An API and demo frontend for monitoring flood likelihood in England, UK. Built to extend the Part IA Lent Term computing activity at CUED. Run the files in the tasks directory to see the standard functionality, or by using

$ ./runtasks.sh

The tasks are documented here.

View the map by running python applications/MapViewer.py.

Map of stations across England:

Hover over a dot to view basic info:

Click on a dot to go to the official gov.uk site for the station:

View some forecasts by running python applications/LevelPredictions.py. Forecasting can be done at each station by training and predicting from an LSTM RNN. Training and evaluation is done with tensorflow.keras and sklearn.

Example 1: forecast of a tidal stations:

Example 2: forecast of a river level shortly after a storm event (predicted the rise, but likely overestimates the future fall)

After (optionally) navigating to any directory using $ cd C:\path\to\download\location, use

git clone https://github.com/lorcan2440/Flood-Warning-System.git
cd Flood-Warning-System
pip install .

The floodsystem module will then be available for use. Alternatively, if git is unavailable, the package can be downloaded manually by using this tool. Unzip the folder, go inside and use pip install . when in the same folder as setup.py and the various other files.

Various uses of the API are given in the tasks folder. To get started,

from floodsystem.stationdata import build_station_list, update_water_levels

stations = build_station_list()
update_water_levels(stations)

Then the functions can be used, e.g.

from floodsystem.flood import stations_level_over_threshold
print(stations_level_over_threshold(stations, 1.0))