Quantifying runoff in arid zone basins of central Australia | Natural Hazards Research Australia

Quantifying runoff in arid zone basins of central Australia

Photo: USGS, Unsplash
Project type

Associate student research

Project status

In progress

This student project is quantifying water runoff in arid zone basins of central Australia, focusing on Lake Eyre Basin as a case study, using machine learning and satellite data to improve flood preparedness.

Project details

In view of recent dramatic floods and droughts, detection of trends in the magnitude of long time series of flood data is of scientific interest and practical importance. It is essential for the planning of early warning system from flood protection, where system design is traditionally based on the assumption of stationary in hydrological processes such as river stage and discharge. This monitoring can be done by installing an adequate number of gauge stations. However, the number of gauging stations is limited and declining in many parts of world. The sparsely distributed networks of gauging stations and data sharing problem poses a considerable challenge to obtaining long time series of discharge data. Despite these difficulties, remote sensing has offered a supplementary discharge estimation technique.
This project will focus on Lake Eyre Basin as a case study. The Basin has seen a number of significant floods in recent years, including in 2010 and 2011, which caused hundreds of millions of dollars in damages. This problem of flooding in the the Basin is caused by a combination of factors, including an increase in rainfall due to climate change. Recent studies have shown that time series of discharge is required for the Basin to relate the biological responses to hydrological events in the rivers. Therefore, in-depth knowledge of discharge estimations on the Basin is essential.
In general, remote sensing-based discharge estimation methods use empirical rating curve equations between observable satellite parameters and in situ measured discharge. Since these methods rely heavily on single hydraulic variables or focus on sole data sources, which does not have better accuracy. To address these issues, this project is using a combination of multiple hydraulic parameters for the rating curve development as an alternative. The result will be validated against the stream flow of the available gauging locations.

Another problem is these methods require a minimum of one gauging station to establish the empirical relation. This project is implementing machine learning method and multi-source satellite data to measure discharge measurement for ungauged basins. This result will be compared with the hydrological models result. All the derived discharges can also be used to perform a trend analysis on extreme hydrological events.