Any unforeseen natural occurrence and, in the event of an emergency that weakens or destroys economic, social and physical capacity, such as loss of life and finances, destruction of infrastructure, economic resources and areas of employment, is defined as a natural disaster, highlights include earthquakes, floods, drought, seawater, volcanoes, landslides, hurricanes and natural pests (Vetrivel et al. 2018). Flooding as one of the major natural disasters that have occurred is very important in terms of economic losses and serious humanitarian concerns, in other words, the flood phenomenon is one of the most dynamic and disruptive natural events that put human life and property and social and economic conditions at greater risk than any other natural disaster (Rahmati et al. 2016; Yariyan et al. 2020). Flooding is an influx of water that overwhelms the surrounding and marginal areas and may cause damage to agricultural land, urban areas and tolls (Perera et al. 2015). Flooding is one of the most important hydro-ecological phenomena of nature that causes damage to human achievements at all times (Woodward et al. 2014; Darabi et al. 2019; Vafakhah et al. 2020). Among the populated areas and centers, there is the highest risk and potential for tangible physical damage caused by the flood. In recent years, the increase in urban flood hazards, particularly along the river banks, has resulted in the risk of flooding for residents and movable property in the area (Choubin et al. 2019). Due to the varying climate, unpredictable temperatures and rainfall in many of Iran's watersheds, several floods occur every year (Tehrany et al. 2014). Limiting environmental resources, reducing and destroying them as a result of the expansion of human activities, poses many challenges for today's society and the next generation. The Kan watersheds have not been impacted by these natural disasters and have been affected by flooding annually; studies show that the Kan watershed is vulnerable to floods. The emergence of restaurants, leisure, tourist and pilgrimage centers adjacent to the steep rivers Kan watershed have increased the potential for damage. Seven important flood events were recorded in this watershed, causing damage to industrial, residential, agricultural land use, and fatalities, according to the available information.
Reducing human casualties as well as damage to property and the environment is a key objective shared by countries most often impacted by natural disasters and they are increasingly conducting feasibility studies with economic analysis to mitigate the effects of these disasters (Molinos-Senante et al. 2011). Although floods cannot be prevented, the analysis of damage can be mitigated through appropriate analysis and forecasting techniques (Heidari 2014). The first step is to identify flood-prone areas with potential for flooding and prepare flood maps (Janizadeh et al. 2019; Hosseini et al. 2020). One way to prevent and reduce flood damage is to provide people with reliable information through flood hazard zoning maps (Cook and Merwade 2009). Mapping the hazard of floods is an essential step to predict the likelihood of a flood and to reduce and control potential floods. To this end, the modelling of flood hazards, which may involve multi-temporal data sets, is required. Recently, with the advancement in computer science, machine learning methods have been successfully applied in assessing flood hazards with higher accuracy. However, there is no agreement which method or set of methods can provide the best prediction.
The prediction and determination of flood susceptibility areas have been investigated by various researchers. Rapid access to satellite imagery based on remote sensing data has increased the use of geographic information systems in the preparation of flood susceptibility maps. A wide range of modelling techniques has been proposed and used in natural disaster assessment including AI based techniques (Sayers et al 2014). In recent years, Bayesian methods, partly because they are over-resistant to the presence of small sample sizes and can control missing or incomplete data, have been developed to assess and model flood sensitivity such as Naïve Bayes (Liu et al. 2016; Pham et al. 2020b; Tang et al. 2020). Also use of regression tree machine learning models such as Random Forest (RF) (Arabameri et al. 2020; Chen et al. 2020; Vafakhah et al. 2020), Decision Tree (Khosravi et al. 2018; Costache 2019; Janizadeh et al. 2019; Pham et al. 2020a), Logistic Regression (Shafapour Tehrany et al. 2017; Al-Juaidi et al. 2018; Tehrany and Kumar 2018) due to the appropriate capability of these models in modeling nonlinear phenomena such as flood, has been considered by researchers in this field.
Machine learning algorithms by default usually present point estimates only, and so decisions are made ignoring the uncertainty surrounding these estimates. In recent years, the use of ensemble models has attracted the attention of researchers in various fields, ensemble models because they benefit from several individual models and therefore have better performance than individual models (Al-Abadi 2018; Tehrany et al. 2019a; Costache and Bui 2020; Shahabi et al. 2020). Bayesian Additive Regression Tree is one of the new ensemble models that combined from Bayesian and Regression tree algorithms and due to this model benefit Bayesian framework can give access to the full posterior distribution of all unknown parameters in the model, which can be useful in decision-making and reduce uncertainty. Due to the fact that the flood is a non-linear phenomenon and has a lot of the uncertainty, use of appropriate models that have the ability to predict this phenomena and reduce uncertainty is essential in the management, planning and prevention of the hazard of these phenomena.
Therefore, the purpose of this study is to develop and present a new flood susceptibility model based on the ensemble type Bayesian Additive Regression Tree (BART) method. The new method will be compared with the Naïve Bayes (Bayesian type) and Random Forest (regression tree type) based models to evaluate the predictive performance of the new method.