OPTIMAL DESIGN OF PRECIPITATION GAUGE NETWORK USING BAYESIAN FRAMEWORK

Abstract:

A clear understanding of precipitation patterns is essential for planning expansion of an existing precipitation gauge network in a river basin. An impediment in this task is sparsity of gauges, paucity of observations, variability in precipitation pattern at different spatial and temporal scales, and their link to global/regional changes in climatic conditions and land use/landcover patterns. Conventional gauge network design procedures cannot account for several of these issues, including large-scale variability and non-stationarity in precipitation. Addressing these issues becomes more challenging when study area has undulating terrain and extends over hundreds of thousands of square kilometers. To address this, a novel two-stage methodology is developed in Bayesian framework. The first stage involves use of Bayesian approach to identify homogeneous precipitation patterns in the study area based on structural and shape characteristics discerned from precipitation records and predictor variables influencing precipitation. The second stage uses fuzzy entropy approach to prioritize gauges in different parts of the study area. The discerned information forms the basis to recommend decommissioning/expansion of the existing gauge network to meet required standards on gauge density. The methodology suggests use of satellite/space and ground-based observations to arrive at potential locations for installing new rain gauges in an expanded network. Other advantages of the proposed methodology include ability to account for non-stationarity in precipitation characteristics and scope for convergence to global optima in discerning precipitation patterns. Potential of the proposed methodology is demonstrated by application to precipitation network comprising 1128 gauges having more than 20 years of record in Karnataka state (191,791 km²) of India. The study is of significance as the proposed framework offers scope for design of gauge network recording various hydro-meteorological variables (e.g., precipitation, temperature, wind speed, humidity, solar radiation), whose reliable observations are essential for studies focusing on different applications such as sustainable agricultural water management, detection of climate variability, and forecasting of floods and droughts.