Fair and systematic resource allocation is one of the most crucial issues in many fields, such as natural resources and environments, social welfare, engineering, and business: Among all, resource allocation in natural disaster management is particularly important as it has to be dealt with limited time and multiple types of resource donated by various aid agencies. According to the recent 30-year statistical data by the Emergency Disasters Data Base (EM-DAT), occurrences of natural disaster have kept increasing continuously. Until now, many researchers have paid attention to how to minimize losses caused by catastrophic unexpected events. Pre-disaster management systems, such as disaster warning systems and simulations of disaster and their consequences, are implemented to prevent damages or to mitigate losses by natural disasters. However, since natural disasters can take place anywhere and anytime, they may not be fully prevented. For this reason, a post-disaster management is also needed to facilitate relieving suffering of victims and bringing back their livelihoods to the sustainable condition as soon as possible.

Those post-disaster management systems which have recently been developed are mostly focused on the information sharing issue. Web-based Geographic Information Systems (GIS) technologies are employed to help agencies understand the big picture of disaster and its current situation as well as share their useful data. However, though these attempts may facilitate each party's understanding of the work done by other agencies in the affected areas and reducing the time for data collection, those agencies may still need to face with the problems of lacking cooperation among parties. In addition, even though some groups of aid agencies may be willing to cooperate with each other, they still need to deal with fair resource allocation issues.

According to the interviews conducted in the areas affected by tsunami disaster, it has been often claimed that a fair resource allocation is an important issue in post-disaster management. The problems arise not only from the scarcity in resources, but also from the need for managing resources from various sources to be used efficiently. Regarding this issue, major problems are: (1) mismatching between victims' needs and aids by agencies which could lead to inefficient allocation of resources; (2) duplication of aids provided by agencies, which may cause some places overwhelming resources while others with little support; (3) unfair resource distribution when comparing between different places; and (4) lack of cooperation and inefficient use of resource as some resources need to be grouped with others for their usability. Without considering these characteristics, the resources, when allocated singly, may not provide any benefits to victims.

In summary, to tackle these problems, it is firstly needed to induce an incentive of aid agencies to cooperate with each other by making them realize the benefit of cooperation. After that, it is important to guide them as to how to create fair resource allocation schemes based upon well-defined principles of resource distribution. In this research, the author has dealt with this issue by proposing a new decision support system, termed as ViTSPRA (Visualization Tool for Spatial Resource Allocation), which can be used as a medium for sharing up-to-date information among users and generating fair resource allocation schemes. This tool integrates the web-based mapping technology and analytic models for resource optimization.

Resource optimization models embedded in ViTSPRA are developed by incorporating socioeconomic aspects. To identify utility of an individual when receiving resources, the author has classified resources into two groups: independent resources and dependent resources. Independent resources are defined as the resource which can provide benefit for receivers on their own. Specifically speaking, the more resources provided, the more benefits passed on to victims. Examples of this type of resource are food, water, houses, and money. On the other hand, the benefits from dependent resources depend not only on how much amount has been provided, but also how they are combined or grouped with others. In other words, dependent resources can provide benefit for receivers if and only if they are formed in a group. Examples of dependent resources in the field of natural disaster are boats and engines, electric generators and light bulbs, etc.

In addition to the resource classification rules, the resource optimization model also employs two well-known resource allocation theories - utilitarian and egalitarian. The former aims at maximizing the social welfare, which is measure as the summation of individual utility in the society; while the latter aims at maximizing the welfare of the individual who is the worst in the society. Through ViTSPRA, these two approaches can be selected to generate their corresponding resource allocation schemes which show in the form of web-based mapping. This helps decision makers to visualize the problems in the same view, thereby could induce better cooperation between aid agencies. Through the geographic distribution of resources generated by the system, aid agencies can see the more benefits passed on to victims when they cooperate with each other. Besides, this system also provides the aid agencies with updated information of distribution of aids by various aid agencies. From this, each agency can know what the others are doing in a particular location; hence the problem of duplication of aid could then be reduced.

We have examined the feasibility of a prototype system in the case study on a tsunami, the natural disaster that took place on December, 26 2004. The main objective is to compare the results generated by the system to the existing approach. The area of the case study covers three districts in Phang Nga province in Thailand, which were seriously damaged by the tsunami and faced with the problems of fair allocation of resources. Data required for the analysis are needs data by victims in the affected area and aid data by various aid agencies. The need data used for the case study were gathered from the local government office in Phang Nga, while aid data were gathered from the Disaster Tracking Recovery and Assistance Center (D-TRAC).

