Water management: new bias correction tool improves modelling for water resources

UNSW Sydney

Managing water effectively through climate change relies on accurate climate modelling.

A novel bias correction tool that improves the accuracy of climate modelling will help us adapt our water management to climate change.

The software, developed by UNSW hydrologists, provides engineers, water managers and impact modellers more consistent data on projected climate impacts on water infrastructure and systems, says Scientia Associate Professor Fiona Johnson from UNSW Engineering.

“Australia's climate is highly variable with multiple climate drivers [that influence weather patterns], leading to frequent floods and extended periods of droughts,” the Director of the UNSW Water Research Centre says.

“As our climate changes with increasing temperatures and shifts to rainfall patterns, water availability is impacted.

“Australia's water policy and infrastructure investment rely on hydroclimate modelling to make informed decisions that promote resilience.”

The tool looks at multiple climate drivers, such as cycles of El Niño and La Niña, and their relationships to each other in time, correcting systematic biases within the modelling.

“Our methods adjust the estimates of rainfall and evapotranspiration [processes that move water from the Earth’s surfaces to the atmosphere] to support better modelling of river flows and therefore planning for water-dependent industries,” A/Prof. Johnson says.

The Bureau of Meteorology (the BoM) has implemented the software in the Australian Water Outlook, opens in a new window that provides the first nation-wide projections of climate change impacts on our water resource.

The interactive website, launched in 2023, provides comprehensive climate data at a five-kilometre resolution, including soil moisture, rainfall (as well as sleet, hail and snow), runoff, evaporation, evapotranspiration and drainage.

“The information enables more effective management of water availability, agriculture and planning for potential hazards, such as flood and drought,” A/Prof. Johnson says.

“It allows us to make more informed and resilient decisions around where and when we use water.”

The Australian Water Outlook produces calculations across past, present and future timescales: historical (daily outputs); seasonal forecasts (monthly updates spanning one to three months); and projections (for aggregated periods until the year 2100).

Its climate projections can be input into user-specific models for policy makers, farm managers, water and natural resource managers, researchers and members of the public.

Within the Australian Water Outlook, the UNSW bias correction software integrates with a range of climate inputs, static grids and near-real time satellite observations and additional downscaling techniques used to create more detailed regional modelling.

Integrating cycles of El Niño and La Niña into climate modelling

Climate modelling helps us understand changes in the global climate as well as examining alternative future trajectories based on variables, such as emission reduction versus business as usual.

“However, these models struggle to represent rainfall well for a number of reasons leading to biases at a number of different time scales,” A/Prof. Johnson says.

The modelling divides the world into grids, conducting calculations based on areas of 50-100 square kilometres.

“If you think of the difference between the weather in Sydney CBD and Western Sydney… you're trying to represent all that [diverse weather] in one box.”

This smooths out the variations that we see in reality, such as increased rainfall, caused when air masses flow over higher land. Additionally, modelling cannot presently calculate how much water is flowing down a river.

“To calculate an estimate of flow, we take the rainfall data from climate models and put it through our existing hydrologic models, making a series of adjustments. But if the rainfall data from the climate model is incorrect, the estimate of flow will be out,” A/Prof. Johnson says.

While many bias correction tools have been developed over the last 20 years, UNSW’s tool developed by A/Prof. Johnson, Professor Ashish Sharma and Dr Raj Mehrotra was world-first in its focus on interannual variability of rainfall, looking at rainfall variations that occur across multiple years.

“Rainfall varies a lot from year to year. And the interesting thing is it varies in clusters, for example, we’ll have multiple wet years – even 30 years where average rainfall is higher than normal – followed by a drier epoch.”

The software also adjusts the rate of evapotranspiration in relation to rainfall and its patterns in time, A/Prof. Johnson says.

“This allows the modelling to take into account the wetness of the catchment in advance [of rainfall] – they have this memory. It’s not just about the rainfall itself but what happened before.

“If you don’t integrate those patterns, the modelling can show you have a lot of water, but the reality is a different story.”

Examining the Murray-Darling Basin climate through projected data and tree rings

Data from the Australian Water Outlook has also been used to compare the severity of projected and historic droughts in the Murray-Darling Basin (MDB) dating back 800 years.

“Managing water security and sustaining ecosystems under future warming poses substantial challenges for semi-arid regions,” says UNSW research associate Dr Philippa Higgins.

“The Murray-Darling region is particularly vulnerable given the considerable demand for water that underpins its agricultural production and contribution to the national economy.” 

Dr Higgins reconstructed the Murray River streamflow by analysing paleorecords, natural records found in fossil pollen, ocean and lake sediments, corals and, in this case, tree rings that aid the study of past climates.

The properties of tree rings can be affected by climate, with wet years producing wider rings.

While most Australian trees have evolved to withstand our variable climate, Dr Higgins examined tree rings from neighbouring regions with strong climate teleconnections (significant relationships or links between weather phenomena) to the MDB.

“Because the climate drivers of MDB drought are global, I could examine a network of tree rings from Asia into South America and other areas affected by El Niño and La Niña to reconstruct these historic flows in the Murray River,” she says.

“We found that the Millennium Drought, which occurred in the 2000s, was the most severe in terms of duration, magnitude and peak [streamflow deficit] during the 800-year reconstruction. However, the data from the Australian Water Outlook projected an increase in future drought severity,” Dr Higgins says.

These results highlight the need for water management strategies to incorporate past, present and future modelling, says A/Prof. Johnson.

“Promoting resilience under future warming will require a robust water security policy and modelling to better understand current and future risks and guide climate change scenarios going forward.”