By Donna O. Burnett

University of Montevallo

Abstract

Community-level water filtration systems depend on sustainable water sources and filtration methods. Satellite remote sensing provides a cost-effective methodology for identifying potential groundwater sources. The objectives of the study were to: (a) map elevation data derived from the ASTER Global Digital Elevation Model (GDEM), (b) analyze land cover using ASTER data, (c) analyze lineaments and surface dips using Landsat 7 ETM+ Band7 data, and (d) analyze soil and rock type using ASTER and Landsat 7 ETM+ data. Material used included (a) ASTER 15m pixel resolution satellite imagery of Port au Prince, Haiti, and surrounding areas acquired on Jan. 21, 2010, (b) ASTER Global Digital Elevation Model (GDEM) 30m pixel resolution data acquired from 1999-2008, (c) Landsat 7 ETM+ 30m pixel resolution data acquired circa 2000, (d) public use shapefile data for Haiti’s population and terrain, (e) ESRI ArcGIS 9.3 software, and (f) ERMapper 7.1 software. In the present study, areas were identified that are rich in carbonate and silica-type alluvium and are at water level. These areas may be suitable for shallow wells. Areas displaying carbonate, and silica, and quartz-rich outcroppings within 250m of (but not directly adjacent to) linear features are prime areas for ground water exploration and potentially productive wells.

I. INTRODUCTION

The United Nations Development Programme ranks Haiti at 148 of 179 countries according to their Human Development Index with a per capita Gross National Income (GNI) in 2008 of $660 U.S. About 56% of the population lives on less than $1 per day, and another 25% on less than $2 (1). Sustainable access to safe drinking water is a public health issue directly linked to the economy in Haiti and around the world.

It is estimated that investment of $23 billion per year is needed to improve water sources for 1.5 billion people and sanitation for 2.2 billion (2). Given the expense of establishing infrastructure, governmental agencies, non-governmental organizations (NGOs), and nonprofit organizations implement point-source filtration or disinfection treatments through household and community-level interventions in an attempt to mitigate adverse health consequences related to consumption of contaminated water. Sustainability is a critical indicator of intervention success and is especially important when interventions are funded by contributions of donors to nonprofit organizations. Sustainability efforts must target water supply and system operations. The purpose of this research is to apply remote-sensing techniques to data from ASTER satellite imagery to produce maps useful to decision-making in placing community-level water-filtration installation systems in Port au Prince and surrounding areas.

II. BACKGROUND

In 2009, Haiti reported an infant mortality rate of 46.7 per 1,000 live births (lb), an improvement over the 57 per 1000 lb rate for 2006. In 2008, reports indicated an under-5 year mortality rate of 72 per 1000 lb, an improvement over the 151 per 1000 lb rate for 1990 (1). For 2003-2008, Haiti reported prevalence rates for moderate and severe underweight in children under 5 years of 18%, moderate and severe wasting at 10%, and moderate and severe stunting at 29% (3). Underweight, wasting, and stunting in children are potential functions of inadequate nutritional intake, inadequate nutrient absorption/utilization secondary to diarrheal illness, or both (4) (5). Diarrheal illness accounts for 10-80% of retarded growth in infants and very young children (5). In their 2003-2008 report, the Pan American Health Organization released prevalence rates of 43% for children under 5 years with diarrheal illness receiving continuous feeding and oral rehydration (3). The World Food Programme reported the prevalence of school-aged children infected with intestinal parasites at 32% (1). Bacterial and parasitic contaminants in surface water, groundwater and water obtained from Haiti’s broken water infrastructure contribute to diarrheal illness and dysentery, cholera, typhoid, and hepatitis (6). In total, the Pan American Health Organization estimated that water-borne diseases, especially diarrheal illness, resulted in half of Haiti’s total deaths in recent years, excluding those resulting from natural disasters (6).

The World Health Organization (WHO) and UNICEF published statistics from 2008 reporting 71% of urban residents in Haiti with access to improved drinking water sources and 55% of rural residents with the same access, for an overall rate of 63% (7). For the same year, WHO/UNICEF reported 24% of urban residents with access to improved sanitation and 10% rural, for an overall rate of 17% for access to improved sanitation for the country. In total, 47% of the population reportedly resided in urban areas in 2008, indicating an annual growth rate of 5.1% for urbanization. While drinking from improved water sources from infrastructure does not guarantee safety as evidenced by water testing, it does improve sustainable access to water gained from filtration systems running from this water source. When infrastructure is not available, Haitians depend on springs, rivers, other surface waters, and wells, including tubewells.

