This weekly digest focuses on budget allocation for natural disaster risk management.
Budget allocation for natural disaster risk management mainly deals with allocation of disaster mitigation funds (which are usually public funds) to different disaster-prone regions for different mitigation strategies. In general, the issue can be divided into the following levels:
- Level 1: Funding level for natural disaster mitigation programs. To answer this question, we need to understand how much are spent each year for public spending including all societal programs and capital investment, and how much percentage should be devoted to natural disaster mitigation programs. The output of this analysis becomes the input for the next levels of analysis. This issue is more of a type of political/policy analysis.
- Level 2: Funding formula for different geographical regions. To answer this question, we need the inventory of public infrastructure that are vulnerable to natural disasters. We also need to perform probabilistic hazard analysis to understand the overall risk level at different regions. In a more detailed manner, this level may also deal with prioritization of systems, structures and components (SSCs) for rehabilitation.
- Level 3: Budget allocation for different mitigation strategies. The mitigation strategies considered include preventive retrofitting, post-event rehabilitation and reconstruction, and disaster emergency management. To answer this question, we need to understand cost effectiveness of pre-event retrofitting, disaster emergency management plan, and post-event loss estimation.
- Level 4: Selection of optimal rehabilitation strategies. This is a typical engineering design decision making issue.
The March issue of ASCE Journal of Infrastructure Systems published a paper titled by ‘Optimization-based regional hurricane mitigation planning’ by Legg et al. This is continuation of the work on resource allocation for natural disaster risk management done by the research group led by Davidson and Nozick. The proposed linear programming model focuses on Level 2 and Level 3 questions and considers mainly the following three types of disaster mitigation alternatives:
- pre-vent retrofitting
- post-event retrofitting or reconstruction, and
- post-event acquisition and demolition
The objective of the LP model is to minimize the total expected expenditures subject to pre- and post-event budget constraints. The model divides buildings into many categories depending on their vulnerability to hurricane, a study region into many census tracts depending on their hurricane susceptibility, and a building resistance into several levels. In order to assess the building damage due to hurricane events, they adopt a detailed, component-based damage assessment approach.
Given the complexity of the proposed model, a heuristic solution procedure is employed. For the case study that they carried out for eastern North Carolina, it was reported that the model contains more than 25 billion variables.
Several critics are given below:
(1) Resource allocation for natural disaster risk management is by nature a stochastic programming issue. The authors used expected number of buildings at different damage states as the state variable of the model. This is nothing wrong. However, the formulation becomes problematic when it represents the budget constraints (both pre- and post-event budget) in terms of the expected costs (Eqn. (5) and (6)) at page 3).
(2) Although the separation of pre- and post-event retrofitting seems intuitive, it may not be necessary particularly when the frequency of hurricane is relatively high, say once every a couple of years.
(3) Time discretization makes the modeling very clumsy. The paper does not show the time trend of the results; however, given the Poisson assumption that has to be made for the disaster events, it is reasonable to perceive that a equilibrium status should be achieved should the annual budgets are constant.