A machine learning based estimation method of beach slopes at a national scale: a case study of New Zealand

Beach slope is a critical parameter for understanding coastal geomorphological dynamics, yet the acquisition of comprehensive datasets at large scales remains a significant challenge. This study bridges this gap by presenting a novel methodology for estimating beach slopes across New Zealand’s sandy coastlines. We developed robust coastal slope estimation models for sandy beaches by integrating 12 environmental factors with high-precision LiDAR-derived slope data, employing four machine learning regression techniques: Random Forest (RF), Gradient Boosting Decision Tree (GBDT), eXtreme Gradient Boosting (XGBoost), and Category Boosting (CatBoost). These models were trained on datasets from 1,241 beaches with LiDAR-derived Digital Elevation Models (DEMs) and subsequently applied to predict coastal slopes for an additional 509 beaches lacking LiDAR data. The results reveal that the XGBoost model outperformed the others, achieving the highest accuracy with an R2 of 0.93 and an MAE of 0.02, demonstrating the effectiveness of machine learning in coastal slope estimation. This innovative approach, leveraging DEM datasets and environmental variables, provides a robust and cost-effective tool for estimating coastal slopes across global sandy beaches compared to high-cost field measurement methods. We also emphasized that our method can estimate beach slopes for beaches without topography data based on constructed machine learning methods and environmental factors. Future studies should focus on incorporating additional environmental covariates, and extending the model’s applicability to diverse coastal environments, thereby enhancing its predictive accuracy and utility, supporting sustainable coastal development worldwide.