Predicting property damage from tornadoes with zero-inflated neural networks

Abstract

Tornadoes are the most violent of all atmospheric storms. In a typical year, the United States experiences hundreds of tornadoes with associated damages on the order of one billion dollars. Community preparation and resilience would benefit from accurate predictions of these economic losses, particularly as populations in tornado-prone areas increase in density and extent. Here, we use a zero-inflated modeling approach and artificial neural networks to predict tornado-induced property damage using publicly available data. We developed a neural network that predicts whether a tornado will cause property damage (out-of-sample accuracy = 0.821 and area under the receiver operating characteristic curve, AUROC, = 0.872). Conditional on a tornado causing damage, another neural network predicts the amount of damage (out-of-sample mean squared error = 0.0918 and R 2 = 0.432). When used together, these two models function as a zero-inflated log-normal regression with hidden layers. From the best-performing models, we provide static and interactive gridded maps of monthly predicted probabilities of damage and property damages for the year 2019. Two primary weaknesses include (1) model fitting requires log-scale data which leads to large natural-scale residuals and (2) beginning tornado coordinates were utilized rather than tornado paths. Ultimately, this is the first known study to directly model tornado-induced property damages, and all data, code, and tools are publicly available. The predictive capacity of this model along with an interactive interface may provide an opportunity for science-informed tornado disaster planning.

Data-Driven Approaches for Tornado Damage Estimation with Unpiloted Aerial Systems

“…Data-driven approaches can help automate tornado damage estimation. For example, fully connected neural networks were used to predict tornado-induced property damage using publicly available data [9], which includes high-level information such as where and when a tornado occurred, tornado path length and width, and how much property damage it caused. Le et al [10] employed a fully connected neural network as a surrogate model for the performance assessment of a vertical structure subjected to non-stationary, tornadic wind loads.…”

Transformation of plant and soil covers of the Botanical nature monument “Pine forest near Venetsiya village” (Russia) as a result of a windfall

“…Currently, hurricanes, for example, have been appearing in new places where they've never met before (Hoffmann et al, 2018;Barrett et al, 2020). With great destructive power, wind storms cause huge loss of life and property damage (Moore, 2017;Diaz and Joseph, 2019;Cui and Caracoglia, 2019). They also have a strong impact on natural complexes, such as coastal wetlands, (Majidzadeh et al, 2020;Mo et al, 2020), tropical dry forests (Novais et al, 2020), deciduous forests (Santoro and D'Amato, 2019), agricultural land (Strader et al, 2018;Baker et al, 2020) and sandy sediments (Meixler, 2017).…”

Interpreting the socio-technical interactions within a wind damage–artificial neural network model for community resilience

“…Such factors include debris impacts and social characteristics. One of the aforementioned studies [ 5 ] specifically evaluates the applicability in creating an ANN to predict property damage, in terms of economic loss, from tornadic wind events. The ANN model was built using data related to tornado intensity, land cover and a significant number of potential sociological vulnerability demographics.…”

“…Machine learning approaches have proven viable within the field of natural hazards. Previous studies have applied such techniques to enhance modelling, forecasts and potentially predict outcomes or impacts from hazards [1][2][3][4][5][6][7][8]. Each of these studies used artificial neural networks (ANNs), a type of machine learning (ML), in which the ANN is best used for modelling a singular problem, for example, forecasting the overall impact, in terms of economic damage, from a hurricane event [1,9,10].…”