Community-level water filtration systems serving small villages require a consistent supply of about 600-800 gallons of water per week to yield approximately 10 gallons of safe drinking water per week, per household. Health education for hygiene and sanitation, including hand washing and use of latrines when possible, is imperative to maximize positive health impacts and outcomes (8) (9). Water filtration systems depend on system operators trained to maintain the system and system administrators to oversee the operational and financial stability of the system. Often, micro-businesses, formed around the water filtration system, generate sufficient capital to replace system parts and compensate system operators. Figure 1 identifies key environmental and anthropomorphic sustainability issues.

Remote-sensing data can provide information about location, quality and potential availability of surface and ground water through analysis of watershed health, surface water modeling, and ground water exploration (10) (11) (12) (13) (14).

A. Watershed Health and Groundwater Potential

Previous water assessments conducted in Haiti provide important information about surface and groundwater sources (6) (15) (16) (17) (18). Groundwater from unconsolidated sediments is usually accessible using simple pumps and shallow wells. Drury and Deller indicate the safest ground water sources are from wells, due to the self-filtering action of aquifers (11). The filtering action and oxidizing micro-organisms essentially destroy the majority of pathogens when surface water infiltrates the aerated zone. Dissolved toxic minerals constitute a health risk, and advance remote-sensing analytical techniques can be employed to detect this risk (11).

Griffith describes the relationship between land cover and water quality, indicating that vegetation patterns are useful in predicting watershed health (12). Padgett-Vasquez identified that percentage land cover of non-vegetated areas within a watershed is predictive of stream health (19). Non-vegetated areas include impervious surfaces and bare ground, found to increase surface run-off and adversely affect watershed health. The present study uses ASTER data acquired by three sensor systems operated on the NASA Terra to produce a vegetation map of the study area delineating non-vegetated areas.

Sander, Minor, and Chesley published research from a study of wells in Ghana indicating that the 23 wells in the study area within 250m of a linear feature exhibited a 57% success rate (the wells closest to the linear feature were not as productive as those off from the lineament, possibly because of the presence of clay) versus the 17 wells located further away that yielded a 41% success rate (14). Other authors describe the usefulness of mapping linear features for groundwater exploration (10) (11) (13) (14) (16) (20) (21) (22) (23). The present study used Band7 of Landsat 7 ETM+ imagery to evaluate the Earth’s surface for lineaments and surface dips that point to potentially trapped ground water and GIS shapefiles to label known faults in the study area. Landsat 7 ETM+ data is registered on a passive sensor, the Enhanced Thematic Mapper Plus (ETM+) instrument. Data is collected in eight bands of electromagnetic data in the visible and near-infrared (VNIR) and infrared regions of the spectrum, calibrated to within 95% accuracy.

Soil types greatly influence movement of water within aquifers. Soils with higher proportions of carbonate and silica/quartz-type soils allow groundwater to flow freely through watershed aquifers as opposed to soils rich in clay that tend to clog the aquifers and obstruct the flow of groundwater (11). Kalinowski and Oliver compiled information describing the ASTER band ratios usefully in the identification of soil and rock types that have higher proportions of carbonate and silica in the study area (24).

III. MATERIALS AND METHODS

Materials used in the study included (a) ASTER 15m pixel resolution satellite imagery of the Port au Prince and surrounding areas acquired Jan. 21, 2010, courtesy of NASA, (b) ASTER GDEM 30m pixel resolution data acquired between 1999-2008, courtesy of the Ministry of Economy, Trade and Industry of Japan (METI) and NASA, merged into a mosaic for analysis, (c) Landsat 7 ETM+ 30m pixel resolution imagery acquired circa 2000 obtained through NASA’s Global Land Cover (GLC) project, (d) public use shapefile data for Haiti’s population and terrain, including population density, 15m sub-basin boundaries, and fault lines, (e) ESRI ArcGIS 9.3 software, and (f) ERMapper 7.1 software. ASTER Band3, registering in the VNIR range of the electromagnetic spectrum, provides GDEM data acquired 1999-2008 found to be accurate to 20m at 95% confidence vertically and 30m horizontally. ASTER imagery is referenced to WGS84/EGM96 (25).

The study objectives included the following: (a) map elevation data for Haiti derived from ASTER GDEM data, (b) analyze the study area for non-vegetated land cover using ASTER data, (c) analyze the study area for lineaments and surface dips using Landsat 7 ETM+ Band7 data, and (d) analyze the study area for soil and rock type using ASTER and Landsat 7 ETM+ data.

IV. RESULTS

The University of Alabama at Birmingham Laboratory for Global Health Observation computers running ERMapper 7.1 and ESRI ArcGIS 9.3 software were used for image analyses.

A. Elevation Data

Figure 2 encompasses Port au Prince and surrounding areas. Much of the Port au Prince area is at water level and on an alluvial floodplain (16). Figure 2 provides a zoomed-in image of the ASTER GDEM image subjected to log transformation and 11 X 11 kernel sharpening in ERMapper 7.1. Note the alluvial fans around Port au Prince and the alluvium in the floodplain joining the northern and southern portions of the country.

