Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation, Hydrology and River Channel Variation

i NEMAIA WAQANIVALU KOTO FLASH FLOODS IN THE NADI WATERSHED, FIJI: MORPHOMETRY, PRECIPITATION, HYDROLOGY AND RIVER CHANNEL VARIATION Orientador: Prof. António Alberto Gomes Masters in Geographic Informations System and Spatial Planning Semester II, 2014 Classificação: Ciclo de estudos: Dissertaçao/relatório/Projeto/IPP: Versão definitiva Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation ii Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation iii Acknowledgements I would like to acknowledge the following people whose contribution to this research work was inordinate, to whom I am forever grateful. Firstly, to my supervisor Professor Alberto Gomes of the Geography Department, University of Porto, I am grateful for the inspiration and wisdom, as well as the amount of time you put in to helping in the completion of this thesis research, of which I will cherish. Also, to Inês Marafuz, Phd research student at the Geography Department, University of Porto, I am grateful for your willingness to share the knowledge that you have, as well as your time in navigating this research until its completion. For the development and production of this thesis research, I feel a deep sense of appreciation to the following people:  Dr John Lowry of the Department of Geography, University of the South Pacific for his continuance guidance in the direction of my research  The entire teaching staff of the Department of Geography, Faculty of Letras, Univversity of Porto. My sincerest gratitude also to the entire department of the International Office at Faculty of Letras.  The Erasmus Mundus Scholarship Programme, for this opportunity of academic advancement with their sponsorship for the two years  My colleagues and friends that accompanied me in this academic journey whose tireless provision I am forever grateful for: Ilaisa Naca, Maluseu Tapaeko, William Young and Baraniko Namanoku. And lastly to my family and friends in Fiji, who played an integral role as my sole support system pushing the boundaries which at times I thought was never possible. Thank You, Vinaka and Muito Obrigado. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation iv Abstract Intense precipitation and flash floods has been a problem in Nadi, a town located in the north western part of Viti Levu island, Fiji. The latest flood events occurred in January 2009 and 2012 respectivefully of which the former was reported to be the worst since the 1931 floods (Chandra & Dalton, 2010). The main aim of this research is to understand the characteristics of flash floods in the Nadi watershed by analysing three main research objectives; (i) basin morphometry of the Nadi watershed using geospatial technology, (ii) precipitation data for a period of 52 years (1961-2013) by calculating time of concentration, peak discharge for a return period of 10, 50 and 100 years using Giandotti formula for the main basin as well as four sub basins. Also to construct a trendline for maximum monthly rainfall for 50 year period and predicted maximum monthly rainfall for 60100 years and (iii) dynamics of river channel changes in the Nadi basin that is caused by a flood event by using satellite images for the years 2005 (pre flood), 2012 (during flood) and 2014 (post flood). The methodology chosen was done in three parts. Firstly, in order to realise how flash floods occur, there is a need to better understand the basin morphometry of the watershed. These include a linear and areal analysis to obtain drainage density, hydrographic density, compact coefficient, elongation and roundness amongst other parameters. Secondly, using the precipitation data, calculation of time of concentration, peak discharges on a 10, 50, 100 year return period was carried out using Giandotti formula. Finally, using satellite images and ArcGIS to digitize river channels of the main Nadi river in two locations and to compare any changes in its physical aspect as well as the flooded alluvial plain. The results show that the basin morphometry of the Nadi watershed is such that the total number and length of stream segments is high in first order streams and decreases as the stream order increases.The time of concentration (tc) of 11.8hours for the main basin is relatively rapid considering the elongated shape of the watershed. Furthermore, the peak discharge of a 10, 50, and 100 year return period also indicates the main basin having a peak discharge of 3762.47 m3/s for a return period of 10 years and doubles in a 100 year return period with 6165.27 m3/s, predicting the increase in severity of intense precipitation. This pattern is the same for the four sub basins. Channel variation results also indicate a small yet significant change in its channel formation for a period between 2005-2014, increasing the extent of flooded alluvaial plain, and consequently increasing the boundary of the river channels. These results are intended to assist future researchers in the field of geohydrological studies. Keywords: Flash Flood, Basin Morphometry, Hydrology, Channel Variation, Nadi Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation v Acronyms CDBCentral Business District CSDCommission on Sustainable Development DEMDigital Elevation Model DTMDigital Terrain Model EUEuropean Union FIBSFiji Islands Bureau of Statistics FJDFiji Dollars FMSFiji Meterological Services GISGeographical Information System JICAJapan International Cooperation Agency NOAANational Oceanic and Atmospheric Administration PICPacific Island Countries SOPACSouth Pacific Islands Applied Geoscience Commission SRTMShuttle Radar Topography Mission UNUnited Nations UNCEDUnited Nations Conference on Environment and Development UNDPUnited Nations Development Program Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation vi General Index Acknowledgements .......................................................................................................... iii Abstract ............................................................................................................................. iv Acronyms .......................................................................................................................... v General Index ................................................................................................................... vi Chapter 1: Introduction .................................................................................................... 1 1. Introduction ........................................................................................................... 2 1.2 Objectives .......................................................................................................... 4 1.3 Conceptual Framework ...................................................................................... 5 1.4 Study Area: Nadi ............................................................................................... 7 1.5 Rationale for this research ................................................................................. 9 1.6 Literature Review ............................................................................................ 11 Chapter 2: Methodology & Materials ............................................................................. 17 2. Methodology .......................................................................................................... 18 2.1 Primary Data .................................................................................................... 