Estimation of disaggregated canal water deliveries in Pakistan using geomatics

Lack of accurate information on water distribution within an irrigation system is a major roadblock for effective management of scarce water resources. Numerical techniques to estimate canal water distribution require large amounts of data with respect to hydraulic parameters and operation of the hydraulic structures; such data are absent in large irrigation systems. In this study, six quantitative geomatic models, with varying complexity and physical representation, were formulated. They all use high-spatial resolution Landsat-7 images and a canal network stored in a GIS. The models require only the flow rates at the head of the distribution system to calculate downstream water distribution. Results were compared with discharge measurements for three tertiary canals (watercourses) within two selected secondary canals (distributaries), Gajiana and Ghour Dour, in Rechna Doab, Pakistan. The model that computed canal water distribution using the shape of the irrigated area was the most accurate at the watercourse scale. The overall deviation was 28% and the standard deviation of differences between modelled and measured flow was 10%. The model that relates canal water distribution directly to normalized difference vegetation index (NDVI) has the lowest accuracy with a 32% absolute deviation and 16% standard deviation. We conclude that geo-information techniques can be used to compute water distribution in large irrigation systems with a minimum of field data if an appropriate geomatic model is selected