Assessing preferences regarding centralized and decentralized water infrastructure in post-earthquake Leogane, Haiti

Overview of centralized and decentralized water infrastructure options

Centralized and decentralized approaches apply different spatial strategies, but also differ in terms of their associated costs, operation and management, latent vulnerability to hazards, and expandability potential. In many urbanized regions of the industrialized and post-industrial world, drinking water is sourced and distributed in a centralized fashion; i.e. source water is made potable (e.g. treated) at a centralized treatment plant and then delivered to users through a pressurized distribution system (Kyessi 2005). Though the most costly component of such systems is typically the excavation required to lay the pipes (USEPA 1991), centralized distribution systems are often considered the only way to provide water and sanitation services to densely populated urban areas where local water sources are often contaminated and unsuitable for consumption even after treatment. The requirement that cities served by such systems have sustained access to relatively high quality, reliable, extra-urban source water supplies, however, carries the potential to create upstream/downstream conflict, especially as urban growth infringes on source water watersheds.

While the planning, design, and construction of large centralized networks is often subsidized by the public or private sectors, recurring operation, maintenance, and repair costs are paid with revenues generated by user fees, often administered by a local governmental, quasi-governmental, or private body. The regular upkeep and repair of pressurized distribution systems requires specialized machinery and skilled labor, and may involve excavation, requiring that such infrastructure networks be accompanied by a trained team of water utility personnel.

Additionally, and because of the high cost associated with laying and maintaining pipes, centralized water distribution systems are typically designed to serve high density population centers only. The user costs associated with extension of the network into low density regions is typically much higher than the cost of providing exurban residents with a decentralized local water source (e.g. a protected well). As of 2010, only about 4% of the world’s urban population relied on unimproved water supplies, compared to 19% of the rural population (UN 2012).

Though their benefits (e.g. improved public health, environmental protection, streamlined operations, economy of scale, reliability) are well known, centralized water and sanitation systems are not always feasible or appropriate. In developing world settings, for example, where financing or other capital funds may not be readily available, the start-up costs to build a centralized water or sanitation system can be prohibitively high (Wilderer and Schreff 2000), and rural populations may lack a reliable source of energy for well pumps. In addition, if an economically stable user base is not present, the revenues needed to pay for recurring operation and maintenance costs cannot be guaranteed, a situation that can lead to gradually deteriorating system performance. Since damage to critical nodes or segments can compromise the functionality of the entire system, centralized infrastructure can also be vulnerable to environmental hazards such as earthquakes or hurricanes. In contrast to industrialized settings, where, on average and as a percentage of household income, the cost of a protected well or other decentralized water supply system is bearable for exurban residents, in developing world settings rural populations often rely on unimproved sources (UN 2v012).

In such contexts, there has been growing interest by infrastructure experts, researchers, and international lending institutions in applying funds earmarked for improving water access in a decentralized manner (Wilderer and Schreff 2000). In a broad sense, decentralized infrastructure replaces centralized treatment and distribution systems with a network of smaller, individual facilities (e.g. point-of-use or on-site treatment, no centrally managed pipe systems). From a management standpoint, it replaces a team of mainly utility personnel with grass roots networks, which could include local water committees, and municipal (as opposed to regional or national) decision-making bodies. Decentralized water sources are accessed by users either through a large number of relatively small distribution systems, or directly at the point-of-use (Peter-Varbanets et al. 2009). When evaluated on per user basis, investments in decentralized water systems in rural areas are as cost-effective as in urban settings.

More generally, and compared to centralized systems, decentralized systems can have lower maintenance costs, can require fewer upgrades, and be installed incrementally in response to actual demand (Venhuizen 1991). Decentralized systems have independent components, making the system less vulnerable to systemic failures, hazards, and extreme environmental events (Venhuizen 1991). Additionally, expansion of a decentralized system can be easier because it is not contingent upon availability of treatment plant capacity, as is necessary in expansions to centralized systems (.Mintz et al2001). A high level of service, however, is contingent upon the consistent availability of labor to operate and manage the entire decentralized network (Wolff and Gleick 2003).