19 2.1.1 Elevation Data: Shuttle Radar Topography Mission based DEM ................. 19 2.1.2 Precipitation Data .......................................................................................... 23 2.1.3 Satellite Imagery: Google Earth .................................................................... 26 2.2 Secondary Data ................................................................................................ 26 Chapter 3: Physical Characteristics of Nadi Basin ......................................................... 27 3. Characteristics of Hydrological Basin for Nadi River............................................ 28 3.1 Morphometric Analysis of Nadi Basin ............................................................ 29 3.1.1 Linear Analysis: ............................................................................................. 29 3.1.2 Areal Analysis ............................................................................................... 33 3.2 Physical Characteristics: Slope, Aspect and Geology ..................................... 36 Chapter 4: Hydrological Characteristics and River Channel Variation ......................... 39 4. Hydrology ............................................................................................................... 40 4.2 Analysing Precipitation Data ................................................................................ 41 4.3 Time of Concetration (tc) ...................................................................................... 42 4.2. Channel Variation ................................................................................................ 44 Chapter 5: Conclusion .................................................................................................... 54 5. Conclusion .............................................................................................................. 55 5.1 Recommendation .................................................................................................. 56 Bibliography ................................................................................................................... 58 Appendix ........................................................................................................................ 62 Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation vii List of Figures and Tables Figure 1 The Fiji islands located east of Australia. .......................................................... 4 Figure 2 The proposed scheme for this research .............................................................. 5 Figure 3 Location of Nadi town ....................................................................................... 7 Figure 4 Cyclone season from November to April........................................................... 8 Figure 5 The Nadi basin covers 490 km2 ......................................................................... 8 Figure 6 The January 2009 flood .................................................................................... 10 Figure 7 Linear and areal morphometry formulas .......................................................... 13 Figure 8 The shape of stream network affects discharge time. ...................................... 14 Figure 9 Different models of hydrographs as adopted from Sarangi (2007) ................. 15 Figure 10 Processed SRTM elevation data..................................................................... 20 Figure 11 Step 1: selecting and exporting the island of Viti Levu ................................ 21 Figure 12 The main Nadi basin as well as the four sub basin (A,B,C,D) ...................... 23 Figure 13 Time of concentration .................................................................................... 25 Figure 14 Giandotti formula for calculating peak flow .................................................. 25 Figure 15 The formula for calculatting Return Period (T) ............................................. 25 Figure 16: Hydrological watershed and four sub catchments. ....................................... 28 Figure 17 Drainage Density for the watershed of Nadi .................................................. 35 Figure 18 The slope map of the Nadi watershed with the stream network .................... 37 Figure 19 The aspect map of the Nadi watershed .......................................................... 38 Figure 20 Flood events as recorded by Yeo (McGree et al, 2010) ................................. 40 Figure 21 Precipitation data tabulated and graphed for a period of 52 years ................. 41 Figure 22 Trendline for the return period ....................................................................... 42 Figure 23 The two selected areas of study ..................................................................... 45 Figure 24 From the Location 1(a) 2005 satellite image showing the Nadi river ............ 46 Figure 25 (a) 2005 channel sketched in ArcGIS showing pre flood event..................... 47 Figure 26 From Location 2 (a) 2005 satellite image showing the Nadi river ................ 49 Figure 27 The segment of the river channel in Location 2 ............................................. 50 Figure 28 Summary of channel variation analysis ......................................................... 52 Table 1 The five main types of urban floods .................................................................... 2 Table 2 A brief summary of the literature review, the authors and their research. ........ 12 Table 3 Fluvial hierachy of river channels in the Nadi watershed ................................. 29 Table 4 Bifurcation ratio adopted from Horton (1945) .................................................. 30 Table 5 Segment length (LU) and relationship .............................................................. 30 Table 6 Order of segments and relation to length .......................................................... 31 Table 7 Important morphometric parameter for basin geometry .................................. 32 Table 8 The elongation ratio ........................................................................................... 33 Table 9 Time of concentration for Nadi watershed ........................................................ 43 Table 10 Peak discharge for the 4 sub basins in a return period of 10, 50, 100 years ... 43 Table 11 Calculating the average, minimum, maximum length of transects ................. 48 Table 12 Calculating the average, minimum, maximum length of transects ................. 51 Table 13 In location 1 segments of 6 labelled A-F......................................................... 52 Table 14 In location 2 segments of 6 labelled A-F......................................................... 52 Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation viii 1 Chapter 1: Introduction Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 8 Figure 4 Cyclone season from November to April tends to produce a significant number of floods. Adopted from 'Flooding in the Fiji Islands between 1840 and 2009 (Source: McGree et al, 2010) According to JICA (1998), the Nadi river discharges 300m2/s into the sea, and with the onset of cyclone events during the cyclone season, increases river discharge at the mouth of the river. Subsequently, during a flood event in 2007, it is reported that infrastructural damages in Nadi town reached an astounding US$1 million dollars (Government of Fiji, 2009). Figure 5 The Nadi basin covers 490 km2 (a) Fiji island location in the worldeast of Australia (b) The chosen study area for this research, the Nadi watershed with Strahler stream order classification Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 9 The chosen study area for this research is the Nadi watershed (figure 5b). According to the Strahler classification system the stream network constructed using ArcGIS 10.1, has 123 1st order streams (total length: 194.65km), 28 2nd order streams (total length: 83.45km), 4 3rd order streams (total length: 79.33km), 2 4th order streams (total length: 23.98km), and 1 5th order stream (total length: 14.89km). It also has a total area of 506.69km2, with the highest altitude of 1042m and the lowest of 3m. All basin attributes will be further discussed in the chapters ahead. 