​Learning and transferring groundwater level dynamics across alluvial aquifers. 

Project leaders: Dan Pagendam and Justin Wu

Project Description: The project aims to develop novel Machine Learning methodology to predict groundwater dynamics in alluvial aquifers in Australia and overseas making use of large volumes of satellite remote sensing data. The project also use of unconventional monitoring information from similar basins across the world to enable the assessment and prediction of groundwater resources in less monitored basins.

Project Goals:

1. Successful application of Machine Learning model for groundwater dynamics using remote sensing.

2. Development and demonstration of methods for transferring knowledge and information between groundwater basins to improve predictions.

3. Facilitate uptake of novel Machine Learning-based alternative methodology for groundwater assessment.

4. Developing insights about transferability of groundwater knowledge across basins.

Project outputs:

1. Paper on machine learning methodology for groundwater level predictions using remotely sensed data (see A log-additive neural model for spatio-temporal prediction of groundwater levels – ScienceDirect).

2. Machine learning code library for model implementation.

3. Methodology and algorithms for transfer learning of groundwater dynamics.

Interested to know more?

Please reach out to Justin Wu (CSIRO Environment) or Dan Pagendam (CSIRO Data61) for more information.