The magnitude 7.0 earthquake in Haiti (Figure 1) on January 12, 2010 resulted in an emergency situation with a high risk of diarrheal disease due to unsafe drinking water. As a result, NGO responders provided household water treatment and safe storage (HWTS) meth-ods including chlorine tablets, liquid chlorine, ceramic filters, and bi-osand filters.1 In a study commissioned by UNICEF and Oxfam Great

Britain, Daniele Lanta-gne and Thomas Clasen investigated the effective use of HWTS methods to improve drinking water in sev-eral locations during the acute phases (within eight weeks) of emergencies through analysis of household surveys and water samples.1

Investigators col-lected data in Haiti from

February 14th to March 13th, 2010, conducting surveys of 442 house-holds that received treatment. Results of the survey relate to factors such as household knowledge of treatment systems and training for use of treatment system. Treated and untreated water samples were al-so analyzed to determine effective use based on E. coli levels and the presence of free chlorine residual.1 Locations of surveyed households were collected with a GPS.

This project explored the spatial relationships between household water treatment and safe storage (HWTS) effective use and household location. The aim was to test the hypothesis that effective use is di-rectly related to density of water treatment use. In other words, HWTS recipient households closer to other households that also re-ceived treatment were more likely to use the method and to use it cor-rectly than those far away from other households. Understanding these correlations would be useful for further assessing the uptake of a water treatment system in emergency situations.

Analysis for this study focused on the rural region of Léogâne, which is shown in Figure 2 below. Here, Deep Springs International (DSI) distributed chlorine based treatment systems. GPS coordinates of households in the region made up the main set of spatial data. Ad-ditional data collected from sur-veys was also used, such as the presence or absence of free chlorine residual in the tested water. The presence of free chlorine residual is an indication of successful use. Basemap data, including region boundaries, roads, image-ry, and eleva-tions were also gath-ered . Additional layers used for spatial analysis are listed in the sources section.

Statistical analysis based on spatial relationships was used to de-termine whether there was statistical significance associated with the number of households surrounding a given household and success of treatment at the given household.

To focus on the study area, households were selected by the attrib-utes of location (Léogâne) and land type (rural or urban). Imagery and slope in the area helped to verify land type (Figure 3). Households were also differentiated based success of water treatment to obtain a preliminary estimation of spatial relationships, as shown in Figure 4.

Seven sets of buffers were created around each household (Figure 5a). Because the rural region under study is mountainous, nearby households considered for analysis were within one mile. Distances of 50 ft, 100 ft, 250 ft, 500 ft, 1000 ft, 2500 ft, and 5000 ft were used. Resulting polygons were joined with surrounding household location points. Indication of spatial relationships are clear through the exam-ple buffers around isolated households without treatment success (Figure 5b) contrasted with buffers around clusters of households with successful treatment (Figure 5c). Information from these joins re-vealed how many surrounding households were within each buffer.

The output of these joins was used to cre-ate contingency tables, which were then ana-lyzed with a Chi-square test for signifi-cance. The table below provides an output of combinations that re-sulted in statistical significance (p<0.05). A total of 28 tests were cal-culated. Figure 6 provides a visualization of the results considering that there was at least one neighboring household within each buffer. Figure 7 show a similar correlation considering at least two house-holds. The same data was analyzed for whether there were at least three and four neighboring house-holds.

Based on the data from the spatial joins and a visual assessment of the households, there is a clear correlation between household den-sity and water treatment success. Chi-square statistical analysis did not provide the same result, however, because of the small sample size. The data appears to support the hypothesis that being closer to other households using water treatment will increase the likelihood of treatment success. Further investigation would be required to deter-mine if similar trends are present elsewhere, since this data is related to a specific and unique context.  