1.5 Rationale for this research In January 2009, Fiji experienced a tropical depression which subsequently resulted in intense periods of high precipitation, especially hitting the western part of the main island of Viti Levu, especially the towns of Nadi, Ba and Rakiraki (figure 6). According a report done by Government of Fiji (2009), this was reported as one of the worst loods in Fiji since the 1930s, with most of the low lying areas experiencing lood levels of up to 3–5 metres. The report goes on to state that the impact was greatest in the Western Division with costs estimated at about FJ$ 81 million. Nadi was one of the most affected towns, where businesses suffered a loss of over FJ$ 20 million. Lal et al. (2009) estimate the total economic cost of the January 2009 floods on the sugar industry (infrastructure, losses to growers and millers) to be FJ$ 24 million. The report stated that with such vast devastation of major townships, visitor arrival numbers declined. The total rehabilitation cost was about FJ$ 73 million, diverting a major portion of government budget. Additionally, humanitarian costs of about FJ$ 5 million were incurred. Most rehabilitation works are still continuing, with assistance from donors and development partners (Government of Fiji, 2009). Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 10 Figure 6 The January 2009 flood caused major infrastructural damage in the western parts of Viti Levu (a) An aerial view of a inundated Nadi town (b) Main road Nadi town submerged in water (c) Villagers located near the Nadi river and tributuaries flooded (d) A roundabout on the main road submerged in flood waters (e) Flood waters reach waist high in some areas as shown up two men cheering for Fiji (f) Shops are forced to shut down businesses when flood waters inundate the town during the 2009 flood. (Source: GoogleImages) Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 11 Floods in Fiji are of high significance and according to FMS (2001), over 160 floods from 18402000 have been recorded by the Fiji Meteorological Services (appendix 1) which to be put into perspective, there are floods occuring at an average of 10 per decade. Furthermore according to Chandra & Dalton (2010), with the increased intensity of floods throughout various parts of the country over the past years as well as future projections, fooding has become a high priority for the Government of Fiji, and thus the need for different type of research concerning floods. 1.6 Literature Review The study of floods in Fiji is not scarce, but at the same time there is not a lot of research that has been done on it. However, several research has been undertaken by both academic researchers as well as organizations to better understand the impacts and the nature of floods, particularly in a developing country like Fiji (Terryand Raj, 2004; Yeo et al, 2007;Yeo, 2000; JICA, 1998). This section will serve as a literature review of a number of published journal articles, books and reports on floods in terms of geohydrological characteristics of drainage basins. All journal articles and reports used in this literature review section was obtained form open access databases such as EBSCOhost, ProQuest and Web of Knowledge and SOPAC virtual library which has been summarized and tabulated in table 2. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 12 No. Author Summary 1. (JICA, 1998) (Chandra & Dalton, 2010) (Terry & Raj, 1999) (Yeo, 2000) (Yeo et al, 2007) - Several authors that research on flood events in Fiji, the causes, the impacts and recommendations on mitigation activities that can be carried out to minimise the impacts of floods in Fiji. - Includes journal articles and reports. 2. (Horton, 1945) (Strahler, 1957) (Strahler, 1964) - Considered as pioneers in the field of hydrology, Horton researched on the functionality of a watershed and how its attributes are vital in determining variables such as peak discharge, time of concentration and its morphometric parameters with relation to intense precipitation. - Strahler, is well known for classifying streams according to the power of their tributaries. - Published book 3. (Hajam, Hamid, & Bhat, 2013) (Rastogi & Sharma, 1976) (Ritter, Kochel, & Miller, 1995) (Wilson, 1990) (McCuen, 2005) (Beven, 2004) (Pilgrim & Cordery, 1993) - Collectively these 8 authors cover the topics of Hydrographs, Geomorphology, and morphometry. - Discussion on factors such as climatology, land use, geology, meteorology that all contribute to the workings of a watershed and subsequently, the formation of the basin and stream functions. - Includes majority journal articles, and published books Table 2 A brief summary of the literature review, the authors and their research. 13 Basin Morphometry & Geomorphology of Watershed Morphometric studies in the field of hydrology were first initiated by Horton (1940) and Strahler (1945). The morphometric analysis of the drainage basin and channel network play an important role in understanding the geo-hydrological behavior of drainage basin and expresses the prevailing climate, geology, geomorphology, structural antecedents of the catchment (Hajam, Hamid, & Bhat, 2013). Various important hydrologic phenomena can be associated with the physiographic characteristics of drainage basins such as size, shape, slope of drainage area, drainage density, size and length of the contributories etc (Rastogi & Sharma, 1976). According to Hajam et al (2013) analysing drainage basin is important in any hydrological investigation as assessments of relation are important between runoff characteristics, and geographic characteristics of drainage basin systems. Increasingly studies have used the patterns of basin morphometry to predict or describe geomorphic processes (Ritter, Kochel, & Miller, 1995). Figure 7 Linear and areal morphometry formulas used to mathematical calculate basin morphometry (Source: Ritter et al, 1995) These are made easy with the introduction of geospatial techonology such as ArcGIS. Geographical Information System (GIS) techniques are now-a-days in use for assessing various terrain and morphometric parameters of the drainage basins and watersheds, as it provide a flexible environment and an important tool for the manipulation and analysis of spatial information of which can be calculated mathematically using formulas illustrated in figure 7. The method of studying basin attributes in Fiji using this method of linear and areal morphometry is not widely recognized, and would be vital considering the input Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 14 needed for calculating these parameters can be easily attained from spatial analysis tool with the ArcGIS. Horton (1945) introduced geometric processes between number of streams, and