Debris is one of the most challenging cascading effects posed by hurricane events, causing large financial and logistical burdens to coastal communities. Moreover, disaster debris can have cascading consequences on the safety and functionality of infrastructure, inhibit community recovery efforts, and lead to public health concerns. However, existing debris predictive models have shown an unsatisfactory performance, with errors up to 50% in debris estimates, and only consider a limited set of predictive variables. Given the importance of debris in coastal community resilience, this study leveraged a convergent research strategy to propose a knowledge-informed data-driven methodology by developing a comprehensive database that expands across human-built–natural systems to inform a probabilistic data-driven model of debris volume. A Gaussian process model was used to generate the debris volume model from a debris removal database for Hurricane Ike and the human-built–natural systems predictors. Moreover, different spatial resolutions (500, 250, and 125 m) were tested to analyze their effect on the model performance. Results showed that the low-resolution (500 m) and the intermediate-resolution (250 m) models have the best performance with a normalized root-mean squared error (RMSE) of 0.49 and 0.50, respectively. These two models were then used to explore the relative variable importance of the predictive variables in the model in order to get insights on the drivers of the debris process and propose more flexible lower-dimensionality models. The influence of different predictors and the trade-offs of resolution and model performance were also discussed before demonstrating application of the model for a synthetic storm in the Galveston, Texas, region. The proposed methodology and the probabilistic estimates of debris quantities are key to develop comprehensive risk estimates of storm impacts on coastal communities and can support informed decision making and mitigation planning strategies.