Senior Lecturer, Babar Mahmood recently supervised final year Bachelor of Engineering Technology student Zhimin Zheng in a research project that used Artificial Intelligence (AI) to predict runoff in the Waikato River. Babar has been invited to present the findings at the 2018 Stormwater Conference this week.
This year the annual conference, held in Queenstown and attracting stormwater experts from across the globe, will explore the effects of stormwater on the environment.
Babar will discuss a model developed here at Unitec that measures the relationship between rainfall and runoff in New Zealand. It’s an exciting piece of work that could potentially be used to predict future floods.
Using data collected in the Waikato region, the team tested the new tool, which uses an AI technique known as Artificial Neural Network (ANN), and found that it could predict the amount of runoff during a period of rain. It’s a tricky thing to measure, taking into account factors such as rainfall duration and intensity, soil type, rate of evaporation, and other catchment characteristics.
“Environment Waikato provided data from previous years and we used that to train our model to forecast how much runoff would occur during periods of rainfall. The Council were very interested in our results, which could potentially help to predict and prepare for flooding in future”, says Babar.
Great project!
Hopefully Auckland City Council can use it also to prevent the flooding such as happened in New Lynn last year.
Exciting project indeed!
Although the New Lynn flooding was related to the intensity of the rainfall, the extent of the damage was significantly exacerbated by the failure of the culvert structure itself.
Thanks and true Cesar.
In this model, we used antecedent rainfall depth data to predict runoff/flow rates for a catchment. Recently, a study was undertaken in Australia where they use ANN model (using the gauged catchment data) to predict flow rates for ungauged catchments. This was an interesting article.
Thanks again Daniel and Cesar.