Potential of simple filters to improve microbial quality of irrigation water used in urban vegetable farming in Ghana

Abstract

Irrigation water used for growing vegetables in urban areas in many low-income countries is contaminated with untreated wastewater. Many wastewater treatment methods are economically prohibitive and continued use of such irrigation water pose health risks for vegetable consumers and farmers. As part of a larger study on possible interventions for health risk reduction, the potential of simple interventions was explored. Column slow sand filters with three levels of sand depths (0.5 m, 0.75 m and 1 m) and fabric filters made of nylon, cotton and netting were assessed. More than 600 water samples were analyzed for helminth eggs and thermotolerant coliforms. Flow rates were also measured. From slow sand filters, 71–96% of helminths and 2 log units (from 7 to 5 log units) of thermotolerant coliforms were removed. Sand depths had no significant influence in the removal. Lower removal rates were achieved by fabric filters, with an average removal of 12–62% for helminth eggs and 1 log unit for thermotolerant coliforms. Nylon filters had higher removal rates especially for helminth eggs (58%). Average flow rates for sand filters were 3 m per day and fabric filters had steady flows of about 1.5 liters per second, but flow reduced with time in cotton filters. The simple filters tested improved the microbial quality of irrigation water and could easily be used in combination with other interventions to further reduce health risks. The unit cost of the filters tested also appear acceptable to farmers and some incentives like better prices will motivate many farmers to invest in such simple interventions.

Keywords:

• Slow sand filters
• fabric filters
• low-cost wastewater treatment
• urban agriculture
• microbial contamination

Acknowledgment

This work was supported by funding from the CGIAR—Challenge Program for Water and Food (PN 38), CIP—Urban Harvest and the International Water Management Institute. We acknowledge Mr. Maxwell S.K. Akple and Mr. Kwame Osei Boateng for their field and laboratory assistance.

Notes

^aFilter 1 had 0.5 m sand depth, Filter 2 had 0.75 m sand depth and Filter 3 had 1 m sand depth.

^bClogged filters.

^cStandard Deviation.