**5. Conclusions**

This study developed an innovative yet practical approach for rapid assessment of post-fire hillslope erosion risk using weather radar, remote sensing, GEE, and GIS. The automated scripts running in ArcGIS and GEE allow rapid processing of time-series and event-based spatial data including satellite and weather radar images to estimate the erosion risk for the 2019–2020 wildfires at SDWC.

This study was the first attempt to precisely estimate where and when high erosion risk is likely to occur at daily and/or hourly steps after severe wildland fires across large drinking water catchments, allowing for more accurate estimation of event-based erosion. With these timely estimates of rainfall erosivity and C-factor, along with the existing datasets on soil erodibility and slope-steepness, we were able to deliver rapid assessment of the post-fire hillslope erosion risk and link it with fire severity. These continuous and consistent estimates of erosion rates were used to analyze the erosion risk before and after the 2019–2020 wildfires and the subsequent impacts of rainfall on erosion rates. With these time series datasets, we identified the locations and times of the highest erosion risk. The sub-catchments near Warragamba Dam have the highest erosion risk because of the bushfires and rainfall events.

The rainfall erosivity is the dominant factor affecting hillslope erosion after severe wildfires over the SDWC area. Severe erosion events are often caused by short but intense storm events such as the case in February 2020. High temporal rainfall data are essential in rainfall erosivity modeling. Weather radar Rainfields data are adequate for erosivity or EI30 estimation at catchment and regional scales.

Field observations and measurements of soil erosion are necessary for model validation and improvement. Other relevant water quality models (such as Source developed by eWater Australia) may be used jointly to validate the modeling accuracy and predict likely pollution risk and delivery to the river or lake system after severe wildfires [48]. Over the next two-years we plan to collect more field data on erosion level, ground cover, land management activity, slope and slope length will be gathered along a set of transects at selected trial catchments for further calibration and validation of methods. A hillslope sediment erosion trap network, consisting of moderate and low burn severity, prescribed burn, rainfall event-based, and control sites are planned to be installed at SDWC sub-catchments along with the water quality stations, and maintained by Water NSW. The goal of the traps is to collect post-fire and unburnt erosion rates to validate model estimates. This will help further calibrate and improve the erosion model and link the estimated erosion rates with sediment transport and water quality downstream. The methodology, once fully validated, will be extended to other areas and wildfire events to provide timely and accurate information on erosion water quality immediately after wildfires.

**Author Contributions:** Conceptualization, X.Y.; methodology, X.Y, M.Z; software, M.Z.; validation, Q.R.O., L.O. and S.F.; formal analysis, X.Y. and M.Z.; writing—original draft preparation, X.Y.; writing—review and editing, X.Y., M.Z., Q.R.O., S.F., and A.R.; visualization, M.Z.; supervision, X.Y. and L.O.; project administration, X.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This project was managed through the New South Wales Department of Planning, Industry and Environment (DPIE) and collaborated with Water New South Wales (WaterNSW). Many DPIE and WaterNSW staff, particularly soil surveyors, contributed to this project and their effort is greatly appreciated. We especially thank the contribution and discussion from Lisa Hamilton, John Bickmore, Charity Mundava, Marlene Van Der Sterren from WaterNSW, and Linda Henderson, Jonathan Gray and Mark Young from DPIE. We also thank

Bureau of Meteorology for providing the radar Rainfields data, and CSIRO (Juan Guerschman) for providing the MODIS-derived fractional vegetation cover data.

**Conflicts of Interest:** The authors declare no conflict of interest.