Map Data Figure 1: “Magnitude 7.0-Haiti Region” USGS Earthquake Hazards Program. http://earthquake.usgs.gov/earthquakes/eqinthenews/2010/us2010rja6/ National Geographic World Map, 2012 [map service] National Geographic, Esri, DeLorme, NAVTEQ, UNEP-WCMC, USGS, NASA, ESA, METI, NRCAN, GEBCO, NOAA, iPC. http://www.arcgis.com/home/webmap/viewer.html?webmap=d94dcdbe78e141c2b2d3a91d5ca8b9c9 World Countries, 2011 [layer package] Esri, DeLorme Publishing Company, Inc. ArcGIS online Figure 2: Haiti Basemap from United Nations, 2010 [map service] MINUSTAH GIS and UN Cartographic Section. ArcGIS online Household data courtesy of Professor Daniele Lantagne SRTM 90m Elevation Data for Haiti, 2010 [layer package] USGS, ESRI. ArcGIS online World Countries, 2011 [layer package] Esri, DeLorme Publishing Company, Inc. ArcGIS online Figures 3-5: Bing Maps Aerial, 2010 [layer package] © 2011 Microsoft Corporation an its data suppliers. ArcGIS online Haiti Basemap from United Nations, 2010 [map service] MINUSTAH GIS and UN Cartographic Section. ArcGIS online Household data courtesy of Professor Daniele Lantagne SRTM 90m Elevation Data for Haiti, 2010 [layer package] USGS, ESRI. ArcGIS online World Countries, 2011 [layer package] Esri, DeLorme Publishing Company, Inc. ArcGIS online

References [1] Lantagne, D. S. and T. F. Clasen. Use of household water treatment and safe storage methods in acute emergency response: case study results form Nepal, Indonesia, Kenya, and Haiti. Environmental Science and Technology. 2012, Web Article. doi: 10.1021/es301842u

Household Water Treatment and Safe Storage in Hai Examining the Influence of Spa al Rela onships on Effec ve Use During the January 2010 Earthquake Response 

Background

Data

Methods Findings

Sources

0 1 2 3 40.5Miles

Area Type

#* Urban Households

#* Rural Households

±

0 1 2 3 40.5Miles

Buffers

No Free Chlorine Residual

Free Chlorine Residual

1000 feet

2500 feet

5000 feet

±

Analysis

Buffer Minimum number of Households in 

Buffer p value 

1000    1  0.010976 

5000    1  0.002581 

5000    3  0.007092 

Figure 1: Loca on of earthquake epicenter 

Figure 2: Households surveyed in Léogâne 

Figure 3: Urban and Rural     Households 

Figure 4 

Figure 5a: 1000 , 2500 , and 5000  Buffers 

Figure 5c: A cluster of households              with free chlorine residual 

Figure 5b: Isolated household with         no free chlorine residual 

Projection: WGS 1984 UTM zone 18N

Catherine Hoar, CEE 187—December 2012

n=6

n=13

n=36n=64 n=79

n=93 n=99

n=0

n=1

n=1n=4 n=5

n=9 n=10

0

20

40

60

80

100

50 ft 100 ft 250 ft 500 ft 1000 ft 2500 ft 5000 ft

Pe

rce

nt

of

ho

use

ho

lds

wit

h a

nd

wit

ho

ut

FCR

ou

t o

f h

ou

seh

old

s w

ith

at

leas

t o

ne

ne

igh

bo

rin

g h

ou

seh

old

wit

hin

bu

ffe

r

Buffer distance

Percent of households with and without free chlorine residual out of all households with at least one neighboring household

FCR

No FCR

n= number of households

Figure 6 

n=0 n=0

n=9

n=27n=54 n=90 n=98

n=0 n=0 n=0

n=3n=4 n=8 n=10

0

20

40

60

80

100

50 ft 100 ft 250 ft 500 ft 1000 ft 2500 ft 5000 ft

Pe

rce

nt

of

ho

use

ho

lds

wit

h a

nd

wit

ho

ut

FCR

ou

t o

f h

ou

seh

old

s w

ith

at

leas

t tw

o n

eig

hb

ori

ng

ho

use

ho

lds

wit

hin

bu

ffe

r

Buffer distance

Percent of households with and without free chlorine residual out of all households with at least two neighboring households

FCR

No FCR

n= number of households

Figure 7 

Figures 6 and 7 illustrate the trend found for the ap-plied buffer distanc-es. In general, out of the households with neighboring homes, the percentage of households with ef-fective use (FCR present) was greater than the percentage of households with-out FCR.