corresponding drainage areas which are now known as Hortons Laws. The first important set of values is linked to the morphology of the catchment. For instance the shape of the stream network, determines the discharge time. Catchment shapes vary greatly and reflect the way in which runoff will be distributed, both in time and space (figure 8). In wide, fan-shaped catchments the response time will be shorter with higher associated peak discharges as opposed to long, narrow catchments. In circular catchments with a homogeneous slope distribution, the runoff from various parts of the catchment reach the outlet more or less simultaneously, while an elliptical catchment equal in area with its outlet at one end of the major axis, would cause the runoff to be more distributed over time, thus resulting in smaller peak discharges compared to that of a circular catchment (McCuen, 2005). In order to understand the interaction between the different variables influencing the catchment response time and resulting runoff, it is necessary to view all the catchment processes in a conceptual framework, consisting of three parts: (i) the input, (ii) the transfer function, and (iii) the output (McCuen, 2005). Floods are generated in catchment areas in which runoff, resulting from rainfall, drains as streamflow towards a single outlet. Rainfall is the input. Figure 8 The shape of stream network affects discharge time. Example: A fan shaped catchment (B) will have a faster stream rise and similarly a faster fall than a dendritic shaped catchment (C) because of shorter travel times (Source: Wilson, 1990). The catchment characteristics define the nature of the transfer function, since rainfall losses occur as the catchment experiences a change in storage while it absorbs (infiltration), retains or attenuates (surface depressions) and releases some of the rainfall through subsurface flows, groundwater seepage and evaporation. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 15 Furthermore, the effective rainfall exits within the catchment as the streamflow output, i.e. the direct runoff contributing to flood peaks. However, runoff generation in catchments is highly variable both in time and space, depending not only on the amount and intensity of rainfall, but it is also affected by the different physiographical parameters, or combinations thereof, which describe the catchment characteristics (Beven, 2004; Chow, 1988; Pilgrim & Cordery, 1993). Hydrographs Hydrographs are defined by Sarangi (2007), as „a graph showing the rate of flow (discharge) versus time past a specific point in a river or channelor conduit carrying flow and is usually expressed in cubic meters or cubic feet per second (cms or cfs)‟. Each depending on the users intent, a hydrograph can be used basically for acquiring run off for a particular time period for events such as floods or storms (figure 9). In the article, the author attempted to evaluate the models for accurate runoff prediction from ungauged watershed using ArcGIS. This was accomplished by running three types of models namely the (i) exponential distributed geomorphologic instantaneousunity hydrograph (ED-GUIH) model, GIUH based Clark model and spatially distributed unit hydrograph (SDUH) was used to predict the direct runoff hydrograph (DRH) for their study area. Furthermore, the researchers attempted to compare the generated DRH against the observed DRH at the watershed outlet. The result of Sarangi (2007) research concluded with main point of using EDGUIH model to predict DRH rather then the other two models because it performed better when predicting the direct runoff hydrograph for short duration storm events (≤6h) with an error margin of 4.6-22.8% for the study watershed. Figure 9 Different models of hydrographs as adopted from Sarangi (2007) (a) A simple flood hydrograph (b) A 'hybrid' hydrograph showing flood and storm DRH Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 16 Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 17 Chapter 2: Methodology & Materials Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 24 noted, that this 52 years of precipitation data also had three years of data missing, from 1971-1973, but was included in the calculation of hydrological dynamics as explained in the following section. Additionally, this precipitation data was analysed to determine longterm average monthly and annual precipitation using Microsoft Excel 2010. Futhermore, this same data was correlated to calculate „Return Period‟, „Time for concentration‟ and „Peak discharge for three return periods‟ for individual sub basins in the Nadi watershed which will be explained in the following chapter. As with all research the need for viable data ensures the researcher with a good foundation to building their research. The principle reason that this data source was chosen is the location of the NOAA station in the study area which contributes to the validity of the data gathered. Additionally, attaining data from Fiji Meteorological Offices in Fiji pertaining to precipitation data dating back to the 1950‟s was unsuccessful, hence this resource was used. (b): Tropical Rainfall Measuring Mission (TRMM) The „Tropical Rainfall Measuring Mission‟ (TRMM) is a joint space mission between NASA and the Japan Aerospace Exploration Agency (JAXA) designed to study and monitor tropical rainfall on the planet from space. The instruments or sensors used to gather data from the satellite include; Precipitation Radar (PR), TRMM Microwave Imager (TMI), Visible Infrared Scanner (VIRS), Clouds and Earth Radiant Energy System (CERES) and Lighning Imaging Sensor (LIS). However, the main instrument for gathering precipitation is TMI. Correlating the data from TRMM was carried out in order to try and ascertain peak precipitation periods in Fiji using Microsofts Excel program. Unfortunately, the data from TRMM had values that were not coinciding with that of the precipitation data from NOAA which illustrates a major set back for this dataset. A major limitation when dealing with TRMM precipitation data is its underestimation of precipitation data due to its high latitude, however, analysis of both datasets (NOAA and TRMM) will be explained in the latter sections, to compare the two datasets. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 25 Calculating Hydrological Dynamics: Time of Concentration, Peak Flow & Return Period Using the precipitation data (1961-2013) obtained from NOAA, the hydrological properties (i.e. time of concentration (tc), peak flow (Q) and return period (T)) for the Nadi watershed as well as the four sub basins was calculated. The first parameter that was calculated was the time of concentration (tc) and this was accomplished by using the Giandotti formula given by Grimaldi et al (2012) (figure 13): The second parameter to be calculated was „peak flow‟ for a return period of 100 years. This was adopted from the Giandotti formula (figure 14) (Giandotti, 1953; Lencastre & Franco, 1984). In order to calculate the return period, the 52 years of precipitation data was arranged in descending order of magnitude. This was to allow a frequency analysis to be carried out, after which the Welbull method (figure 15) was applied to get the return period (T): Figure 15 The formula for calculatting Return Period (T) as adopted from Dawood et al (2012), using the Welbull method Figure 13 Time of concentration Giandotti formula as adopted from Grimaldi et al (2012) Figure 14 Giandotti formula for calculating peak flow as adopted from Giandotti (1953) Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 26 2.1.3 Satellite Imagery: Google Earth Determining channel variation by using satellite images was a method adopted for this research. Several authors have used this simple method to determine channel varioation like (Erskine, 2013) (Friedman & Lee, 2002) (Lyons & Beschta, 1983). Satellite images from the years 2005, 2012 and 2014 were downloaded for selected areas that had the main Nadi channel running through it. This is illustrated in chapter 5, figures 25 and 27. The main purpose of choosing these specific years was to try and ascertain if the width of the main channel changed before the 2012 flood event. The river channel was digitized in ArcGIS 10.1 and thewidth of respective segments was averaged. Furthermore, it is important to note that by simple observation of the 3 different satellite images, one is able to note the change in river channel formation. This is solidified by the changing width of the river channels. 