Geographic and infrastructure setting

The commune of Leogane is distributed across a coastal plain and mountainous region located approximately 30 km to the west of Port-au-Prince. Approximately one-third of the commune’s 300,000 people live in the city’s urban center. The ~114 km² coastal plain is underlain by a productive shallow unconfined aquifer that typically can be accessed at depths of 5–10 meters below the surface. At greater depths (25–30 meters) a confining layer separates the unconfined aquifer from a deeper confined one. Two rivers bisect the commune and are accompanied by a multitude of smaller tributaries, irrigation canals, and drainage ditches.

The climate is marine tropical with a hot and humid summer and a cooler and drier winter season. There are two distinct rainy seasons; one April to June the other October to November. Leogane has no climate station but because it has similar topography, and is only ~30 km west of Port-au-Prince, its weather can be characterized as similar to that of the capital (average annual rainfall = 1370 mm, daytime temperature 25-30C). Haiti is hit, on average, by one cyclone (defined as a climatic event with windspeed over 250 km/h) with devastating effects every five years (Aquastat 2000).

Portions of the city of Leogane were, at one time, supplied with drinking water by a centralized gravity-driven distribution system built with international aid money in the early 1980s. The system was fed by an artesian spring on the eastern side of the Momance River. A main distribution line was buried in the river bed and resurfaced on the western side before continuing under the main road to Leogane’s urban center, where it subdivided into a gridded distribution system covering about a 1 km by 1 km square region. Two smaller lines branched off the main line, one going to the north (the Matthieu branch) and one to the south (the Belle Fortune branch). Another small line branched off the downtown service area to serve Ca Ira in the northwest. In addition to this centralized distribution system, a smaller subsidiary system fed by another artesian spring supplied drinking water to an unknown service area. While these distribution pipes may have sustained additional damage due to the earthquake, the system had actually been rendered inoperable during the 2008 hurricane season when floods washed out the river bed and destroyed the piping buried in the bed. At the time of this study, responsibility for the defunct system was being transferred from the Service National d’Eau Potable, the national water utility located in Port-au-Prince, to a regional office of the Direction Nationale de l’Eau Potable et de l’Assainissement, as part of the decentralization and reorganization of the water sector in progress throughout Haiti under the terms of a 2009 national water law.

The goal of this study was to elicit local perspectives on how reconstruction funds could be used to improve post-earthquake access to water services in Leogane. By documenting local preferences, the intention was to assist the reconstruction effort, namely by helping foreign donor agencies with the financial wherewithal to make key infrastructure investments in Leogane better understand local needs. Assessing whether local opinion favors a decentralized or centralized strategy for meeting the city’s water needs from afar is difficult. Though evidently Leoganais have coped with informal decentralized infrastructure for some time, at the time of this study post-earthquake reconstruction was viewed by many in Haiti as an opportunity to “build back better” (Fountain, 2010). Accurate or not, a higher quality of life is associated by some with centralized systems. To the extent that they are more reliant on local management structures, decentralized strategies could be viewed favorably by individuals with a low level of confidence in the government’s ability to effectively solve development problems.

Structured Paper Survey of Infrastructure Practices and Preferences

The survey included the “core questions” used by WHO and UNESCO (2005) for quantifying the percentage of a population that has reasonable access to water and sanitation. The responses were analyzed to determine the number and percentage of participants who used different water systems prior to the earthquake, and also to characterize reconstruction preferences. Because all respondents did not respond to all questions, there were slight differences in the number of responses to individual questions. We used the survey results to explore whether different water sources were used for different water needs (e.g. cooking, agriculture, hygiene, and religious or ceremonial use). Contingency tables were developed using Microsoft Excel and SPSS 19 (IBM Corporation, Armonk, NY), breaking down the results based on age, gender, geographic origin, and employment status. For categorical variables, Pearson’s chi-square test was used to assess if response frequencies differed significantly by any of these factors. For continuous variables, t-tests were used to assess differences in means.

