A data-driven approach to identifying and selecting temporary disaster debris management sites: The case of Istanbul

Abstract

Disasters manifest in various forms, encompassing both natural and human-induced events. They may arise suddenly, as in the case of earthquakes, fires, floods, tsunamis, hurricanes, and volcanic eruptions, or develop gradually, such as during civil conflicts or droughts. The impacts of these events can vary significantly in both their physical and social dimensions. The type and intensity of a disaster can result in the generation of substantial amount of debris, posing significant challenges for affected communities. To address such impacts, disaster management is typically divided into four key stages: mitigation, preparation, response, and recovery. During the recovery phase, debris management is essential, emphasizing the processes of collecting, reducing or recycling, and ultimately disposing of the remaining debris. Disasters can generate debris volume that considerably surpass the yearly debris production of the impacted community often straining the resources of existing solid waste disposal systems and staff. Addressing such immense volume of debris requires substantial financial investment and significant time for its clearance, removal, and disposal. Despite its critical role in disaster recovery, disaster debris management has received relatively limited attention in the literature compared to other aspects of disaster management. By putting forth a combined structure designed to optimize disaster debris management procedures, this thesis seeks to close this gap, paying particular attention to the particular difficulties Istanbul presents. This thesis's primary goal is to solve the difficulties associated with managing disaster debris, with a focus on enhancing the selection Temporary Disaster Debris Management Sites (TDDMS) and allocation debris to these sites. Through the integration of transportation optimization, clustering techniques, and spatial analysis, the thesis aims to: •Establish identification criteria for TDDMS that consider the environment, engineering, and economy. •Develop a clustering-based transportation model to optimize debris allocation to selected TDDMS. •Develop a systematic and replicable approach to site evaluation and debris allocation. •Validate the methodology by applying it to Istanbul's particular disaster debris management demands. Introduction chapter discusses the important difficulties faced by disasters, focusing on their increasing frequency and intensity because of climate change, population increase, and unsustainable development. Communities are disrupted by natural and human-caused disasters like floods and earthquakes, which cause significant death tolls, economic losses, and environmental degradation. Turkey's geographical and geological circumstances cause it highly vulnerable to disasters caused by nature, particularly earthquakes. Historical disasters such as the 1999 Marmara earthquake and the 2023 Kahramanmaraş-Hatay earthquakes are emphasized for their devastation, which included massive debris generation, fatalities, and community displacements. These instances demonstrate the critical necessity for effective disaster management strategies. The chapter defines disaster management as a cyclical process that focuses on the recovery stage, when debris management becomes an important component, necessitating the removal, sorting, and disposal of debris to promote community reconstruction. Major research questions concentrate around the criteria for selecting TDDMS and optimizing the number and operation to save costs while increasing efficiency. The thesis seeks to solve two critical challenges in disaster recovery: •Establishing the optimal number, locations, and service zones for TDDMS. •Estimating and allocation debris volume to aid in effective resource planning. Chapter 2 digs into disaster debris management, exploring its complexities and significance in the overall disaster management cycle. It starts by defining disaster debris and categorizing it according to its origin and features, stressing how different types of disasters result in distinct debris compositions. The volume of debris generated after disasters frequently exceeds typical waste production, posing problems to disposal systems and straining municipal waste disposal ability. The chapter emphasizes the significance of precisely calculating debris amount, as miscalculations can result in inefficiencies and financial losses. Various tools and methodologies, such as past record data analysis, mathematical models, and remote sensing techniques are highlighted as important means of forecasting debris amount. TDDMS have been emphasized as critical components for disaster debris control. These facilities perform a variety of functions, including debris storage, processing, and recycling, as well as serving as logistical buffers. In order to balance economic effectiveness and environmental preservation, the selection of TDDMS requires a thorough evaluation of engineering, economical, and environmental considerations. The disaster debris supply chain is investigated, with a focus