A Spatially Explicit and Dynamic Approach to Flood Risk Management in South-east, Westmoreland, Jamaica Tracy-Ann Hyman, Mphil Student The University of the West Indies, JA Supervisors: David Smith & Leonard Nurse
Statistics on Flooding: The Caribbean Flooding associated with tropical cyclones, troughs, cold fronts and heavy / prolonged rainfall, is the most common and widespread hazard in the Caribbean (Carby, 2011; Carby et. al, 2012).
1979 Debris Floods Western Jamaica June 12, 1979 Tropical Depression stalled over the Western Section of the island PARISHES: Hanover, St. Elizabeth, Trelawny, Westmoreland, St. James Amt of rainfall: Western Jamaica 87cm (10hrs) vs. mean monthly rainfall (26.2cm) Intense Rainfall, high ground water levels, high soil moisture content COMBINED Persons Affected: 160,000 Deaths = 41 Economic losses= US $27 Million (as at 1979) OVERLAND FLOWS DEBRIS FLOWS HEAVY SEDIMENTATION Loss of life, crops, livestock, destruction of bridges, roads and infrastructure
Overall Aim of Research To construct a basic SIMULATION of a 1979 flood-like event on a 2016 Landscape - assessing the MOVING VULNERABILITY of Residents (BUSINESS AS USUAL SCENARIO) RESEARCH HYPOTHESIS The features and characteristics of a 1979 Flood event, will produce similar patterns of flooding and loss of life in 2016. (null)
Expected Outputs & Outcomes EXPECTED OUTPUTS (simulation) o Exact locations / zones likely for residents to be in (stuck/ injured/ killed/safe) o The number of people in these locations /zones o The socio-demographics of persons likely to be stuck/ injured/ killed i.e. age, gender, physical impediments, ability to swim, occupation EXPECTED OUTCOMES DRM / DRR o to assist with search and rescue operations o the placement of evacuation shelters or treatment centres (disease outbreak) SAFE ZONES o development of COMMUNITY flood early warning systems (FEWS) - determination of optimum response time (s) o The creation of a Decision Support Tool for LOCAL GOVERNMENT RESPONSE TO FLOODS Overall :reduction in Resident / Community vulnerability
Methodology Agent Based Models (ABMs) Creation of a VIRTUAL environment to test hypothesis and theories Computer simulations that model: the choices / decisions people make how they interact with other persons the impact of their decisions on themselves, other agents or the ENVIRONMENT in which they operate
Agent Based Model - Applications Source: Macal and North 2009 Source: AIDS Model, NetLogo Library
PILOT SITE: CAVE, WESTMORELAND POPULATION SIZE= 1095 (ED Districts) Houses are Actual GPS points Average Household Size: 3 pax The information to populate the model was collected through the use of social surveys - in community meetings; 1 to 1 73 persons (direct) (43 F) (30 M) 148 (indirect) (76 M) (72 F) AREA= 4.8 Km2
Flood Features Volumetric Flow Rate (Q value) Water depth Elevation of Land (DEM) Catchment areas SECONDARY AGENT PROFILE Age Gender Occupation Relationship to Primary Agent ABM Landscape & Infrastructure Physiographic regions Zones Roads & Bridges Social & Commercial Infrastructure e.g. hospitals, churches, schools COMPONENTS (Vulnerability Calculation) PRIMARY AGENT PROFILE Age Gender Experience with 1979 flood Occupation Income Swim Physical Impediments HOUSE Height Value Location Stilts AGENT SCHEDULE Daily Schedule Day Day of the Week Time Time of Day / Night Flood Time MODE OF TRANSPORTATION Taxis / Buses Motor Car Bike Walking VULNERABILITY Stuck Safe Injured Killed 9
