The Effect of Early Flood Warnings on Mitigation and Recovery During the 2010 Pakistan Floods - Reducing Disaster: Early Warning Systems for Climate Change - page 241

experienced only small perennial floods were heavily impacted. At least 20 million
people were affected, with an estimated 1.6 million houses destroyed, roughly 2,000
deaths and 3,000 injuries, and several hundred thousand remaining without shelter,
even 6 months or more after the flood waters had receded (EM-DAT 2012 ). As a
result of lost crops, livestock, and livelihoods, many households also experienced
illness or death due to malnutrition.
The severe 2010 flooding in Pakistan provides an appropriate context for exam-
ining the impact of early disaster warnings, for two main reasons. First, there was
considerable variation both in flooding levels and in access to external information
and assistance. For example, Rawalpindi area (near Islamabad) received an average
8.6 in. of rainfall, while Lahore received an average 11.3 in., creating the potential
for wide variation in flood impacts within districts and even within villages at a
household level. This variation makes it possible to compare groups that did and did
not receive external warnings, at different levels of flood impact, controlling for a
variety of characteristics.
Second, the 2010 flood was a severe flood not experienced previously during this
generation (Webster et al. 2011 ), but many areas had experienced perennial flooding
at less severe levels, which created a wide variation in personal experience and
second-hand knowledge of flooding dangers and appropriate mitigation activities.
For example, between 1980 and 2010, seven of Pakistan's top ten worst natural
disasters (in terms of people affected, lives lost, and economic damage) were flood
events. 1
By collecting a unique data set of households in both flood-designated and non-
lood-designated villages, we are able to compare the effects of warnings on house-
hold mitigation activities, actual loss values, and recovery outcomes, while
controlling for personal characteristics and previous flood experiences. We can
therefore measure the effect of different types of warnings on certain mitigation
behaviors, as well as observing variation across types, which is potentially useful
for disaster management authorities who would like to improve the targeting and
timing of warnings.

13.2 Local Context of Early Warnings and Mitigation

Options in Pakistan

Numerous authors have pointed to the need for better hydrological modeling and
predictions of severe flooding to help the Pakistan government prepare for disasters.
On the other hand, the 2010 floods sparked a debate about whether existing
warnings were adequately transmitted to local governments or heeded by the popu-
lation. For instance, Fair ( 2011 ) says that flood warnings that were issued by the

1 EM-DAT Global Disaster Database: http://www.preventionweb.net/english/countries/statistics/?cid=129