2.2 Secondary Data As illustrated in table 3, secondary sources of data for this research was largely from open source journal databases such as web of knowledge, ScienceDirect, EBSCOHost and published books as well as online reports. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 27 Chapter 3: Morphometric Analysis and Physical Characteristics of Nadi Basin Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 28 3. Characteristics of Hydrological Basin for Nadi River TheNadi river is located on the northwestern (NW) side of the main island of Viti Levu. Since the Nadi river is on the lee side of Viti Levu, it experiences long periods of dry spells. The valleys and hills along the Nadi river are rich agriculture lands. Moreover, the monthly average temperature ranges from 24 – 27°C and rainfall ranges from 1440 – 1993 mm/yr. The Nadi district produces the largest volume of sugarcane annually in Fiji, and sugar is the main economic activity, followed by tourism, manufacturing, forestry and fisheries respectively (Ledua et al, 1996). Nadi town is the main business centre of the Nadi district and is located on the western side of the river delta approximately 8 kilometres inland from the coast. The population of Nadi as a province from the last census in 2007 is approximately 231 760, and Nadi town has a population of 42 284 (Fiji Bureau of Statistics, 2010). This chapter will discuss the result of the anaylsis carried out on the Nadi watershed based on its hydrological and physical characteristics. Also, precipitation data from NOAA was analysed which was used to try and channel variation which will be discussed in the latter sections. Figure 16: Hydrological watershed and four sub catchments (A,B,C,D) for Nadi created from SRTM based DEM with a 90m resolution. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 29 3.1 Morphometric Analysis of Nadi Basin A major emphasis in geomorphology over the past several decades has been on the development of quantitative physiographic methods to describe the evolution and behaviour of surface drainage networks (Horton, 1945). All hydrologic and geomorphic processes occur within a watershed and because of these the morphometric features at the watershed level may contain vital information regarding its formation and development (Singh N. , 1990). For this research, the author has used SRTM 9 based Digital Elevation Models (DEM) with a resolution of 90m and GIS to evaluate linear and areal analysis as discussed in the following sub sections. 3.1.1 Linear Analysis: The hydrological basin for Nadi covers 506.69 km² with a total of 5 stream orders according to the Strahler Stream Hierarchy Classification system. The length of the main Nadi river under the 3rd, 4th and 5th order sums up to 45.60 km. The sum of the 1st order stream magnitude, and has a total of 162 segments, with the total length of 194.56 km. This is tabulated in „Fluvial Hierarchy of Nadi river‟ table 3 below: STREAM ORDER LENGTH (M) LENGTH (KM) Nº OF SEGMENTS 1 194566 194.566 162 2 83456 83.456 64 3 79338 79.338 65 4 23984 23.984 18 5 14899 14.899 12 Table 3 Fluvial hierachy of river channels in the Nadi watershed based on Strahler classification of stream orders Bifurcation Ratio (Rb) The bifurcation ration is defined by Pareta et al. (2012) as „ratio of the number of stream segments of given order ‘Nu’ to the number of streams in the next high order’. Moreover, according to Strahler (1964) the bifurcation ratio is dimensionless property and generally ranges from 3.05.0, whilst the lower values of Rb are characteristics of the watersheds which have suffered less structural disturbances and the drainage pattern 9 Shuttle Radar Topography Mission Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 30 has not been distorted because of the structural disturbances (Nag, 1998). In contrast, the higher the Rb indicates more structural damages undergone in areas that that the stream order passes through. BIFURCATION RATION (RB) STREAM ORDER VALUE 1 2.531 2 0.984 3 3.611 4 1.5 5 - Table 4 Bifurcation ratio adopted from Horton (1945) for streams in the Nadi watershed According to the index parameter for bifurcation ratio given by Strahler (1964), and the results in table 4, it can be inferred that the watershed being studied has suffered less structured disturbances. Segment Length (LU) and Stream Length Ratio (RL) Horton (1945, as cited by Pareta et al., 2012) states that ‘length ratio is the ratio of the mean (Lu) of segments of order (So) to mean length of segments of the next order (Lu – 1), which tends to be constant throughout the successive orders of a basin’. Furthermore, Horton‟s law of stream length refers to the mean stream lengths of stream segments of each successive orders of a watershed tend to approximate a direct geometric sequence in which the first term (stream length) is the average length of segments in first order (Pareta et al., 2012). SEGMENT LENGTH STREAM ORDER VALUE 1 1201.02 2 1304 3 1220.58 4 1332.44 5 1241.58 Table 5 Segment length (LU) and relationship According to Singh et al. (1997), change of stream length ratio from one order to another order indicating their late youth stage of geomorphic development. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 31 RELATION TO LENGTH ORDER OF SEGMENTS VALUE 1 - 2 1.085739543 3 0.936031147 4 1.091644469 5 0.931808706 Table 6 Order of segments and relation to length The area of the watershed is another important parameter like the length of the stream segment. Schumm (1956) established an interesting relation between the total watershed areas and the total stream lengths, which are supported by the contributing areas. The author has computed the basin area, basin length and basin morphometry parameter for Nadi basin using ArcGIS 10.1 software as displayed in table 7 below and other morphometric parameters for areal anaylsis which is explained in the following sub sections. 