Vulnerability Metrics Water Depth (feet) Exposure Safe Stuck Injured Killed References Direct Exposure of Agents 1 feet of water and below Above 1 feet to less than 3 feet 3 feet and above This Flood outcome is determined by The Ability to Swim, Physical Impediments, Age and Gender. 3 feet and above This Flood outcome is determined by The Ability to Swim, Physical Impediments, Age and Gender. as they cavorted in the new rivers and helped to push cars out of 3 feet deep water into the safety of 1 foot (32 perish in the flood, Jamaica Gleaner archives June 14, 1979) Agents in Vehicles Water below the wheel height < 3 feet of water (Taxi and Motor Car) < 4 feet of water (Bus 1) < 3.6 feet of water (Bus 2) Above 3 / 4 feet of water, occupants are assumed to exit the vehicle, and are then exposed to the direct exposure criteria below Above 3 / 4 feet of water, occupants are assumed to exit the vehicle, and are then exposed to the direct exposure criteria below..on one road as the car stalled in water reaching as high as the door (32 perish in the flood, Jamaica Gleaner archives June 14, 1979) Agents in Building (with stilts) Below stilt height From stilt height to below 3 feet of water 3 feet of water and above inside the house, then agents are exposed to the direct exposure criteria 3 feet of water and above inside the house, then agents are exposed to the direct exposure criteria Once building height is exceeded by water, every-one dies Mr. Keiling Gordon, a 53 year old farmer and his three year old granddaughter Natasha, who were trapped in the two storey six room structure when the flood water burst in (32 perish in the flood, Jamaica Gleaner archives June 14, 1979) Agents in Building (without stilts) 0 1 feet 1feet < X <3 feet 3 feet and above, agents go to the direct exposure criteria 3 feet and above, agents are exposed to the direct exposure criteria Once building height is exceeded by water, every-one dies before this he reported how the flood waters tore off the bottom section of his door and poured into his shop. The water rose to about a foot inside the shop. He opened other doors at the back and the water rushed out.. (Night of terror aftermath of agony, Gleaner archives June 24, 1979) 10
Preliminary Results Flood Impact Flood Period: 4pm to 11:59pm Safe Stuck Injured Killed Total Tuesday 191 17 4 9 221 TUESDAY FRIDAY Friday 197 11 6 7 221 Saturday 195 17 5 SATURDAY 4 221 11
Socio-Demographic Analysis (killed) Male Age (s) Female Age (s) Physio- Geo Region Tuesday 5 58,42,46,26,70 4 48,38,56,19 Coastal Friday 3 26,58,46 4 56,38,48,19 Coastal Saturday 1 70 3 48,56,19 Coastal TOTAL 9 11 Total = 20 pax INTERESTINGLY, MORE MALES DIED IN 1979 THAN FEMALES 12
1979 Flood Damage Cave, Area Source: Geological Survey Division 13
Zone Analysis Model Runs 2 5 6 3 7 THE MODEL HAS SHOWN THAT IT CAN CAPTURE THE FEATURES OF THE 1979 FLOOD So the aim is to model another past event; then a future event 12 4 8 ZONE 1 (20 pax) 14
WHY ABMs? Vulnerability studies for the most part, use indicators or indices for assessments that are static cant reflect a moving vulnerability (Hyman 2013) A systems based approach to vulnerability (Adger 2006) The inclusion of human movements in flood modelling Coupled Human and Natural Systems - SILOS Limited work on ABMs in the Caribbean - ESP. on DRR Examples to date: fisheries, telecommunications, agriculture
The Future: ABMs DRR/ DRM USE OF MOBILE DATA IN SCIENCE - BIG DATA TO ADVANCE THE SENDAI FRAMEWORK, SDG S Analysis of human mobility data, and patterns through agent based models Mobility patterns should best assist in the allocation of safe zones, EVACUATION SHELTERS, Search and rescue, relief distribution efforts
Acknowledgements RAJIV MANDERSON LAWRENCE BARRETT BARBARA CARBY
For Queries or Comments Email: hymannic@yahoo.com 18
What does this mean? These results can lead to improvements in the allocation of relief supplies i.e. water, food (efficiency) Real time monitoring of populations movements as it relates to infections disease outbreaks & early containment (Haiti Example) Better / Quicker Estimations of mortality (non- responding SIMS) & Reduction in Mortality: Rescue operations # of Buried But Alive SIM card users Mapping Population Movement can help in NON-DISASTER EVENTS such as terrorist attacks and conflicts Bengtsson et. al. 2011 Mobile Data can be used to rapidly produce detailed and up-to-date population distribution maps, sidestepping the need for the cumbersome once-a-decade census Deville et. al. 2014 19