white sunshading (7)
3. Yard lantern (6)
4. House number 106 on the garbage can ( )
5. The number 106 on the wall (1)
6. The house opposite to the ghost house with solarpanels on its roof (4)
Confirmed!!! This is the ghost house.
Conclusion
By now it is clear that the privacy of the neighbours and the ghost house are not 100% private. It is probably not a big issue for these people but if it was they were found within 30 minutes.
People do share more geospatial information than that they know or are aware of. One reason for that is that they do not understand the possibilities (or risks) of (geo)-information. For example the house number on the photo combined with opensource geospatial data like OpenStreetMap.
In 2018 a reporter of the dutch KRO-NCRV made an item about the privacy of victims and criminals who were filmed by a police vlogger. Bellingcat published an article about it https://www.bellingcat.com/resources/case-studies/2018/02/10/dutch-police-social-media-activity-raise-privacy-concerns/. So it is not a new issue but something that’s going way back. Youtube and other media sources are full of this kind of ‘exercises’ for OSINT investigators.
