Application of a Genetic Algorithm in a Collaborative Process to Resolve Hydrology and Physical Reality with Both Western and Maori Cultural Values

Lakes Rotorua and Rotoiti are two sizeable, culturally and economically important lakes on the North Island of New Zealand. Rotorua outflows traverse the short Ohau Channel before entering Lake Rotoiti. Ohau channel flows are partially controlled by a stoplog structure. Rotoiti outflows to the Kaituna River are fully controlled by the Okere Gate structure. The structures are managed by Environment Bay of Plenty (EBOP), a government agency. Management objectives include maintaining minimum lake levels to support recreational boating, restricting maximum lake levels to avoid residential flooding, minimum instream flows below the lower lake to maintain aquatic ecosystems, limits on maximum releases to control erosion and prevent flooding. In addition, management seeks to provide for a minimum annual variation in lake levels to control the growth of aquatic plants in the littoral zone and to periodically expose beaches with important cultural value to the indigenous Maori population. The levels necessary to expose beaches may be lower than the minimum levels desired to support boating. Records of beach exposure are scant; the existence of beaches may depend on climate cycles. There is flow dependent recreational rafting below Okere Gates. This rafting is economically valuable, but is also contentious because the reach of Kaituna River flows through important Maori cultural areas, including grave sites. The Maoris have expressed a preference for replacing the Okere Gates with a fixed stepped weir, although the existing gates can be relatively easily operated to reproduce the flows over any of the fixed weir designs so far proposed. HydroLogics created a model of the two-lake system using its OASIS software system. The inflows to the lakes were estimated based on available historical flow and lake level data and on flow estimates derived from Mike-11 modeling of historical lake outlet configurations. A custom genetic algorithm (GA) was created to “wrap” the simulation model and develop well-performing parameter sets for alternative forms of operating rules. Initial results from the genetic algorithm created operating rules which significantly outperformed previous options on all of the intial objectives; those objectives may not fully reflect the objectives of the Maoris, however. A collaborative process has begun which must reach a conclusion (hopefully a consensus) by mid November. A ten year climatic cycle is hypothesized to exist in the basin. The effectiveness of a particular operating rule can be improved by modifying its form and parameters to follow the climatic cycle or in response to climate change driven changes. The ability to adapt to cyclic changes may be included in the forms of operating rules searched by the GA. The presentation will briefly describe the problem setting, the structure of the custom GA and OASIS model, and the results of the collaborative process.