RESUMO
People have used rainwater harvesting (RWH) technology for generations to a considerable extent in semi-arid and arid regions. In addition to meeting domestic needs, this technology can be utilized for agricultural purposes as well as soil and water conservation measures. Modeling the identification of the appropriate pond's location therefore becomes crucial. This study employs a Geo Information System (GIS) based multi-criteria analysis (MCA) approach and satellite rainfall data, Global Satellite Mapping of Precipitation (GSMaP) to determine the suitable locations for the ponds in a semi-arid area of Indonesia, Liliba watershed, Timor. The criteria for determining the location of the reservoir refer to the FAO and Indonesia's small ponds guideline. The watershed's biophysical characteristics and the socioeconomic situation were taken into consideration when selecting the site. According our statistical analysis, the correlation coefficient results of satellite daily precipitation were weak and moderate, but the results were strong and extremely strong for longer time scales (monthly). Our analysis shows that about 13% of the entire stream system is not suitable for ponds, whereas areas that are both good suitability and excellent suitability for ponds make up 24% and 3% of the total stream system. 61% of the locations are partially suited. The results are then verified against simple field observations. Our analysis suggests that there are 13 locations suitable for pond construction. The combination of geospatial data, GIS, a multi-criteria analysis, and a field survey proved effective for the RWH site selection in a semi-arid region with limited data, especially on the first and second order streams.