Basin-wide water demand estimation using RGB color detection of built-up area from satellite imagery

The unavailability of as-built drawings and population data in primarily rural east central Java, Indonesia, presents a challenge to effectively estimate the total demand of a proposed large-scale retrofit water supply scheme. A simple RGB color detection method is employed to analyse satellite imageries extracted from Google Earth. The applicable range of RGB spectrum for typical rooftops is ascertained based on image inspection and the candidate areas identified are filtered, smoothened and manually cleaned-up to produce the total built-up area by pixel calibration. Water demand is then estimated based on assumed population equivalent per standard residential unit and consumption rate. The preliminary estimate is further corroborated with the most recent census and population growth rate. Result show that the approach taken is a cost-effective and efficient way for large-scale areal approximation.