PiNC Lab researchers join new Data to Model (D2M) project
The People in Nature and Climate (PiNC) Lab, in collaboration with Rhodes University and others, is pleased to announce its participation in the Data to Model (D2M) project, with PiNC Lab’s Dr Petra Holden and Dr Assumpta Onyeagoziri as part of the team.
The D2M project aims to improve hydrological modelling practice to reduce uncertainty and better inform critical water management decisions in South Africa.
The project is funded by the Water Research Commission (WRC) and led by Dr Julia Glenday, a Research Associate at Rhodes University and the South African Environmental Observation Network (SAEON).
Uncertainty in hydrological modelling
Research by Rebelo, Glenday, Holden and colleagues, published in the journal, Ecological Modelling, highlighted how structural differences in hydrological models create uncertainty that is frequently overlooked in both research and applied cases.
“It is critical to actively consider this uncertainty for water security planning and decisions,” says Holden.
Indeed, a policy brief by the same authors makes the case for collaborative standardisation of uncertainty analysis and communication approaches in the sector, especially to improve decision making based on model output that involve nature-based solutions. Implementing this would require investment in capacity, time, tool development, and data accessibility.
Reality-check and refine models
The Data to Model (D2M) project is based on the assumption that hydrological modelling can be more accurate and less uncertain when data on multiple catchment processes are used to reality-check and refine models.
“Many current models rely mainly on streamflow data, which can result in significant uncertainty,” explains Onyeagoziri.
“Additional information such as evapotranspiration, soil moisture, and groundwater observations can be used together with field measurements and remote sensing to evaluate and calibrate models.”
These data are not yet routinely used, and methods for applying these data strategically in the modelling process are not well established. “As such, the project will provide practical guidance for improving model accuracy and reducing uncertainty,” Onyeagoziri says.
Informing practical guidelines
The project has identified three case study catchments to test what happens to model output when additional variables are used.
“Catchments in KwaZulu-Natal, Mpumalanga and the Western Cape were selected because they have relatively rich datasets compared to many other parts of the country,” says Onyeagoziri.
“They all have long-term streamflow and weather observations, as well as additional field measurements such as evapotranspiration, soil moisture, or groundwater levels.”
The project findings will be used to inform practical guidelines for using data on multiple hydrological variables to improve model realism and reduce uncertainty. This will also include suggested approaches for different cases of data availability, for example, with or without other field measurement data.
By reducing uncertainty in hydrological models, the project will support more robust water management decisions and help decision-makers better assess the potential benefits and trade-offs of nature-based solutions in catchments.
