Introduction

This exercise uses a training version of the Botswana National Water Resource System model built in the Pywr modelling software and visualised and run in the Water Strategy interface. The model is able to be run under a number of different scenarios including different domestic demand scenarios composed of per capita consumption and population growth, agricultural and mining demand growth, groundwater availabiliy and climate change impacted hydrological scenarios. All together, up to 72,032 scenario combinations of water supply (river flows and ground water aviality) and multi-sector demand scenarios are considered for the years 2030, 2060 and 2080. Each decade is represented by timeslices of 30 years. The model is run using monthly flow time-series on a 2 day time step. The use of a 2-day time step allows for storage in the system to be updated at a faster frequency than a monthly time step would allow.

The model includes different interventions which can be activated or left inactive. These include:

  • Chobe-Zambezi transfer which supplies the North South Carrier (NSC) System as welll as an agricultural scheme and offshoots to the west and Maun as well as smaller demands on its way to the south.W

  • Walvis Bay Desalination from Namibia

  • Losotho Highalnds Transfer

  • Non-revenue Water Reduction

  • Brine deslination

  • Wellfield Development

  • Mosetse dam

  • Floating Solar Panels on Reservoir

  • Waste Water Reclimation

  • Stormwater Harvesting

  • Additional NSC Pipeline capacity Expansion

The training model includes a subset of of these options.

The different combinations of future possible interventions (portfolios) can be run over a the full range of future uncertainty on a computer cluster and the performance over the range of uncertainty quantified to aid in strategic planning and decision making under uncertainty.

Analysts can also run the key portfolios of interest on the Water Strategy interface over selected future supply and demand scenarios and analyse system performance on a more granular scale.

Last updated