jcrattz/senegal-floods

# DEAfrica-CSE Senegal Flood Analysis This repository contians Jupyter notebooks and associated data and documentation for conducting urban flood analyses in Dakar, Senegal. One of the goals is to create a model for determining flood hazard maps (flooding frequency) and flood risk maps (product of flooding frequency and estimated damages caused by floods in different areas) for the coastline of Senegal. ## Organization The **floodareas** directory contains the citizen science data (**citizen-data-flood-zones**), EO4SD flood hazard (**eo4sd_dakar_fhazard_2018**) and risk (**eo4sd_dakar_frisk_2018**) maps, and a GeoJSON file enclosing an area of Dakar particularly vulnerable to flooding (**Dakar_flooding.geojson**). The **landuse** directory contains a shapefile of landuse data (**landuse_DK.shp**). There is also a notebooks that loads and shows the data (**landuse_DK_test.ipynb**). The **precipitation** directory contains precipitation data. Information on obtaining CHIRPS precipitation data is included at the bottom of this document (place in a **chirps** subdirectory of **precipitation**). The **shapefile** directory contains population data in the **Commune_DK.shp** file. The shapes of Dakar's districts is contained in the **Dakar_districts.geojson** file. The **tests** directory contains various testing notebooks (e.g. loading and inspecting data). There are several methods that have been used to try to detect flooding in Dakar. ## Method 1 - Machine Learning **Summary: A larger dataset than the current one for Dakar is needed to create an accurate model for determining a flood hazard map for Senegal. The models that can be created with the current flood hazard map for Dakar do not perform well. If you are unfamiliar with machine learning, try one of the other methods mentioned below first.** The first attempt was to train and use machine learning models. These notebooks are in the **training_and_mapping** directory. The algorithm used in these notebo …