on the optimization of TDDMS locations and debris allocation. The chapter emphasizes the significance of reducing transportation costs, time, and hauling distances to provide an efficient response and recovery process. A review of the most recent literature demonstrates considerable advances in debris management approaches, including GIS-based analysis and optimization algorithms. However, it also highlights key shortcomings, such as the absence of integrated frameworks that combine site selection, transportation optimization, and multi-criteria decision-making. This chapter sets the stage for the thesis by identifying these shortcomings and suggesting novel techniques to improving disaster debris management systems, with a particular emphasis on tackling the unique challenges given by large-scale disasters. Chapter 3 outlines the methodology employed to deal with disaster debris management challenges, with a particular emphasis on selecting TDDMS, estimating debris amount, and optimizing transportation. TDDMS selection begins with determining and prioritizing evaluation criteria based on a comprehensive literature study of 20 recently published national and international articles. These articles shed light on the environmental, engineering, and economic factors that influence the choice of TDDMS. The methodology strikes a balance between including different criteria, which typically range from 9 to 13 for reliability, practicality and meeting the dynamic and short-term demands of disaster recovery scenarios. A crucial aspect of the methodology is the accurate sizing of TDDMS, which directly impacts the efficiency and capacity of disaster debris management systems. This involves using historical disaster data and estimation models to predict debris amount. In this regard, FEMA guidelines, which take into consideration storage requirements as well as operational areas such as safety zones, access roads, fire pits, and hazardous waste management zones, identify a minimum of 100 acres per million cubic yards (about 40 hectares per 0.76 million cubic meters) of debris. As part of the approach, the selected criteria are also given weights according to their significance in relation to others. The Equal Weight Method, Analytic Hierarchy Process (AHP), and Analytic Network Process (ANP) are three popular methods to criteria weighing. To ensure comparability, criteria are standardized into a single unit utilizing Boolean or Fuzzy Logic, which simplifies the assessment process. Geographical Information Systems (GIS) are critical to the integration and analysis of spatial data. It includes comprehensive capabilities for correlating various data types, implementing zoning and buffering rules, and visualizing geographical relationships using criteria map overlays. This improves decision-making by providing a broader grasp of spatial dynamics. When integrated with Multi-Criteria Decision Analysis (MCDA), GIS allows for the integration of spatial and non-spatial data, resulting in a methodical and data-driven approach to site selection. This integration not only saves cost and time, but it also establishes a strong digital repository for continuing monitoring and adaption, guaranteeing that site selection processes are efficient and sustainable. Chapter 3 presents a tailored approach to estimating disaster debris amount in Istanbul, based on lessons learned during the 1999 Marmara Earthquake, which produced 13.18 million tonnes of debris. The model fits to Istanbul's metropolitan characteristics by employing similar parameters, such as 4.2 housing units per building covering 100 m² and 1.3 tonnes per square meter of debris generation. Formulas are offered to estimate debris at the district and neighborhood levels, allowing for more precise design of TDDMS and logistical activities. This system ensures scalable and effective debris control that is specific to Istanbul's demands. Chapter 3, additionaly, explores cluster-based transportation optimization for disaster debris management, with a focus on efficient allocation of debris to TDDMS. To address unequal debris allocation and restricted site capacity, the system incorporates clustering algorithms and optimization models to reduce costs and environmental implications. Traditional optimization methods, such as Mixed Integer Linear Programming (MILP) and Multi-Objective Optimization (e.g., NSGA-II), are discussed for their ability to reduce costs and save time. Clustering approaches, such as K-medoids, provide more adaptability and precision for large-scale geographic and quantitative data sets. The K-medoids algorithm divides neighborhoods into clusters, then selects core spots (medoids) to optimize debris transportation while being robust to noise and outliers. The Haversine formula is used to calculate precise distances, taking into account the Earth's spherical shape and ensuring spatial relationships are accurate. The transportation model's goal is to reduce the total costs associated with transporting debris from neighborhoods to TDDMS. Constraints include meeting the neighborhood debris levels, adhering to TDDMS capacity constraints, and ensuring sites are only used when activated. The