Flood Predictions In The Seasonally Arid Tropics

In recent years Australia and many other parts of the world have faced extreme and extended adverse weather conditions. These events, coupled with increased industrial, agricultural and urban development on floodplains, have amplified strain on both physical and social environments.

Climatic extremes in the seasonally arid tropics are notoriously difficult to interpret and hence predict. The only rule that can be applied is that nothing is average.  Communities living in Australia’s seasonally arid tropical environments have adapted to a seasonal climate of long periods of negligible rainfall and moderate temperatures followed by high temperatures, humidity and variable rainfall.  More irregular events, including periods of extreme heat and short periods of intense rainfall often leading to flooding, pose significant challenges to both environment and community. The unpredictability of such events hinders effective warning systems and relief during and after the events.

Through integration of soil composition, river flow patterns, and real-time flood hydrological data, we accurately modelled the Central Queensland 2008/09 flood event that severely impacted the Bowen Basin coal mining industry.  An outcome of our study suggests that recent flood classifications of 1:100 events are not necessarily applicable to seasonally dry tropical settings.  Rather the rainfall events fall within the uncertainty limits surrounding 1:50 events.

The Emerald 2008 flood resulted from one, catchment-scale, extremely intense rainfall event. By contrast, multiple smaller localised events across multiple saturated catchments caused the 2011 flooding.  These numerous smaller events, through a combination of timing, catchment topography, land-use changes, vegetation clearing and urban growth, produced a confluence of runoff river peaks that resulted in widespread flooding and considerable damage. The contrast between these two floods highlights how little is understood about the causes of extreme events and the interplay between their numerous formative factors.

C&R Consulting have incorporated years of research in observation and statistical knowledge to produce the software package Monte Carlo Rainfall Simulator (MCRS) to assist in flood predictions.  This software is capable of taking observational rainfall data, and produces years of additional rainfall data that is similar to the original.  In regions where poor rainfall data is available, spatial interpolation techniques such as SILO can be used to create a base dataset.  This software can be used to analysis risk based on a strong probabilistic model, with significantly reduced uncertainties.

The hazards and risks associated with flooding are enormous. The devastation associated with flooding events throughout central and south-east Queensland, northern NSW and Victoria in recent years are potent examples.  Isolation of communities; loss of life (both human and animal); loss of income; damage and destruction of homes, businesses, agricultural and mining infrastructure; health risks, are all realities of extreme flood events.

C&R identified widespread inadequacies in available data suitable for meaningful and realistic calculations of Average Recurrence Intervals (ARIs) and Average Exceedance Probabilities (AEPs).  Primarily this is due to the lack of appropriate long-term rainfall and stream discharge data, but is compounded by the application of standard across-the-board modelling approaches to data sets unsuited to the chosen model and/or specific climatic event and is exacerbated by a lack of understanding of the scale and distribution  in which rain actually falls, the lack of rainfall and river gauging stations, radar stations, and the inability of most modellers and planners to accept and acknowledge the value of qualitative and non-official recordings (i.e. landholders’ recorded rainfall and physical flood height indicators).

From our experience in flood and catchment analysis in the seasonally arid tropics over the last decade, we advocate the need for a regionally appropriate, whole-of-system approach to understanding the dynamics of extreme flood events. Addressing all relevant environmental and infrastructure parameters that may impact, and in turn be impacted by, extreme climatic events in flood modelling is crucial to improving predictive models.  Integration of local knowledge is the key to timely warnings and response.  In the Emerald 2011 event, C&R’s knowledge of catchment dynamics, flow velocity and rainfall timing and a willingness to integrate local ‘non-official’ historical data, allowed for accurate flood height prediction to within a few centimetres, four days in advance of official reports. The consequent evacuations of residents, land holders and commercial properties significantly diminished adverse human impacts.

Accurate flood prediction in the seasonally arid tropics requires much more realistic, regionally appropriate integration of local environmental parameters. Importantly, this includes appropriate statistical treatment of non-regular climatic datasets and an understanding that use of ARIs and AEPs are largely inappropriate in this setting.

Australasian Mine Safety Journal.