A project for Strengthening data bases on disasters
Commissioned by Working Group 3 (on Risk, Vulnerability and Impact Assessment) of the Inter-Agency Task Force
( IATF ) on Disaster Reduction of the International Strategy on Disaster Reduction ( ISDR ), a comparison was
carried out of two types of database, one with world coverage and the other concentrating on South America.
The databases included in the analysis were EmDat and DesInventar. Four countries were selected - Chile,
Jamaica, Panama and Colombia.
In the two databases, the fields possible for comparison were identified (date, type of event, number of
deaths and number of people affected) and the terms of comparability established. Equivalent entries in each
database were identified, ie those DesInventar entries that corresponded to the same disaster recorded in
EmDat. It was not possible to establish equivalences for all the EmDat entries, either because of the general
nature of the EmDat information, or because the event was not recorded in DesInventar.
The non-equivalent entries in EmDat could not be analysed in relation to DesInventar, because there is no
reasonable way of establishing possible comparisons or equivalences.
The non-equivalent entries in DesInventar were analysed in two ways; firstly, by identifying those entries
that individually or grouped together (resembling an EmDat-type entry) meet the criteria for inclusion in the
EmDat database, viz. ten or more deaths and/or a hundred or more people affected; and secondly, by analysing
the remaining entries, ie those non-equivalent entries that do not meet the EmDat criteria, in relation to
all the entries in the DesInventar databases.
Sections 1 and 2 of this report describe the general framework and details of the methodology used in
carrying out the comparison. Sections 3 and 4 set out the overall results of the comparative analysis for the
equivalent and non-equivalent entries respectively. Appendices III to VI give the analyses for each country.
The overall results show that equivalent entries represent 58% of the EmDat entries and 6% of the DesInventar
entries. In terms of number of deaths, the entries in EmDat are generally of the same order of magnitude as
in DesInventar; in terms of number of people affected, the differences are substantial, due to the
under-recording of this variable in DesInventar for three equivalent entries.
Among the non-equivalent entries in DesInventar, 7015 were identified that, either grouped together or
individually, meet the EmDat criteria; these have been called "virtual EmDat entries". They represent a total of 2968
EmDat-type entries. In defining the "EmDat disaster universe" as the sum of
the available entries plus the virtual entries, the former represent 5% of
this universe; in terms of number of deaths and people affected, they
represent 90% and 40% respectively.
Those non-equivalent DesInventar entries that do not meet the EmDat criteria represent approximately 60% of
the total number of DesInventar entries. In terms of the variables analysed, they represent 7% of the total
number of deaths recorded and 10% of the number of homes destroyed.