BNHCRC Project: Pre-disaster multihazard damage and economic loss estimation model

Funded through the Bushfire and Natural Hazards Cooperative Research Centre (BNHCRC), this project will identify the optimum economic policy options to recover or minimise the adverse effects of natural hazards.The economic impacts resulting from natural disasters are often overlooked in the economic planning process. This is because the immediate focus in the wake of natural hazards is typically placed on the emergency response, and it takes time to realise the economic effects of the disasters. In Australia, the disaster management arrangements across all stages (mitigation, preparedness, response and recovery) have proven to be very successful at saving lives, however, less attention and resources have been devoted to the economic impacts of natural disasters. One of the problems identified in this setting is the lack of estimates of the full economic impact of natural hazards covering all the affected sectors of the economy. An ideal estimation should reflect both the primary and secondary effects of the natural disasters so that persistent losses throughout the economy originating from various sectors are taken into account. A case study was conducted on the 2010-2011 Queensland floods, which involved comparisons between the flooded areas and unaffected areas of 19 economic sectors. This involved comparing the economic conditions of individuals residing in flooded and non-flooded areas before and after the floods, and revealed the sectoral decomposition of income and employment differences reflected in household well-being because of the floods. more details.

My contributions contain three parts:

(1) analysis of extreme rainfall, drought, temperature from historical weather data (obtained from BOM, daily record at 0.05 arc degree from 1900 to 2014, total size 500GB) with various thresholds (0.95,0.9,0.85,0.8,0.75,0.7,0.6,0.5) at grid level across Australia. Five NeCTAR virtual machines (80 cores in total) were used to perform parallel computation. The outputs were fed into a multivariate regression model for economy loss estimation analysis.


techs: R, PostgreSql, ArcGIS, QGIS
project timespan: Jun 2014 – Dec 2016



Intelligent Disaster Decision Support System (IDDSS)

This project aims to develop a prototype system for urban disasters integrating a smart Geospatial Information Platform with an advanced optimisation and simulation engine. The IDDSS will support an end-to-end process from scenario planning to disaster response and recovery. The platform will perform real-time collection, management, analysis, distribution, and visualization of information for enhanced situation awareness, aligning the impact of information with its availability. This real-time stream of critical information will populate the optimisation/simulation engine whose goal is to increase the cognitive abilities of decision makers when faced with an urban disaster of large magnitude and uncertainty.

The project technical report is available here:   IDDSS_Project_Technical_Report

I designed and implemented the entire system including backend data/model services and frontend user interface.

techs: GeoServer, PostgreSql, GeoTools, Cesium, ExtJS, GRASS, R, JAVA
project timespan: Nov 2013 – Now



Understanding Walkability and Public Transport Using Ped-Catch

This project aims to test this hypothesis by utilising new methods for measuring walkability whilst bringing together separate data bases from Department of Sustainability and Environment such as road network, residential density, rail networks, and aerial imagery with information from Transnet including public transport frequency and stop locations, with traffic volume information provided by VicRoads. This project will build directly upon another AURIN-funded project that created a walkability index which will be calculated and applied to census collection districts surrounding public transport nodes in the North West Metropolitan Region. This indexing will be overlayed with an animated agent-based pedestrian modelling tool called ‘Ped-Catch’ which uses artificially intelligent ‘agents’ who navigate street networks based on predefined rules to measure catchments around a selected nodes such as schools or railway stations. Comparisons can be made between an existing street network and potential improvements through urban interventions such as pedestrian links, pedestrian bridges, alternative street layouts, or potentially freight volume as a barrier to walking. more details.

techs: GeoServer, PostgreSql, GeoTools, Openlayers, ExtJS
release: Jun 2013
developed by: Gus Macaulay, me



Examining How Social Infrastructure Data Can be Used in Local Health Planning and Innovation for Type 2 Diabetes Management

The aim of this project is to improve access to an integrated set of health-related (prevalence and service use) and social and physical infrastructure data to aid policy makers and planners. This Demonstrator Project is based on collaborative work previously undertaken by The Department of General Practice at the University of Melbourne and the Department of Health North West Metropolitan Region (NWMR). more details.

techs: GeoServer, PostgreSql, GeoTools, Openlayers, ExtJS
release: May 2013
developed by: Gus Macaulay, me, Azadeh Keshtiarast



Housing Affordability and Land Administration

The focus of this project is to demonstrate the link between availability of developable land and space and affordable housing development. This requires a rigorous analysis of residential land development potential that is essentially linked to datasets such as land value and capital improved value information (Agunbiade et al 2011). The core analysis is intended to determine Residential Development Potential Index (RDPI) for a cross section of the NW Corridor region. Spatially enabled tools provide this (Kalantari 2007). The tool aims to provide ways of analysing and communicating, the challenges and prospects of discovering developable land for housing production. more details.

techs: GeoServer, PostgreSql, GeoTools, Openlayers, ExtJS
release: May 2013
developed by: Ghazal Karami, Azadeh Keshtiarast, me, Alireza Shamakhy



