Flood risks on assets and livelihoods in tropical catchments within oil palm landscapes

Simbiwen, Terence (2016) Flood risks on assets and livelihoods in tropical catchments within oil palm landscapes. Masters (Research) thesis, James Cook University.

PDF (Thesis)
Download (14MB) | Preview
View at Publisher Website: http://dx.doi.org/10.4225/28/5a9f309681a...


There are currently more people than before seeking a livelihood through oil palm agriculture around the world – approximately 14 million people with properties vulnerable to flood risks, and more than half of the settlements contain livelihood assets being exposed to seasonal floods. Flood risks cannot be eliminated but they can be effectively analysed and reduced in many oil-palm dominated landscapes. To assist disaster planning, hazard and risk mitigation, tools and models must be used to understand flood characteristics and behaviour, the areas affected, potential flood risks, levels of exposure and loss probabilities of livelihood assets. However, little is known about riverine flood risk to, and vulnerability of, livelihood assets in oil-palm dominated landscapes. This study assessed flood risks on livelihood assets in the Dagi River catchment, PNG.

Fieldwork data and hydrological models were used to determine flood characteristics and behaviour. Water depths, velocity and duration were observed during two wet seasons in 2010 and 2014. Flood discharge in terms of its peak and magnitude were analysed using rainfall data simulated using HEC-HMS. Both Log-Pearson Type III Distribution and Gumbel's extreme value distribution techniques were used to analyse flood frequency. In HEC-HMS program, the millimetre unit of depth is written as MM (USACE, 2013; p.124). Rainfall-runoff simulation results show that upstream reaches have a steep hydrograph with a peak discharge of 1326m³/s (130.10MM) for 2010 at 13:00pm compared with that of 2014 which peaked at 12:30pm with 729m³/s (86.96MM). Downstream reach generally showed a broader hydrograph with a peak discharge of 1158.4m³/s (72.47MM) at 14:00pm in 2014, while in 2010 it was 2424.4m³/s (109.67MM) that peaked by 14:30pm. Floods in Dagi rise quickly upstream and flow fast downstream and this provides less time for warning and evacuation.

Velocity in all cross-sections decreases towards the banks and in the floodplains as they encounter roughness. The 2014 floods had an average velocity of 4.35m/s in the main channels upstream while downstream average velocity was 2.75m/s. The 2010 floods had an average velocity of 5.38m/s upstream while downstream average velocity was 3.76m/s. The flood duration from upstream in 2014 receded to normal levels in three days (72 hours) but took seven days downstream. In 2010 the flood receded within four days upstream but took 11 days to reach normalcy downstream. Flood height varies for all sites in response to slopes (1-20⁰) from 0.01m to as much as 7.5m. The stream power during the 2014 flood increased from 1915.12 watts/m² (N/m s - where N is expressed as stream power per unit weight of 1 N and work and energy are metres per second written as one unit - m s) from sub-catchment 1 to as much as 9575.58 N/m s at sub-catchment 4. The flood event of 2010 had a recurrence interval of 11 years (9.09%) while that of 2014 had a recurrence interval of 7.33 years (13.64%). This means floods in Dagi catchment are highly variable and this depends on many factors but primarily rainfall and slope characteristics.

SOBEK 1D2D, HEC-RAS Beta 5.0, HEC-RAS 4.1 and HEC-GeoRAS modelling software were used to model and visualise flood inundation and hazards. Modelling focused on flood velocity and depth in 1D and 2D channels, where inundation and hazards were visualised. The spatial extent of inundation and hazards were determined by using flood characteristics (velocity and depth) obtained after the 2010 and 2014 flood simulation. Because of the relatively flat topography (1-20⁰) with no flood protection in all reaches, high peak discharge breached the bankfull stage and inundated within an average radius of 500m. The 2010 flood event inundated a large area (79.9 ha) compared with the 2014 inundation (55.2 ha), with most inundation occurring in the middle and lower reaches. Results validated using depth and velocity data together with various roughness figures revealed a difference, on average, of 0.40m in the modelled and observed data. The peak difference during the 2014 flood event ranged between 0.24m and 0.55m. On average, the 2010 flood revealed a difference of 0.45m in the modelled and observed results and at peak discharges the difference ranged from 0.32m to 0.54m. Thus, the modelled peaks agreed with those observed. The simulated results with appropriate input data, initial condition, boundary conditions, model assumptions, roughness coefficient values and coarse representation of the grid resolution were able to generate a very good simulated flood inundation extent.

Flood risks were assessed using a risk-based approach by analysing risk comprehensively. It estimated flood hazards using geo-processing tools and hydrodynamic models that represented flood intensity. It estimated vulnerability based on the percentage of livelihood assets damaged, dependent on flood velocity and depth. This study used a combination of qualitative and quantitative risk assessment methods. Focusing on depths of inundation and maps of vulnerable land use, an assessment of risks was performed qualitatively. The level of vulnerability and risk zones were identified based on the assigned land use weights, hazard and vulnerability assessment criteria. Using the weights and criteria and land use curves, raster-based vulnerability and risk maps were drawn in relation to three exceedance probabilities. Quantitative risk assessment involved estimating the total costs of exposed elements (direct tangible) based on the damage functions and classified according to their type. Results showed that houses and buildings incurred the greatest costs (34.3%), leaching of fertilisers (27.2%), deaths (23.7%), subsistence gardens (11.4%), roads (2.6%), damaged oil palm trees (0.8%), non-pick-up of bunches (0.02%) and formal job income loss (0.01%). The total economic costs based on the elements at risk is PGK77,869,451. This is equivalent to US$26,545,696 (23rd September 2015 exchange). The results reveal that the level of damage varies with flood probabilities.

Flood risks have been exacerbated by the increasing population and the need for more people to seek a livelihood. This has increased the extent and impacts of floods through the replacement of vegetation with bare and levelled land surfaces associated with increased oil palm cultivation and subsistence gardening in riparian zones. These land uses have interfered with the water cycle and the stream channel morphology. Prevention plans and mitigations recommended for flood disaster should include the implementation of conventional and unconventional structures.

Item ID: 51304
Item Type: Thesis (Masters (Research))
Keywords: Dagi River catchment, flood risk, flooding, floodplains, floods, hydrological models, livelihoods, palm oil, Papua New Guinea, seasonal floods, Tropics
Date Deposited: 23 Oct 2017 05:02
FoR Codes: 04 EARTH SCIENCES > 0406 Physical Geography and Environmental Geoscience > 040608 Surfacewater Hydrology @ 100%
SEO Codes: 96 ENVIRONMENT > 9610 Natural Hazards > 961005 Natural Hazards in Fresh, Ground and Surface Water Environments @ 100%
Downloads: Total: 92
Last 12 Months: 12
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page