model employs practical assumptions, such as fixed transportation costs, consistent site capacities, and adequate resources for debris transports. By integrating clustering, spatial analysis, and optimization, the system assures effective and scalable disaster debris allocation, addressing logistical challenges while minimizing costs and environmental implications. This method offers a strong framework for managing debris allocation in disaster recovery scenarios. Chapter 4 describes the practical application of the proposed methodology for opting for TDDMS in Istanbul, a city with high seismic risk due to its proximity to the North Anatolian Fault Zone (NAFZ). Applying GIS and land suitability analysis, the study assesses possible sites against a set of predetermined criteria customized to Istanbul's geographic, environmental, and urban characteristics. Istanbul, home to approximately 20 million people, has significant seismic dangers due to its proximity to the NAFZ. Historical statistics, including as the 1999 Marmara Earthquake, illustrate the city's vulnerability, with a 7.5 magnitude earthquake expected to yield almost 25 million tonnes of debris. This demands effective debris management systems, which include the identification and selection of TDDMS. The study focuses on Istanbul's 38 districts, totaling 5,461 km². The land suitability study includes nine important criteria derived from a thorough literature assessment. These include proximity to lakes and rivers, built areas, major roads, airports, land slope, aspect and wind direction, as well as ownership and land use. Data were gathered from reliable sources, including HydroSheds, ESRI, OpenStreetMap, and the General Directorate of Land Registry and Cadastre. The analysis used raster and vector data for spatial evaluations, with buffers and thresholds set for each criterion. The land suitability analysis evaluated TDDMS locations utilizing ArcGIS 10.8.2. Euclidean distances measured proximity to important characteristics, whereas Boolean logic reclassified land to be suitable (1) or unsuitable (0). Two overlay approaches were employed: Fuzzy Overlay ("AND") for tight binary outcomes and Weighted Sum for a flexible the suitability index that allows constraint relaxation as needed. This approach ensured a thorough assessment of potential sites. The Weighted Sum analysis was selected to identify TDDMS. Satellite imagery and property records confirmed the findings, excluding undesirable zones such as woods and overcrowded metropolitan regions. Reclassification maps indicated that proximity to built areas and land use were the most restrictive criteria, but distance from lakes and rivers posed fewer limits. Fuzzy Overlay produced limited results due to stringent constraints, whereas Weighted Sum allowed for more flexible categorization, identifying 46 eligible sites of varied sizes and ownership. Some industrial properties were allowed despite looser requirements to have adequate capability for managing the expected 25 million tonnes of debris. After identifying potential TDDMS, the focus shifted to Istanbul's Asian side, enhancing the analysis by narrowing it to specific neighborhoods and utilizing neighborhood-level debris amounts estimates for greater precision. These estimations, computed using the debris estimation formulas outlined in the methodology section, are based on the Istanbul Earthquake Loss Estimation Update Project (2019) and consider a Mw 7.5 earthquake scenario. The projections include buildings classified as Very Heavy Damage, Heavy Damage, Moderate Damage, and Light Damage, along with the number of buildings categorized by story count (1-4, 5-8, 9-19 stories). The final application in Chapter 4 compares clustering-based and non-clustering methods to optimize disaster debris transportation in Istanbul's Asian side. These methods were tested for their success in allocating debris amounts from neighborhoods to TDDMS, emphasizing the importance of balancing cost efficiency, operational complexity, and site usage. The clustering-based method applied the K-Medoids algorithm to group neighborhoods based on their geographical proximity. Among several cluster configurations, a 15-cluster scenario demonstrated a balanced trade-off between transportation cost and operational complexity. The method focused on allocations to seven active TDDMS, lowering transportation costs and simplifying logistics by concentrating on fewer, larger sites. Conversely, the non-clustering method assigned debris amounts from neighborhoods directly to TDDMS, leading in a modest increase in transportation costs. This method used a broader variety of TDDMS, including smaller sites, which provided greater flexibility but increased logistical constraints. The comparative analysis revealed important trade-offs: clustering decreased costs and operational overhead by centralizing allocations, whereas non-clustering allowed greater flexibility and scattered allocations at the cost of increased complexity. These insights provide significant recommendations for disaster debris management, emphasizing the importance of tailoring methods to particular operational priorities and constraints.