Food Desert Index For Greater Melbourne Area

This research investigates possible barriers to accessing fresh and healthy food in metropolitan Melbourne. The idea is derived from the concept of ‘food deserts’ – spaces within cities that lack adequate access to affordable, healthy food. The methodology comprises two steps: (1) devising an initial travel accessibility index (TAI) for each mode – walking, driving and public transportation (tram); (2) generating a more complex and realistic indication of each area’s overall travel accessibility by integrating travel accessibility index in the step (1) with car ownership rates. To calculate TAI, a travel information database has to be built up first, in which, every record consists of the details (i.e., time, distance, travel mode) of travelling from a residential Mesh Block(MB, a spatial unit applied in the 2011 Australian Census by ABS) to its nearby supermarkets. To obtain realistic travel information, public navigation APIs were applied in this project. Within nine hours, over 1.1 million API calls were sent to MapQuest and Google servers through four Nectar VM nodes controlled by R parallel backend ‘doMC’ module.

techs: R, MapQuest/Google Navigation APIs, Nectar Research Cloud
release: May 2013
developed by: me



Impacts of Planned Activity Centres on Local Employment and Accessibility: Evidence of Progress Toward a Polycentric City

This project responds to a strong consensus among policy makers, that Melbourne needs to adopt a multi-nodal metropolitan planning strategy in order to foster local economic development, reduce commute burdens on households, and make jobs more accessible to where people live – particularly lower-income and disadvantaged workers. For decades, metropolitan planning strategies have sought to promote non-CBD activity centres in metropolitan Melbourne. Since the 1980s and before, Victoria’s metropolitan urban planners and state governments have been trying to develop activity centres – local job, shopping, and recreational centres that serve the local population. The idea is that activity centres reduce the need for commuters to travel to the city centre, and supply firms with incentives to locale in a jobs cluster, rather than choosing dispersed locations. In theory, there is an overall benefit to commuters and taxpayers, through reduced commutes and more local jobs. The core focus of this demonstration is to identify spatial patterns in employment locations, commuting behaviour, worker job accessibility, and their relationships to each other, in the Northwest Corridor of metropolitan Melbourne. Specifically, we wish to understand whether spatial policies aimed at cluster development have resulted in employment clusters, reduced commuting burdens, or a better set of accessible job choices for workers in the Northwest. more details.

techs: R, Openlayers, ExtJS
release: Apr 2013
developed by: me, Philip Greenwood



eMU-Public & eMU-OHS

eMU-Public is designed for the public to identify emergencies and send out notifications around the University of Melbourne, Parkville campus. Users will receive a message once an OHS officer makes a response to it. Thousands of geo-coded QR tags will be placed around the campus so users can quickly scan it to get accurate location information, not matter outdoors or indoors. FirstAid and AED toolkits are also marked on the map, and the calculation of fastest walking route between user’s current position and any toolkit is supported. download from App Store.

eMU-OHS provides a swift way for OHS officers and department heads (the observers) in the University of Melbourne to respond and monitor newly detected emergencies around the campus (Parkville campus only at current stage). For OHS users, they will get an alarm once an emergency notification is sent out by a public user ; they can respond to the sender directly or assign this task to their colleagues. For observers, they will also received an alarm when an emergency appears and will be confirmed once any OHS officer does make a response to the case. download from App Store.

techs: MapQuest, Objective-C, MySQL
release: Mar 2013
developed by: me



Space Node Betweeness

Betweenness is a measure of a node’s centrality in a network. It is equal to the number of shortest paths from all vertices to all others that pass through that node. It is a more useful measure of both the load and importance of a node. This project aims to construct a multi-floor space network for the old ABP building in the University of Melbourne, then analyse and visualise the betweenness of space nodes in a 3D interface. For each floor, an axial map was created in depthmapX and exported to Pajek (.net) network format, which then was fed to the ‘igraph’ module in R. An additional ‘bridgeIdx’ data file was edited to link the separate floor space node network into a wholesome network for the old ABP building. ‘brokenvertices’ and ‘brokenedges’ files were created to control the space node accessibility to support different scenarios simulation. Check here to see a video clip of the final results.

techs: R, QGIS, depthmapX
release: Mar 2013
developed by: me



Using Public Navigation APIs for Urban Concentration Measurement

The global urbanisation information congregated by the UN is reported by various countries, however divergencies exist due to absence of universally accepted definition of “urban”. It will become particularly difficult to conduct a cross-national analysis or calculate the aggregate urbanisation status of various regions. Uchida and Nelson (2008) propose a measure of urban concentration called agglomeration index(AI), which is based on three factors: population density, the population of a “large” urban center, and travel time to that large urban center. The main objective is to provide a globally consistent definition of settlement concentration to conduct cross-country comparative and aggregated analyses. This work largely respects the factors used by Uchida and Nelson for AI computation, and employs public navigation APIs (MapQuest and Google) to calculate the third factor, which is a more realistic and accurate than Uchida and Nelson’s cost-distance model. The r script was developed to measure the AI of different countries in a particular year at any given administrative district levels and had been used to generate urban agglomeration index of eight south-east asian countries for the World Bank Report 2012.

techs: R, MapQuest/Google Navigation APIs
release: Dec 2012
developed by: me

Leave a Reply