After the analysis, we have compared the resource allocation schemes generated by ViTSPRA and the traditional approach. Using the social welfare as the indicator, results by ViTSPRA show the more benefit could have been passed on to victims. Also, the author has investigated the potential of this system to be implemented in the real context by asking for user responses. This prototype system was proposed to Disaster Tracking Recovery and Assistance Center (D-TRAC) who is working as the coordinator for creating a venue for various aid agencies. The director of this agency has supported the possibility of implementing ViTSPRA in practice. However, some concerns have also mentioned about the quality of data which are needed to be used as input of the system. She has argued that collaborative systems, such as a data collection system and standardized need assessment system, may be required in order that this system can be effectively implemented.

In conclusion, this prototype could be viewed as a backbone structure or a starting point of post-disaster resource optimization system for managing and utilizing shared data in the way that it could help agencies make a better decision. Some important issues of further exploration and development include: the development of data collection systems, bargaining systems for aid distribution schemes, and other algorithms for mixed resource allocation schemes.

　大規模自然災害後の救援活動においては，必要な救援物資を迅速かつ公平に配分することが必要となる．しかしながら近年多くの事例で援助物資の空間的偏在や現地需要との不一致が指摘されており，系統的な救援物資配分の支援手法が求められている．

　一方これまでの防災研究は自然災害を予測し，必要な対策を講じることを目的とした被災前活動に重きが置かれてきた．災害予測警報システムの開発や地震動の構造物への影響評価などがその例である．しかしながら自然災害が地球上のあらゆる場所で常に起こり得ることを考慮すれば，事前対策のみによって災害の被害を皆無にすることは原理的にも現実的にも不可能である．このことからも災害後の対策を対象とした防災研究の必要性は明白である．

　被災後の活動支援を目的とした防災研究の例としては，被災状況に関する情報を共有するための媒体の提供などが挙げられる．このような試みは研究目的にとどまらず，現在では被災状況をWeb上に公開された地図に表示するシステム等，専用の情報システムが実際に運用されることも珍しくない．しかしながら多数の援助機関が短時間に救援活動の連携と調整を行う必要が生じる状況においては情報共有を目的とした既往のシステムは未だ機能的に不十分である．

　このような背景を踏まえ，論文提出者は公正な資源配分の空間的決定を支援する情報システムViTSPRA (Visualization Tool for Spatial Resource Allocation)を提案・開発した．本研究では2004年のタイ国における津波被害を対象に，最初に被災者及び被災後の救援活動に携わった諸機関への聞き取り調査を行った．その結果救援物資配分に関して次のような問題点が明らかになった．すなわち，i)被災者のニーズと援助機関が供与する物資との間に乖離が生じたこと，ii)一部地域において異なる援助機関が同じ物資を配分する等の重複が見られたこと，iii)地域間において物資配分の公平性が保たれていないこと，iv)複数の援助機関の連携が不足しているために，本来他の物資と組み合わせなければ機能を発揮しないような物資も単独で不規則に配分されていること，の4点である．

　これらの問題を解決するために，論文提出者は開発したシステムに資源配分の効率性と公平性を評価するモデルを組み込み，各評価基準に応じて最適と判断された物資配分を地図上に視覚化することのできるユーザーインターフェイスを実現した．本システムで用いられた資源配分モデルの特徴は，上記iv)の問題を踏まえ救援物資を2種類に分類したことである．すなわち，単独で機能を発揮する物資(独立物資)と，他の物資と組み合わせなければ機能を発揮しない物資(従属物資)との2つである．独立物資は需要が存在する限り配分量に応じて受領者の効用が増大するが，従属物資は他の指定された物資との組み合わせ配分量に応じて受領者の効用が増大する．前者の例は食料や住宅などであり，後者の例は電源と電球の組み合わせ，ボートとエンジンの組み合わせなどである．配分状態の評価基準としては効率を重視したベンサム型の厚生関数と，公平性を重視したロールズ型の厚生関数の2つを用い，いずれかの基準をユーザーが選択できるものとした．

　本システムをタイ国津波被害の救援事例に適用した結果，現在の配分状況がいずれの評価基準についても改善しうる水準であることを示した．さらに代替的な配分を求め，視覚化することによってそれぞれの状況を系統的・定量的に比較した．

　以上，本研究は自然災害被災後の救援活動をより効果的かつ公正に行うための手法論を提供するものであり，時宜を得た社会的有用性の高い研究であると評価できる．よって本論文は博士(工学)の学位請求論文として合格と認められる．