Semi-structured Interviews

Interviews were carried out during both trips, and also by telephone from the United States. The 19 semi-structured interviews were conducted by convenience sample, and consisted of current residents of Leogane, expatriate former residents of Leogane, and non-governmental organization (NGO) representatives. These interviews were conducted by one or two team members (with a Kreyò l translator when required,) and digitally recorded. Based on a standardized set of “discussion points” and follow up questions, each interview lasted approximately 45 minutes. The recorded interviews were transcribed in English and coded for analysis with ATLAS.ti (Berlin, Germany). Analysis identified broad patterns of types of responses, common themes, actors mentioned, linkages between problems identified and solutions suggested, and enabled comparison of different groups of interviewees. A separate protocol was also used to check for the presence or absence of certain basic ideas and concepts. This protocol used contingency tables to identify trends similar to those in the structured surveys: by age, gender, geographic origin, and employment status.

Workshop

During the second trip in July/August 2010, the research team conducted a full-day participatory workshop. The team trained 10 Haitian facilitators, and the workshop involved 76 community stakeholders who were identified during the first trip. These local stakeholders and community leaders represented a range of organizations, such as youth associations, women’s organizations, professional organizations, and small farmer’s organizations. During the workshop, participants were asked to propose, discuss, prioritize, and elaborate solutions to Leogane’s water and sanitation problems. First, five groups of ten to fifteen brainstormed solutions. By vote, these groups brought their five top ranking ideas to the general assembly of all workshop participants. The general assembly voted on the 25 solutions proposed by the five groups, and the five top ranking solutions were elaborated more thoroughly during afternoon breakout sessions into proposals. Pre- and post-workshop surveys with Likert Scale questions were used to gage participant satisfaction with the process and results. Overall, ~95% of the respondents characterized the workshop as “a success”, with ~79% strongly agreeing and ~19% agreeing with its conclusions. ATLAS.ti was used to quantify the frequency with which different kinds of centralized and decentralized water and sanitation options were mentioned in all of the proposed ideas. The final five proposals emerging from the workshop are also scrutinized to infer preferences regarding the preferred degree of centralization of post-earthquake infrastructure.

Data Deposition

The paper surveys, as well as the pre- and post- workshop survey results were transported back to Drexel University where the results were entered into digital data files for analysis in Excel and SPSS. With the assistance of native Kreyò l speakers, the interviews were recorded, translated, and transcribed for analysis with ATLAS.ti.

Synthesis of paper survey results

Figure 1 depicts the percentage of respondents who indicated that they accessed different sources of water at different levels of frequency. The three most common water sources accessed for daily needs were well using bucket (35%), public tap/standpipe (28%), and well using pump (12%).

Figure 2 is a bar chart depicting the different sources of water accessed by the participants for different uses prior to the earthquake. The bars break down the responses for each use by source. Overall, the top two sources for cooking, clothes washing, hygienic needs, and drinking were “well using bucket”, and “public tap/standpipe”. Prior to the earthquake, approximately 40% of all respondents relied on “public tap/standpipes” for cooking, clothes washing, hygienic purposes, and drinking. For this same set of uses, the second most frequent source was “well using bucket”, but it is noteworthy that the percentage of respondents who relied on “well using bucket” for drinking purposes (<20%), was significantly lower than for the other uses (30-40%). Stated differently, groundwater accessed using wells and buckets was used for nonpotable purposes more frequently than for drinking. Water that was intended for ingestion was more frequently accessed from the “public tap/standpipe” (which may actually have originated in a local borehole).

