Dr Ash Shokri

Ashkan Shokri is a Geospatial Data Scientist based in Melbourne, Australia. He has more than 6 years of experience in developing predictive spatial data models and utilizing machine learning techniques. he has a Ph.D. in Civil Engineering from Monash University, where he worked on Multiscale data assimilation and its application to water availability prediction.

Ashkan has worked as a Research Scientist at CSIRO since January 2023, where he is developing a toolkit for evaluating water resources in a catchment river system under different climate change scenarios. He is automating the modelling system for multiple runs and conducting data aggregation and statistical analysis to provide insights into water resource management under different climate scenarios.

Previously, he worked as a Geospatial Scientist at the Australian Bureau of Meteorology (BOM). During this time, he developed three operational services for BOM. He was a scientific developer for Australian Landscape Water Balance version 7 (AWRA-L v7), where he significantly improved the modelling system through advancements in theoretical equations, implementation of advanced spatial calibration, and introduction of new datasets. The model converts multiple climate spatial data streams to water balance daily maps and serves as the core of the seamless Australia Water Outlook (awo.bom.gov.au) and several downstream services.

Ashkan also designed and implemented an ML-based operational predictive model to forecast fine-resolution grassland curing maps forecasting up to 3 months ahead and serves the Australian Fire Danger Rating System (AFDRS). He implemented a complex statistical and Bayesian method using Ensemble Kalman Filter (EnKF) to improve water balance accuracy in the AWRA-L geospatial model through assimilation of GRACE satellite observations.

Prior to joining the Bureau of Meteorology, Ashkan worked as a Software Developer at Monash University, where he developed a model to simulate a water market by creating an API to iteratively launch a river system model, eWater SOURCE, to simulate stakeholder decisions. He also developed a long-term water resource planning model in MATLAB for Urmia Lake, Iran, to simulate recovery scenarios under different climate change scenarios.

Ashkan is proficient in Python, MATLAB/Octave, and associated libraries and frameworks for data analysis, visualization, and web development (Flask, Django, React). He is familiar with R and JS and has experience with data pre-processing, handling big spatial data, and SQL and NoSQL databases. In addition, he is experienced in machine learning frameworks such as Scikit-learn, distributed computing with Dask, and data preprocessing libraries such as Pandas, NumPy, Scikit-learn, Xarray, and Geopandas. He is also skilled in statistical analysis, optimization techniques, including linear programming, dynamic programming, stochastic dynamic programming, Monte Carlo methods, and heuristic optimization, and data assimilation techniques, including Kalman filtering techniques (ensemble, extended), triple collocation, and particle filtering.