• Responding to natural disasters

Responding to natural disasters

The analysis in this paper identifies an underlying upward trend in the numbers of people affected by climate related disasters since 1980. This paper predicts that, by 2015, on average over 375 million people per year are likely to be affected by climate-related disasters. This is over 50 per cent more than were affected in an average year during the decade 1998–2007.

Limitations in the quality and coverage of data currently available, coupled with the natural volatility in numbers affected in a given time frame, will limit the robustness of statistical forecasting models. Nonetheless, this analysis provides a broad-brush indication of the rising scale of humanitarian need due to climate-related disasters in the relatively near future.

The author discusses issues such as:

• the CRED EM-DAT Database: maintains a publicly accessible database on emergency events.
• the forecasting model: forecasting the number of people that are affected by natural disasters is an imprecise science, and the figures presented here should be treated accordingly. However, the 2015 forecast predictions shown here were made using a simple two-step process:
  

1. double exponential smoothing: takes into account historic data from across the time series but gives more importance to more recent disaster events than to those longer ago
2. linear regression: estimate a time series trend, it is possible to calculate the upper and lower probable ‘range’ of a future forecast.
   

Recommendations include:

• The data shows a significant variation in the number of people affected from one year to another. Such ‘volatility’ in the data means that different forecasting models could lead to different results. It also means, however, that a highly sophisticated model is unlikely to be more precise than the relatively simple model used here
• It is likely that the reporting of natural hazards as well as data collection standards, definitions and sources, have changed considerably both over time and in different locations
• In addition estimates of people ‘affected’ by an event are likely to be less consistent and more volatile than counts of people actually killed
• It is reasonable to assume that more recent data is likely to be more accurate than older data.
  

.

• Full text of document
• Publisher details
• Send to a friend