Spatial and Temporal Analysis of Orographic Precipitation along the Darling Escarpment in Western Australia

Transcription

1 Spatial and Temporal Analysis of Orographic Precipitation along the Darling Escarpment in Western Australia Elinor Sterrett Supervisor: Dr Kyungrock Paik School of Environmental Systems Engineering The University of Western Australia

2 Cover Photo The Darling Escarpment in the Mist Courtesy of Wes Cooper 28 (i)

3 ABSTRACT Orographic precipitation in the Darling Escarpment has important implications for current and future water resources planning and management in the south west of Western Australia (SWWA). This rainfall mechanism is particularly important since this region has experienced a decline in average annual rainfall since the 197 s, largely thought to be attributable to climate variability and change. This study aimed to quantify the rate of precipitation in relation to elevation, to identify any annual or seasonal trends in orographic precipitation, to determine the influence of climate variability and change in relation to orographic precipitation, and to determine the mechanisms of orographic precipitation, in the Darling Escarpment. Precipitation data was analysed from twenty-one gauging stations from the greater region of the Darling Escarpment (GRDE) over the period of 19 to 27. A significant linear relationship was found between precipitation and elevation along the Darling Escarpment with an average rate of 1.6 mm/m. The rate of orographic precipitation was found to be strongest in winter and non-existent in summer. The dominant mechanisms of orographic precipitation along the Darling Escarpment were identified as unstable convection and stable upslope ascent, and spectral analysis linked winter rainfall in the SWWA to the mean sea level pressure (MSLP) off the coast. Time-series analysis confirmed a downward step change in annual and winter precipitation for the region in the mid-197 s, however this step change occurred one year earlier within the Darling Escarpment than on the adjacent Swan Coastal Plain. The same step change was observed in the annual and winter rate of orographic precipitation, suggesting precipitation has decreased to a greater extent within the Darling Escarpment than on the Swan Coastal Plain. It is suggested that this decrease is due to variability and change in atmospheric circulations, both natural and anthropogenic, rather than a change in the mechanisms of orographic precipitation. (ii)

4 (iii)

5 ACKNOWLEDGMENTS I would like to extent my sincerest thanks to my supervisor Dr Kyungrock Paik, for your knowledge, assistance and guidance. Your enthusiasm for this project and for always making time for me is very much appreciated. To everyone in SESE (past and present), thankyou for making the degree fun, providing distractions when needed and for putting up with me the computer labs over the last few months. I wish you all the best in the future. Thankyou to the MacGillivray s for putting up with me in your house over the past few months. A big thankyou to my family for supporting me through my entire education. To my sisters Isobel and Lola for always being there in the good times and the bad, and to my Mum for instilling in me the importance of a good education, you are my inspiration. Finally, thankyou Alastair for not letting me accept near enough as good enough, without you I would not have made it through this year. (iv)

6 TABLE OF CONTENTS Abstract... ii Acknowledgments... iv Table of Contents... v List of Figures... vii List of Tables... viii 1. Introduction Background Orographic Precipitation and Effect Air Saturation and Temperature Adiabatic Lapse Rate and Atmospheric Stability Airflow Dynamics Cloud Microphysics Knowledge Gaps South West of Western Australia General Climate Climate Variability and Change Approach Study Site Description Gauging Station Details Precipitation Analysis Orographic Effect Regression Analysis between Precipitation and Elevation Annual Time-series Analysis Spectral Analysis Seasonal Analysis Results and Discussion Orographic Effect and Precipitation Along the Darling Escarpment Rate of Orographic Precipitation Relationship between Precipitation and Elevation along the Darling Escarpment Seasonal Variability Comparisons with Previous Studies Mechanisms of Orographic Precipitation along the Darling Escarpment Variability and Change in Precipitation within the Greater Region of the Darling Escarpment Over Time Annual Variability and Trends Dominant Frequencies Seasonal Variability and Trends Rate of Orographic Precipitation Implications (v)

7 5. Conclusions and Recommendations References Appendix A Greater Region of the Darling Escarpment Precipitation relationships over time Appendix B Seasonal Analysis Results Appendix C Annual Time-series Analysis Results Appendix D Spectral Analysis Results Appendix E Annual and Winter Time-series Analysis Results and Trends (vi)

