UC Berkeley

Access to electricity remains an impediment to development in many parts of the world, particularly in rural areas with low population densities and prohibitive grid extension costs. In that context, community-scale run-of-river hydropower – micro-hydropower – is an attrac- tive local power generation option, particularly in mountainous regions, where appropriate slope and runoff conditions occur. Despite their promise, micro hydropower programs have generally failed to have a significant impact on rural electrification in developing nations. In Nepal, despite very favorable conditions and approximately 50 years of experience, the tech- nology supplies only 4% of the 10 million households that do not have access to the central electricity grid. These poor results point towards a major information gap between techni- cal experts, who may lack the incentives or local knowledge needed to design appropriate systems for rural villages, and local users, who have excellent knowledge of the community but lack technical expertise to design and manage infrastructure. Both groups suffer from a limited basis for evidence-based decision making due to sparse environmental data available to support the technical components of infrastructure design.

This dissertation draws on recent advances in remote sensing data, stochastic modeling techniques and open source platforms to bridge that information gap. Streamflow is a key environmental driver of hydropower production that is particularly challenging to model due to its stochastic nature and the complexity of the underlying natural processes. The first part of the dissertation addresses the general challenge of Predicting streamflow in Ungauged Basins (PUB). It first develops an algorithm to optimize the use of rain gauge observations to improve the accuracy of remote sensing precipitation measures. It then derives and validates a process-based model to estimate streamflow distribution in seasonally dry climates using the stochastic nature of rainfall, and proposes a novel geostatistical method to regionalize its parameters across the stream network. Although motivated by the needs of micro hydropower design in Nepal, these techniques represent contributions to the broader international challenge of PUB and can be applied worldwide. The economic drivers of rural electrification are then considered by presenting an econometric technique to estimate the cost function and demand curve of micro hydropower in Nepal. The empirical strategy uses topography-based instrumental variables to identify price elasticities.

All developed methods are assembled in a computer tool, along with a search algorithm that uses a digital elevation model to optimize the placement of micro hydropower infrastruc- ture. The tool – Micro Hydro [em]Power – is an open source application that can be accessed and operated on a web-browser (http://mfmul.shinyapps.io/mhpower). Its purpose is to assist local communities in the design and evaluation of micro hydropower alternatives in their locality, while using cost and demand information provided by local users to generate accurate feasibility maps at the national level, thus bridging the information gap.