B. Land Cover

Figure 3 is an ASTER 321 image acquired Jan. 21, 2010, subjected to Normalized Difference Vegetation Index (NDVI) analysis in ERMapper 7.1. The image was further manipulated to identify non-vegetated areas (negative NDVI values) in white and vegetated areas in shades of green, with healthier vegetation in brighter shades of green. Very dark to black cells indicate positive NDVI values close to zero. Known faults and linear features identified on a GIS shapefile are mapped on a vector surface overlay.

C. Lineaments and Surface Dips

Results of analysis of the areas surrounding and including Port au Prince for terrain features of lineaments and surface dips using Landsat 7 ETM+ Band7 are displayed in Figures 4 and 5. Lineaments and surface dips are useful for locating trapped groundwater. Following methodology described by Ayday and GÌ_mÌ_ÙlÌ_o€Ùlu, Band7 was subjected to processing with eight seismic directional filters (N dips, NE dips, E dips, SE dips, S dips, SW dips, W dips and NW dips) layered in an algorithm (20). Principal component analysis (PCA) indicated that PC1 through PC3 explained 92.7% of the variance in the image. Figure 4 displays PC1 and Figure 5 displays PC2, overlain with known faults mapped as GIS shapefiles. Additional lineaments are noted in the image. Tables 1-3 and Figure 6 present results of the PCA. Special attention is given to places where surface lineaments and dips intersect.

D. Soil and Rock Type

Results of analyses of the areas surrounding and including Port au Prince for soil type and outcropped rock type using ASTER and Landsat 7 ETM+ images are displayed in Figures 7-9. Known faults and linear features identified on a GIS shapefile are mapped on a vector surface overlay.

Figure 7 is a Landsat 7 ETM+ RGB 453 band GeoCover compressed image circa 2000. Sediment-laden surface water appears blue. Wet soils are darker than dry soils due to the absorption of infrared waves. Known faults and linear features identified on a GIS shapefile are mapped on a vector surface overlay.

Figure 8 is a Landsat 7 ETM+ RGB 742 band GeoCover compressed image circa 2000. In this image, kaolinite (clay) would appear blue, if present. Known faults and linear features identified on a GIS shapefile are mapped on a vector surface overlay.

Figure 9 is a layered image. The top layer is an ASTER image displayed at 25% transparency, with the following band ratios: R = 14/12; G = 13/14; and B = 12/13 (24). This layer identifies carbonate-rich areas in turquoise and silica-rich areas in red. Quartz-rich alluvium or outcropped rocks may appear as red or blue. Unconsolidated sediment rich in carbonate, silica or quartz may be conducive to groundwater flow through potential aquifers. The second layer in the image displays NDVI results with vegetated areas in green and non-vegetated areas (negative NDVI values) in white, resulting in relatively brighter pixels representing bare soils, impervious surfaces, and water. The third layer in the image is PC1 of Landsat 7 ETM+ 2005 image GLS, demarcating linear features through the image. Known faults and linear features identified on a GIS shapefile are mapped on a vector surface overlay.

V. CONCLUSIONS

Results of this research indicate that freely accessible ASTER and Landsat 7 ETM+ satellite data available through NASA provide information potentially useful in the decision-making process for water filtration system installation. Potentially useful information includes watershed health indicators, soil and rock type, and lineaments/surface dips. It is noted that more complex algorithms and software are available to provide more specific information, such as surface water modeling, impervious surface determination, and linear feature modeling. However, the processes used in the present study were described in peer-reviewed articles and are suitable for first-level assessment designed to include or exclude particular areas for further investigation using more sophisticated methods. Ground-truthing is needed to verify that what is seen on the satellite imagery is indicative of what is on the ground. In the present study, areas were identified rich in carbonate and silica-type alluvium that are at water level. These areas may be suitable for shallow wells. Areas displaying carbonate, and silica, and quartz-rich outcroppings within 250m of (but not directly adjacent to) linear features are prime areas for potentially productive wells.

Implications for future research include the use of remote sensing to assess ground water toxicity levels and soil moisture, and application of ASTER imagery correction techniques to improve the quality of data analysis. Future studies are needed using lineament and surface dip data to predict productive wells.

Donna O. Burnett, PhD, RD, is an assistant professor of dietetics at the University of Montevallo, in Alabama. Dr. Burnett is a member of the development team for Living Waters for the World.

Acknowledgements: The author wishes to acknowledge the contribution of Dr. Jeff Luvall, Senior Research Scientist at NASA, and Dr. Sarah Parcak, director of the University of Alabama at Birmingham Laboratory for Global Health Observation.

