The development of a daily stochastic streamflow model for probabilistic water resource management
Date
2023-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: An ever-increasing water demand with limited supply of water in South Africa means that the focus of resource management needs to shift from a macro-level to micro-level. Well-defined research methodology regarding the management of larger water resource systems does exist in models such as STOMSA (Stochastic model of South Africa) and WRYM (Water Resources Yield model), which use monthly timesteps. In analysing smaller catchments, these macromodels need to be adapted to daily timesteps to enhance applicability in the management of
resource systems for smaller local authorities. This research focused on the development of a daily stochastic streamflow model to be used in small, single site catchments for resource management by local authorities in South Africa. Such catchments usually consist of abstraction weirs with off-channel storage dams that should deal with the effects of short runoff response time associated with small catchments, where the monthly timestep analysis typically cannot account for the short-term variability in daily streamflow. The methodology used in the current research focused on the generation of daily stochastic streamflow data by retaining the day-to-day relationship of the historical streamflow series without the reliance on disaggregation models to generate daily data from larger timesteps. This was achieved by implementing a Markov process, as the core element, to generate the stochastic data, based on the day-to-day relationship of the historical daily dataset. To address seasonality associated with daily datasets, the concept of daily duration curves was introduced, which served as both a normalisation process of the historical data, as well as a statistical distribution for the random selection of stochastic streamflow data. To ensure repeatability, a Pseudo-Random Number (PRNG) generator was used in the randomisation process of generating the stochastic datasets. The Daily Markovian Stochastic Streamflow model (DMASS) was developed consisting of four modules. The Pre-processing Module used primary streamflow data from the Department of Water and Sanitation (DWS) to generate the daily streamflow time series. The Analysis Module analysed the daily streamflow time series to create the Daily Duration Curves (DDC) and the Cumulative Transition Probability Matrix (CTPM). The Generation Module used the DDC and CTPM to generate the stochastic sequences. The Climate Change Module provided the option to adjust the DDC according to the selected adjustment parameters.
AFRIKAANSE OPSOMMING: Geen opsomming beskikbaar.
AFRIKAANSE OPSOMMING: Geen opsomming beskikbaar.
Description
Thesis (PhD)--Stellenbosch University, 2023.