![Watering a rooftop vegetable patch GIZ](/sites/default/files/watering-vegetables.jpg)
![Watering a rooftop vegetable patch GIZ](/sites/default/files/watering-vegetables.jpg)
A hydroponic technology system was set up on the rooftops. The model consists of 3–4 water beds, directly attached to on each rooftop. They are made of wooden frames, plastic sheets, foam panels and cups filled with peat moss and pyralite substrate. The 15-cm deep water is supplied by a water pipe through an electricity connection from downstairs and maintained by a water pump and water filter. As an alternative, the project installed boxes filled with soil, having the advantage that no electricity is needed for water circulation. The biggest challenges for the rooftop farmers were irregular water supply and electricity cuts, as well as the hot summer temperatures which negatively impacted the growth of the plants.
The choice of which crop to grow heavily relied on the feasibility and market study; finding the right balance between a crop that has a high demand in the market and used most by the people, a crop that yields high returns to strengthen income generating opportunities, and a crop that can withstand environmental and climatic conditions of the area.
Monitoring: A monitoring system should gather data on e.g. food production, income, farming activities, economic costs/benefits, etc. Measurements should be conducted in order to assess the actual microclimatic impact. Technical setup: The hydroponic model is relatively cheap and thus affordable. However, other factors (maintenance, water and electricity consumption, usage of organic waste and grey water, impact on microclimate conditions) should be considered for upscaling. Other technical options aside from the hydroponic system should be explored. Remaining technical bottlenecks have to be tackled by on-site experimentation. Impact: the microclimate rooftops should be protected from direct solar radiation, enhance cooling-off through evaporation and enable a higher effect by spatial clustering and a bigger implementation area.