References

1. World Food Programme. Countries: Haiti. Available online at http://www.wfp.org/countries/ haiti

2. The World Bank Group. Haiti. Available online at http://www.worldbank.org/

3. Pan American Health Organization. Health Surveillance and Disease Prevention and Control / Health Analysis and Statistics. Available online at http://www.paho.org/english/ad/dpc/ha-unit-page.htm

4. R. E. Black, “Would control of childhood infectious diseases reduce malnutrition?” Acta Paediatrica, pp. 133-140, 2008.

5. K. Brown, “Diarrhea and malnutrition”, Journal of Nutrition, Vol. 133, pp. 328S-332S, 2003.

6. J. A. Treaster, A drinking water crisis in Haiti long before earthquake destruction. Available online at http://trueslant.com/josephtreaster/2010/01/29/a-drinking-water-crisis-in-haiti-long-before-earthquake-destruction/

7. World Health Organization & UNICEF. Water, sanitation and hygiene. Available online at http://www.unicef.org/wash/index_statistics.html

8. Living Waters for the World. Available online at www.livingwatersfortheworld.org

9. S. P. Luby, et al., “Effect of Intensive handwashing promotion on childhood diarrhea in high-risk communities in Pakistan”, Journal of the American Medical Association, Vol. 291, pp. 2547-2554, 2004.

10. K. Bischoff, Examination of Water Drilling Success Rate Using Satellite Imagery in the Central Plateau Region, Haiti. Bowling Green : Bowling Green State University, 2008, pp. 1-149.

11. S. A. Drury, and A. Deller, Remote Sensing and Locating New Water Sources. Milton Keynes, UK : The Open University, 2003.

12. Jerry A. Griffith, “Geographic techniques and recent applications of remote sensing to landscape-water quality studies”, Water, Air, and Soil Pollution, Vol. 138, pp. 181-197, 2002.

13. S. Mutiti, J. Levy, and N.S. Gaturu. “Assessing ground water development potential using Landsat imagery”, Ground Water, Vol. 48, pp. 295-305, March-April 2010.

14. P. Sander, T.B. Minor, and M.M. Chesley, “Ground-water exploration based on lineament analysis and reproducibility tests”, Ground Water, Vol. 35, pp. 888-894.September-October 1997,

15. R.B. Knowles, et al., Water Assessment of Haiti. Mobile, AL : US Army Corps of Engineers Mobile District and Topographic Engineering Center, 1999.

16. E. Rathje, J. Bachhuber, and B. Cox. Geotechnical engineering reconnaissance of the 2010 Haiti earthquake. Available online at http://www.geerassociation.org/GEER_ Post%20EQ%20Reports/ Haiti_2010/Cover_Haiti10.html

17. Scot E. Smith, and Daniel Hersey, “Analysis of watershed vulnerability to flooding in Haiti”, World Applied Sciences Journal, Vol. 4, pp. 869-885, 2008.

18. World Health Organization. Progress on Sanitation and Drinking-Water 2010 update. s.l. : Author, 2010, pp. 869-885.

19. Steve Padgett-Vasquez, Tracking Landscape Changes in the Upper Cahaba River Watershed and its Tributaries (1974-2007) using Landsat and ASTER multispectral imagery. University of Alabama at Birmingham Master’s thesis. 2010.

20. C. Ayday and E. GÌ_mÌ_ÙlÌ_o€Ùlu, “Detection and interpretation of geological linear features on the satellite images by using gradient filtering and principal components analysis.” The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XXXVII, pp. 1207-1210, 2008.

21. L. Q. Hung, O. Batelaan, and G. De Smedt, “Lineament extraction and analysis, comparison of LANDSAT ETM and ASTER imagery. Case study: Suoimuoi tropical karst catchment, Vietnam.” Remote Sensing for Environmental Monitoring, GIS Applications, and Geology, Vol. 5983, pp. 1-12, 2005.

22. K. S. Kavak and H. Cetin. “A detailed geologic lineament analysis using Landsat TM data of Golmarmara/Manisa Region, Turkey”, Online Journal of Earth Sciences, Vol. 1, pp. 145-153, 2007.

23. Yoshiki Ninomiya and Bihong Fu. “Spectral indices for lithologic mapping with ASTER thermal infrared data applying to a part of Beishan Mountains, Gansu, China”, IEEE, pp. 2988-2990, 2001.

24. A. Kalinowski, and S. Oliver, ASTER Mineral Index Processing Manual. s.l. : Geoscience Australia, 2004, pp. 1-36.

25. ASTER GDEM Validation Team. Routine global digital elevation model. Available online at https://lpdaac.usgs.gov/lpdaac/products/aster_products_table/routine/global_digital_elevation_model/v1/astgtm

26. L. Berman. Geospatial data for relief and reconstruction efforts in Haiti: Demographics (3rd-level sections), Haiti [vector] population density, male/female population, number of houses. Available online at http://cegrp.cga.harvard.edu/haiti/?q=resources_data