This repo contains a model that given a longitude, latitude pair returns the risk that flooding will occur at the address. We compute two types of flood risk: The data used comes from dataforsyningen,datafordeler and the danish environment ministry
Flooding caused by excessive rain. (Skybrud in danish). The risk is a combination of the following factors (english/danish). To read more about a factor visit the notebook. The factors include:
-
Hollowing/Lavning: A hollowing is a "hole" in the ground where water will collect.
-
Fastning/Befæstelsesgrad: The amount of area around the point that has buildings, roads or other surfaces the water can't drain through.
-
Conductivity/Hydraulisk ledeevne: The ground type in the area.
-
Housing data: The properties of the house located at the specified point. For instance if the house has a basement it increases the risk of flooding. The danish building registry provides the building data.
Flooding caused by rising sea levels due to storm. This affects people who live close to the sea. The notebook provides more detail
The data is primarily given as images served by the Web Map Service (WMS) protocol. As an example the flow for retrieving a hollowing image given the longitude latitude is as follows:
- Convert the longitude and latitude to The ESPG 3857 projection. This projection uses meters as it's unit making computations easier.
- Create a bounding box around the specified point
- Request an image with the box at the projected point with.
The model consists of two modules the code module containing plain python code, it has functions for data retrieval and image handling. The risk analysis is in the notebooks module, the notebook files contain both python code and markdown text describing the computation.
The model uses the pipenv package as environment manager, it specifies the python version and packages required to run the code. To set up pipenv run see the install section in it's readme.
Before activating a pipenv shell a .env file should exists, copy the
.env.example to .env and fill out the variables. Creating credentials for
Datafordeleren and Dataforsyningen is free.
To run the model start pipenv and run jupyter notebook from the project root.
See the notebook to read more about the models and see them in action.
To run the tests issue the command python -m unittest in an activated
pipenv shell from the project root.
The code on the master branch is automatically deployed to an AWS Lambda setup. The master branch has protection and requires the tests to pass before a merge is possible. OBS: Be careful when upgrading packages. We are close to a max file size supported by Lambda. If possible, avoid upgrading too much.
The react page "When the Water comes" presents this model as a webpage.