32 NO. PARAMETER FORMULA DESCRIPTION REFERENCE RESULT (i) Basin Area - AArea (km²) - 506.69 (ii) Basin Length - LLength (km) - 40.8 (iii) Basin Perimeter - PPerimeter (km) - 134.91 (iv) Compact Coefficient P – Perimeter(Km) A – Area (Km²) Gravelius (1914) 1.67 (v) Relationship between width and length I – Width (Km) L – Length (Km) - 0.62 (vi) Elongation Ratio Re = 2 / Lb * (A / π)0.5 A – Area (Km²) L – Width (Km) Schumm (1956) 0.62 (vii) Circularity Ratio Rc = 12.57 * (A / P²) A – Area (Km²) P – Perimeter (Km) Miller (1960) 0.35 (viii) Relationship between length and area L – Length(Km) A – Area (Km²) - 1.81 (ix) Drainage Density A – Area (Km²) C – Total length of segments (Km) Strahler (1957) 0.78 (x) Coefficient of Maintenace Dd – Drainage Density - 1.28 (xi) Segment density of basin 𝛴n – Sum of all segments from all stream orders A – Area (Km²) - 0.63 (xii) Hydrographic density or Frequency of Thalweg N1 – Nº of segments in 1st Order A – Area (Km²) - 0.31 (xiii) Torrentiality Coefficient Dh – Hydrological Density Dd – Drainage Density - 0.24 Table 7 Important morphometric parameter for basin geometry 33 3.1.2 Areal Analysis Compact Coefficient (Cc) According to Gravelius (1914) compact coefficient of a watershed is the ratio of perimeter of watershed to circumference of circular area, which equals the area of the watershed. The Cc is independent of size of watershed and dependent only on the slope. The calculated Cc for Nadi basin is 1.67 that was calculated using the formula provided in table 7 and calculated in a simple Microsoft Excel spreadsheet. The calculated Cc suggests an elongated basin as it falls within the range of 1.5 – 1.75 as suggested by Singh (1997). Relationship between Width (I) and Length (L) The parameter calculated for this particular index provides comparative data between the two variables of width and length (Pareta & Pareta, 2012). Morphometric index for the parameter indicate if the value is close to then width and length are closer to each other.The index calculated as shown in table 7 for the Nadi watershed being studied is 0.62. Elongation Ratio (Re) According to Schumm (1956) elongation ratio is defined as the ratio of diameter of a circle of the same area as the basin to the maximum length. However, Strahler (1957) states that this ratio runs between 0.6 – 0.1 over a wide variety of climatic and geologic types. The varying slopes of watershed can be classified with the help of the index of elongation ratio such as: ELONGATION RATIO (RE) SHAPE 0.9 - 0.10 Circular 0.8 – 0.9 Oval 0.7 - 0.8 Less elongated 0.5 - 0.7 Elongated < 0.5 More elongated Table 8 The elongation ratio describes the overall shape of the basin using Strahlers classification for this index Furthermore, the calculated elongation ratio for Nadi basin is 0.62 calculated by the author using the formula provided in table 7. Subsequently this would deem our basin as elongated according to the classification provided in table 8. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 40 4. Hydrology The recurrence of extreme precipitation anomalies that result in floods or droughts is a normal component of natural climate variability (WMO, 2009). The use of precipitation data to analyse time of concentration, peak discharge for the return period of 10, 50 and 100 years as well as the return period for the Nadi watershed was carried out to better understand the nature of flash floods in the Nadi watershed. It is widely known that floods are more often caused by intense precipitation. In recent years, heavy precipitation events have resulted in several damaging floods in Fiji, the most recent ones being the flood events of January 2009 (figure 21) which was one considered on of the worst flood occurrence. In this chapter, precipitation data for a period of 52 years will be analysed (19612013) to calculate several parameters such as time of concentration, peak discharge and return period. Addiotionally, a sketch of a segment of the Nadi river for the years 2004, 2005, 2009, 2013 and 2014 was also done to compare channel changes. All images used in this analysis was from GoogleEarth pro, and chosen based on period of flood events or heavy precipitation, minimum cloud cover and most importantly availability of images on Historical Imagery tool on GoogleEarth. Figure 20 Flood events as recorded by Yeo (McGree et al, 2010) Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 41 4.2 Analysing Precipitation Data The precipitation data from NOAA was graphed according to the daily maximum values recorderd for each year so as to demonstrate the nature of heavy precipitation in Nadi. The day which recorder the maximum amount of rainfall was then graphed as illustrated in figure 21. Additionally this was also done to figure out whether the reported flood event coincided with the value of heavy precipitation from NOAA. In order to to accomplish this, the sum of daily precipitation was summed up and sort according to the days that recorded the highest amount of rainfall and using the Webull method (figure 12) as adopted from Dawood (2012), the return period was calculated. Several studies have examined changes in occurrence of larger scale flooding events under climatic change (Booij, 2005), but little research has been done on changes in small scale flood events. This may be driven by the limited availability of high spatial and temporal resolution precipitation output from climate change models (typically climate model precipitation estimates are daily or monthly values and at spatial scales of 100s of kilometers). For our study area, fortunately Flash flood occurrence is often associated simply with “heavy precipitation”, Doswell et al. (1996). However, Brooks and Stensrud (2000) compared the climatology of heavy precipitation, with the flash flood climatology, and concluded that flash floods occur 17 times less frequently than heavy precipitation. Flash flood occurrence is not driven by heavy precipitation alone but by meteorological, climatic and physiographic influences including precipitation intensity, topography, and soils properties (Georgakakos, 1986). Figure 21 Precipitation data tabulated and graphed for a period of 52 years (NOTE: Missing data for 1970-1972) (Source: NOAA) Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 42 A trendline as illustrated in figure 22 illustrates a maximum monthly rainfall versus a return period for upto 50 years and further, a predicted value (in red) from 60100 years using a regression line method on Microsoft excel. Basically, it illustrates that in 100 year return period the maximum rainfall predicted will be 440.32mm, and with the return period of 50years the calculated amount of precipitation received was 365mm. In hydrograph analysis, time of concentration is the time from the end of excess rainfall to the point on the falling limb of the dimensionless unit hydrograph (point of inflection) where the recession curve begins (Beven, 2004). In the next sub section, the time concentration for Nadi watershed, as well as for the four sub basins was calculated using the methods defined in chapter 2. 