Of 170 respondents, 74.1% said they treated water before drinking it (Figure 3), and the most common form of household treatment was a locally available bleach tablet, known as Aquatab (Figure 4). Perhaps because they treated water before drinking, very few respondents listed “sickness” or “death” as among the problem with their pre-earthquake water source. More commonly cited problems were “a long queue,” “water is not always available” or “tastes bad or looks dirty” (Figure 5). Figure 6 displays the average number of minutes required to access different water sources, (includes travel to and from the source, waiting and extraction time, but excludes all socializing time). On average, springs require about an hour, while public taps, artesian wells and wells with pumps require about half that time. Based on travel time, wells accessed using buckets appear to have been most readily accessed at 11 minutes. Women were more likely than men or children to be responsible for fetching water from all sources except artesian wells (Figure 7). We speculate that because they are very expensive to install, artesian wells may be culturally associated with male-dominated economic activities in Haiti and are therefore less accessible to women (see Sheller et al. 2013for a wider discussion of gender, water access, and sanitation issues relating to this study). In Figure 8 the various reported sources are classified as either centralized or decentralized. With the exception of water used for sanitary purposes and water used for commercial food production, decentralized sources were more frequently used than centralized ones. Table 3 is a comparison of pre-earthquake drinking water sources used with the source type that the respondents would prefer to use after the rebuilding process. More than half (53%) of those surveyed stated that they used a decentralized water source to obtain drinking water prior to the January earthquake. However, when asked, “As Leogane is rebuilt after the earthquake, which source of water would you prefer to be available to you?” 83% said they would prefer a centralized one. More detail is provided in Figure 9, which shows that for cooking, clothes washing, hygienic purposes and drinking, survey respondents expressed a preference for “piped water to yard/plot” and “piped water into dwelling” above all other sources.

Synthesis of Workshop Results

The workshop was structured to allow the participants to discuss a wide range of water and sanitation issues for Leogane. Figure 10 displays the number of times different phrases were mentioned during the workshop, as computed using ATLAS.ti. Of those phrases pertaining to water sources, “artesian wells” (a decentralized water source) were mentioned the most frequently (15 times), followed by “extending the piped water system” (13 times), and “capped” or "captured” sources (12 times). Sometimes these suggestions co-occurred, i.e. with “digging of artesian wells” mentioned in the same sentence as “extending the piped water system” (5 times) so that overall the preference seemed to be for a mixed system, including both centralized and decentralized strategies.

These proposals contain a variety of references to both centralized and decentralized water systems. The second proposal makes a clear reference to both centralized (e.g. “piped water system”), and decentralized (e.g. “artesian wells”) drinking water systems. Water in “public fountains” could be sourced either from an artesian aquifer or reservoir connected to it with pipes. The joint references to “public fountains”, “artesian wells” and “piped water” is more indicative of a goal of improving access to water services in general, than to a preference for a particular modality of infrastructure. The third proposal also suggests a desire for consistent levels of service “in each neighborhood”. Although there are different ways that water services could be provided to a neighborhood, no preference for how centralized the approach should be was articulated. The fifth proposal also cites both centralized (e.g. “pipes”) and decentralized (e.g. “artesian wells”) solutions.

A number of phrases and terms in these proposals suggest that generally the participants view decisions related to water as also related to a variety of other local issues. For example, the first proposal suggests that the participants see a relationship between water issues and protection of the larger Leogane environment. The fourth proposal recognizes both the need for treatment (consistent to the survey results) and mentions agricultural water needs, a topic not specifically emphasized in the workshop preamble (and generally invisible in the larger post-earthquake planning processes taking place nationally and internationally). The fifth proposal suggests that deforestation (usually attributed to charcoal making) is viewed as having an important role in local water issues.

Results of Contingency Analysis

The survey results indicate that while most participants seem to have relied on decentralized water sources prior to the earthquake, the majority would prefer reconstruction funds to be used to construct a centralized water system. However, the interviews and workshop results also suggest that the study participants see a blend of centralized and decentralized strategies as essential in the broader quest to improve the post-earthquake water situation. We investigated whether relationships exist between the employment status, gender, age group, or household location of the participants and a) the degree to which they were reliant on decentralized sources and, b) the likelihood with which they expressed a preference for centralized water sources.