pyEmdat is a library for analysis of the EM-DAT natural disasters dataset (developed as an in-house tool at GFDRR). It allows to:
-
Load and clean EMDAT data
-
Summarize disaster statistics by country (single time period & annual time series)
-
Summarize disaster statistics by hazard type (single time period & annual time series)
-
Combine with population and GNI data (via World Development Indicators API)
EM-DAT is a global database of natural disaster events and impacts maintained by the Centre for Research on the Epidemiology of Disasters (CRED) at Université catholique de Louvain. It is used widely to inform policy-making on disaster risk management: for example, it helps policymakers identify the disaster types that are most common in a given country and that have had significant historical impacts on human populations.
Researchers can access EM-DAT through CRED's website: http://public.emdat.be. The data can be downloaded by researchers in Excel format (having created an account and agreed to applicable conditions). pyEmdat makes analysis easier as researchers were rewriting the required functionality each time or using Excel.
First, download EM-DAT data which comes in Excel format. Abide by all terms and conditions of CRED including downloading only the data you need.
Second, create an emdat
object
from emdat_df import emdat
ED = emdat('home/data/my_EMDAT_download.xlsx')
Third, use emdat
's built-in methods to filter, group and aggregate the data.
Fourth, use the built-in utils
to compare natural disaster impacts with population and GNI data from World Development Indicators.
See example notebook for usage.
Load EM-DAT data as above, then:
df = ED.disaster_count_timeseries(1960, 2000, countries = 'all', disastertype = ['Storm','Flood', 'Earthquake','Volcanic activity','Landslide'])
df.plot(title = 'World: Number of events by year');
df = ED.country_stats_timeseries(1980, 2020, ['Georgia','Armenia','Azerbaijan'],'all','total_damages')
df.plot.area(title = "South Caucasus: Total damages 1980 - present (all hazards, current USD '000)");
Documentation for pyEmdat is available at: https://pyemdat.readthedocs.io/en/latest/
- Merge with country polygons to make choropleth maps
- Damage stats in real 2010 USD
- Annual average loss for selected countries/hazards/period (real 2010 USD)
- Possible additional data structure for Desinventar data