4.3 Time of Concetration (tc) Time of concentration represents the hydrologic response time of an urban watershed, is measured in hours and is defined by Akan et al (2003) as the time required for all parts of a basin to contribute to discharge at the outlet simultaneously. In simple terms, it is the time needed for water to flow from the most remote point in a watershed to the watershed outlet. It can be calculated using several methods, however for this particular research the Gumbel method (Grimaldi et al, 2012) was chosen as the formula for obtaining tc..The principle reason for choosing this formula was mainly attributed to the fact that the required input (area, length and height) was easily obtained from the demarcated Nadi watershed carried out in ArcGIS. The main Nadi watershed had a time concentration value of 11.8hrs as illustrated in table 9 below. Figure 22 Trendline for the return period versus recorded maximum rainfall data. Predicted values for precipitation data for 60, 70, 80, 90 and 100 years are in red (Source: NOAA rainfall data) Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 43 Sub-Basin Parameter Unit Main watershed A B C D Area (km2) 506.7 224.6 92.6 129.3 60.2 Length (km) 40.8 34.5 11.2 14.7 6.5 Height (m) 254.8 149.7 183.3 169.6 165.2 Tc (hrs) 11.8 11.4 5.1 6.5 4.0 Table 9 Time of concentration for Nadi watershed and the four sub basins using the Gumbel method Considering that the size of the watershed is 506.7 km2, this illustrates the rapid flow of river water towards the mouth of the river often incurs the quick rise of river water subsequently flooding the areas near the mouth. The same can be deduced from the time of concentration for the four sub basins as well: Sub basin A: area of 224.6 km2 and tc of 11.4 hrs, Sub basin B: area of 92.6 km2 and tc of 5.1 hours, sub basin C: area of 129.3 km2 and tc of 6.5 hoursm sub basin D: area of 60.2 km2 and tc of 4.0 hours (which exhibit the fast concentration time). These results depict an direct relationship between area of basin and time of concentration i.e. the more the area the more rapid the rain water flows and conctrates towards the mouth. Peak Discharge for Return Period of 10, 50 and 100 years The calculation of the peak discharge for 3 return periods (10, 50, 100 years was made through the application of Giandotti‟s formula as explained in chapter 2. In the main watershed, the value of peak discharge expected for a return period of 100 years is 6165.27 m3/s (table 10). As shown in table 10, sub basin „A‟ for a return period of 10 years the value of the peak discharge is 1702.5 m3/s and these values almost double in 100 years. This can also be highlighted in sub-basin D mainly because it is the smallest sub-basin of the four, with an area of 60.2, and gather great volumes of water in 4hours, for example 1726.96 m3/s in 100 years. In contrast, the sub-basin B is a little larger than D, but the peak discharge is smaller, and takes more time (5h) to achieve the most downstream point of the sub-basin. In quantitative terms, the values reach 815.47 m3/s for a return period of 10 years and 1277.54 m3/s in 100 years. Main basin and Sub-Basin Peak Discharge for Return Period (Years): 10 50 100 Main Watershed 3762.47 5462.10 6165.27 A 1,702.51 2,469.25 2,785.45 B 1,090.88 1,550.72 1,726.96 C 1,336.50 1,911.22 2,136.56 D 815.47 1,151.85 1,277.54 Table 10 Peak discharge for the 4 sub basins in a return period of 10, 50, 100 years Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 44 4.2. Channel Variation One of the objectives of this research is to investigate the physical channel changes on the main Nadi river on two areas of the watershed. For this, satellite images of the Nadi river was digitized on ArcGIS, and statistics (width and area) of the channel was calculated to estimate if any physical variation occurred on the river channel for 3 time periods i.e. 2005, 2012 and 2014. These three time periods were chosen to compare if any changes occurred, especially preflood event which is 2005, during a flood even2012, and post flood event, 2014. The methodology adopted to obtain the width of the digitized river channels was by drawing transects between the two sides of the river. Using ArcGIS, the length of the transects weres calculated to obtain the average length, as well as minimum and maximum length of the width. Furthermore, with these data a comparative analysis was done between the three time periods in the two different location to try and ascertain any channel changes. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 45 Selected Areas The two selected areas were from two two segments of the main channel of the watershed, that is Nadi river. The first near the source or headwaters (L1), and the seond being near the mouth, on the NE end of the watershed (L2) (figure 23). These areas were selected to demonstrate two the impacts of intense flood waters that occurred in 2012 in two different regions of the watershed. Figure 23 The two selected areas of study. Location 1 (L1) was towards the headwater and Location 2 near mouth of river (Source: SRTM Elevation Data) Location 2 contains the Nadi town within its limit as illustrated in figure 25 which is located right next to the Nadi river on an area of low elevation. L2 L1 46 (a) (b) (c) Figure 24 From the Location 1(a) 2005 satellite image showing the Nadi river preflood event(b) 2012 satellite image showing flood event occuring (c) 2014 post flood satellite image (Source: GoogleEarth) 47 a. Sketched Nadi river channel from 2005 (a) (b) (c) Figure 25 (a) 2005 channel sketched in ArcGIS showing pre flood event (b) 2012 channel during flood event (c) 2014 channel post flood event. 48 For the three time periods, 3 segments were delimitated and transects were drawn across from one end of the channel to the other end. The statistical output of the length of the transects enabled the researcher to make deductions on channel change. As can be seen from merely looking at the satellite images (figure 25) for the three time periods, a slight change in the width of the channel can be noticed. The most significant of which occurs in the levee which can be a result of the overflow of food waters from the river channel, thus extending the width. In table 11, a slight increase in the average length from the 2005 channel as compared to the 2014 channel further consolidates the that channel variation does indeed occur post flood event. This also can be said for the other parameters that increases such as the area and minimum length of the width of the river channels. Year Area (m2) Average Length (m) Minimum Length (m) MaxLength (m) 2005 152214.9 19.27 12.75 45.41 2012 284561.7 44.17 27.88 61.61 2014 187003.9 27.66 15.25 39.96 Table 11 Calculating the average, minimum, maximum length of transects that were drawn on different segments of the river channel in Location 1 It should be noted however that channel changes however, are not soleyly caused by extreme rain events. The impact that humans have on the physical landscape is also a contributing factor. The river channels in question are located near places of residents, and perhaps further study into the impacts of development on river channel formation is warranted. 49 Figure 26 From Location 2 (a) 2005 satellite image showing the Nadi river preflood event(b) 2012 satellite image showing flood event occuring (c) 2014 post flood satellite image. NOTE: The new meander formed post flood event. (Source: GoogleEarth) (a) (b) (c) Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 56 5.1 Recommendation As a recommendation, the researcher would like to suggest the development of a proper elevation data that would ensure the construction of a Digital Elevation Model efficient enough to delimitate the boundary of the watershed properly. Secondly, in terms of channel variation and using satellite images to make deductions, more ground work is needed. Collection of data such as channel depth, proper measurements of width, area, identifying human impacts towards channel variation is required and can be furthered more with more research. However, with recommendation, it should also be noted that this is the step in the proper direction for researchers, by input of research work, setting the platform for more to be done in the future. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 57 Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 58 Bibliography (GoF), G. o. (2009). Consolidated report of flash floods 816 January, 2009 damanges sustained and necessary responses, rehabilitation and reconstruction. Suva, Fiji: Office of the Prime Minister. (MRD), S. (. (2007). Report on GEF hotspot workshop for Fiji. Suva, Fiji: Government of Fiji. (WMO), W. M. (2009). Integrated Flood Management Concept Paper. Geneva, Switzerland: World Meteorological Organization. Aderogba, K. (2012). Substantive causes and effects of floods in western Nigeria and sustainable development of the cities and towns. Journal of Emerging Trends in Educational Research and Policy Studies (JETERAPS), 551-560. Akan, A., & Houghtalen, R. (2003). Urban hydrology, hydraulics and stormwater quality. New Jersey: John Wiley & Sons. Beven, K. (2004). Robert E. Horton's perceptual model of infiltration process. Hydrological Processes, 3447-3460. Chandra, A., & Dalton, J. (2010). Mainstreaming Adaptation within intergrated water resources management (IWRM) in Small Island Developing States(SIDS). Building Resilience to Climate Change: Ecosystems based adaptation and lessons learnt from the field, pp. 4761. Chow, V. (1988). Runoff. In V. In. ed. Chow, Handbook of Applied Hydrology (pp. Chapter 14: 1-54). New York, USA: McGraw-Hill. Dawood, G., Mizra, M., & AlGhamdi, K. (2012). GIS based estimation of flood hazard impacts on road network in Makkah city, Saudi Arabia. Environmental Earth Science, 22052215. Dilley, M., Chen, R., Deichmann, U., LernerLam, A., Arnold, M., Agwe, J., et al. (2005). Natural disaster hotspots: a global risk analysis. Washington, D.C: International Bank for Reconstruction and Development/ The World Bank . Erskine, W. (2013). Synchronous linked changes in rainfall, floods and river channel changes in southeastern Australia since European settlement. Climate and Land Surface Changes in Hydrology, p.3037. Fiji Bureau of Statistics. (2010). Fiji Facts and Figures as at 1st July 2010. Suva: Fiji Bureau of Statistics. Friedman, J., & Lee, V. (2002). Extreme Floods, and Riparian Forests along Ephemeral Streams. Ecological Monographs, 409-425. Ghumman, A., Ahmad, M., Hashmi, H., & Kamal, M. (2012). Development of geomorphologic instantaneous unit hydrograph for a large watershed. Environmental Monitering Assessment, 184: 31533163. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 59 Giandotti, M. (1953). Considerazioni idrologiche Sulle Piene de Po: con speciale riguardo alla piena del Novembre 1951. Instituto poligrafico dello Stato. Gravelius, H. (1914). Flusskunde. Berlin: Goschen'sche Verlagshandlung. Grimaldi, S., Petroselli, A., Tauro, F., & Profiri, M. (2012). Time of concentration: a paradox in modern hydrology. Hydological Sciences Journal, 217-228. Hajam, R., Hamid, A., & Bhat, S. (2013). Application of morphometric analysis of geohydrological studies using geospatial technologyA case study of Vishav drainage basin. Hydrology Current Research, 1-12. Horton, R. (1945). Erosional Development of streams and their drainage basins . Bulletin of the Geological Society, 275-370. JICA. (1998). The study on watershed management and flood control for the four major Viti Levu rivers in the Repulic of Fiji Islands. Tokyo: Japan International Cooperation Agenc/ Yachiyo Engineering C. Ltd. Karanth, K. (1987). Ground water assessment development and management. New Delhi: Tata McGrawHill Publishing . Kibler, D. (1982). Urban Stormwater Hydrology. Washington, D.C: American Geophysical Union, Water Resources Monograph 7. Kumar, S. (2006). Importing ASCII Grid Data into GIS/Image processing software. Oak Ridge, Canada: The Oak Ridge National Laboratory Distributed Active Archive Center . Lal, P., Rita, R., & Khatri, N. (2009). Economic costs of 2009 floods in Fiji sugar belt and policy implications . Gland, Switzerland: Internation Union of Conservation of Nature (IUCN). Ledua, E., Matoto, S., Sesewa, A., & Korovulavula, J. (1996). Freshwater Clam Resource Assessment of the Ba River. Noumea: South Pacific Commission. Lencastre, A., & Franco, F. (1984). Lições de Hidrologia. Lisboa: Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa. Lencastre, A., & Franco, F. (1987). Lições de Hidrologia. Lisbon: Faculdade de Ciências e Technologia, Lisbon. Lyons, J., & Beschta, R. (1983). Land Use, Floods and Channel Changes: Upper Middle Fork Willamette River, Oregon (1936-1980). Water Resources Research, 463-471. McCuen, R. (2005). Hydrolic Analysis and Design 3rd Ed. New York, USA: Upper Saddle River. McGree, S., Yeo, S., & Devi, S. (2010). Flooding in the Fiji Islands between 1840 and 2009. Suva: Risk Frontiers. Miller, O., & Summerson, C. (1960). Slope Zone Maps. Geographical Review, 194-202. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 60 Nag, S. (1998). Morphometric Analysis using remote sensing techniques in the Chaka subbasin, Purulia district, West Bengal. Journal of Indian Society of Remote Sensing, 6976. NCDC, N. C. (2003). Our strategic vision for stewarding the nationas climatic data. Asheville: National Climatic Data Center of NOAA. Pallard, B., Castellarin, A., & Montanari, A. (2009). A look at the links between drainage density and flood statistics. Hydrology and Earth System Sciences, 1019-1029. Pareta, K., & Pareta, U. (2012). Quantitative Geomorphology Analysis of a Watershed of Ravi River Basin, H.P India. International Journal of Remote Sensing and GIS, 41-56. Pilgrim, D., & Cordery, I. (1993). Flood Runoff. In D. Maidment, Handbook of Hydrology (pp. Chapter 9: 1-42). New York, USA: McGraw-Hill. Rastogi, R., & Sharma, T. (1976). Quantitaive Analysis of drainage basin characteristics. Journal of Soil and Water Conservation in India, 18-25. Ritter, D., Kochel, R., & Miller, J. (1995). Process Geomorphology 3rd Edition. Dubuque, I.A: W.C. Brown Publishers. Sarangi, A., Madramootoo, C., Enright, P., & Prasher, S. (2007). Evalution of three unit hydrograph models to predict the surface runoff from a Canadian watershed. Water Resource Management, 21: 1127-1143. Schumm, S. (1956). Evolution of drainage systems & slopes in Badlands at Perth Anboy, New Jersey. Bulletin of the Geological Society of America, 597-646. Services, F. M. (2001). Lists of floods occuring in Fiji Islands between 18402000. Nadi, Fiji: Fiji Meteorological Services (FMS). Sharifi, F., Samadi, S., & Wilson, C. (2012). Causes and consequences of recent floods in the Golestan catchments and Caspian sea regions of Iran. Natural Hazards, 533-550. Singh, N. (1990). Geomorphology of Himalayan Rivers. Jammu Tawi: Jay Kay Book House. Singh, S., & Singh, M. (1997). Morphometric analysis of Kanhar river basin. National Geographical Journal of India , 31-43. Strahler, A. (1957). Quantitive Analysis of Watershed Geomorphology. New York: Defense Technical Information Centre. Strahler, A. (1964). Quatitative geormorphology of drainage basin and channel network . Handbook of Applied Hydrology, 39-76. Terry, J., & Raj, R. (1999). Island environment and landscape response to 1997 tropical cyclones in Fiji. Pacific Science, 13: 257-272. UNDP, U. N. (2004). Reducing disaster risk: A challenge for development. New York: United Nations Development Programme, Bureau for Crisis Prevention and Recovery. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 61 Vojinovic, Z., & Abott, M. (2012). Flood Risk and Social Justice: From Quantitative to Qualitative Flood Risk Assessment and Mitigation. London: IWA Publishing. Wilson, E. (1990). Engineering Hydrology. London: Macmillan. Yeo, S. (2000). The Ba Community Flood Preparation Project. Suva, Fiji: SOPAC. Yeo, S., Blong, R., & McAneney, J. (2007). Flooding in Fiji: Findings from a 100 year historical series. Hydrological Sciences Journal, 52(5): 1004-1015. Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 62 Appendix Flash Floods in the Nadi watershed, Fiji: Morphometry, Precipitation and Channel Variation 63