8 LIST OF FIGURES Figure 1 Adiabatic Lapse Rate in the context of orography (Krolak 21)... 6 Figure 2 Orographic Precipitation Mechanisms Illustration (a) Stable Upslope Ascent, (b) Partial blocking of the approaching air mass, (c) Down-valley flow induced by evaporative cooling, (d) Lee-side Convergence, (e) Convection triggered by solar heating, (f) Convection due to mechanical lifting above the level of free convection, and (g) Seeder-feeder Mechanism (Roe 25)... 9 Figure 3 South west region of Western Australia defined as the region southwest of the line Figure 4 Average May-October rainfall over the period as a percentage of the average May-October over the period (IOCI 22) Figure 5 Relationship between MSLP and SWWA Rainfall (IOCI 22). Note the inverted pressure axis Figure 6 Topographic Map of the Darling Escarpment and the Swan Coastal Plain Figure 7 Precipitation Gauging Stations Locations Schematic Figure 8 Precipitation in the GRDE (from west to east) illustrating the enhancement of precipitation around the Darling Escarpment... 3 Figure 9 Characteristic Precipitation of the Swan Coastal Plain, Darling Escarpment and leeside of the Darling Escarpment Figure 1 Seasonal comparison of the rate of Orographic Precipitation along the Darling Escarpment Figure 11 The rate of orographic precipitation in winter in Britain s Lakes District (Malby et al 27) Figure 12 Cross section of mean annual rainfall across Sri Lanka illustrating orographic precipitation and the rain shadow effect (Puvaneswaran & Smithson 1991) Figure 13 The rate of orographic precipitation under the influence of the NE monsoon in Sri Lanka (Puvaneswaran & Smithson 1991)... 4 Figure 14 The rate of orographic precipitation in winter in central Chile, where R w is winter rainfall (Falvey & Garreaud 26) Figure 15 The rate of orographic precipitation in Kenya, New South Wales, and Taiwan (Basist et al 1994) Figure 16 Percentage Reduction in Average Annual Precipitation from (west to east) before the 1975 step change to after Figure 17 The rate of orographic precipitation over time Figure 18 The rate of orographic precipitation in winter over time... 5 Figure 19 Locations of SWWA water supply dams (Water Corporation 28b) Figure 2 October 28 water storage in the main water supply dams of the SWWA (Water Corporation 28a) (vii)

9 LIST OF TABLES Table 1 Precipitation Gauging Station Specifications Table 2 Regression Analysis Results for the rate of precipitation in relation to elevation along the Darling Escarpment Table 3 Period of the Step Change in Annual Precipitation in the SWWA Table 4 Step change Analysis Results (viii)

10 1. INTRODUCTION Orographic precipitation is a central component of the interactions between the land surface and the atmosphere and therefore plays an important role in the earth s climate and hydrological cycle. Orographic precipitation describes the precipitation enhancement generated from interactions between topography and the atmosphere. Orographic precipitation affects the hydrology, vegetation, landscape, geology, climate, agriculture and economy of many regions (Fuhrer and Schar 24). As such, an understanding of the mechanisms involved in the enhancement of precipitation from the influence of topography has important implications for meteorology, weather forecasting and water resources planning and management. Water is an extremely precious resource, particularly in Australia, one of the driest continents in the world. Climate change coupled with Australia s highly variable climate is expected to have significant impacts within Australia; particularly in relation to Western Australia s water resources considering that the south-west of Western Australia (SWWA) has experienced a decline in average annual rainfall since the 197s (IOCI 22). The Darling Escarpment is a dominant topographical feature of the SWWA. However, the influence of the Darling Escarpment on the generation of precipitation is relatively unknown. The main objective of this research was to improve understanding of orographic precipitation along the Darling Escarpment. To achieve this, this research focused on answering the following questions in relation to the Darling Escarpment: What is the rate of precipitation in relation to elevation? Are there any annual and/or seasonal trends in orographic precipitation? What is the influence of climate variability and change in relation to orographic precipitation? To determine what are the mechanisms of orographic precipitation? The results of this study may be of benefit to current and future water resources planning and management in the SWWA, particularly in relation to surface water since the majority of dams and storage reservoirs are located within the Darling Escarpment. 1

11 2. BACKGROUND 2.1. OROGRAPHIC PRECIPITATION AND EFFECT The influence that terrain has over precipitation is widely acknowledged (Smith 1979, Barros & Lettenmaier 1994, Roe 25, Smith 26) however; explanations of the processes that are involved are often misunderstood and oversimplified (Smith 26). There is an inaccurate notion that topography such as mountains can create precipitation all by themselves (Smith 26). Topography has a profound influence on pre-existing atmospheric conditions that modify and amplify precipitation events rather than creating precipitation. Therefore, the focus of study into orographic precipitation should be on meteorological disturbances that temporarily alter the regional environment (Hemond & Fechner-Levy 2). In addition, an understanding of the interactions between all of the physical mechanisms involved in orographic precipitation such as fluid dynamics, micron-scale cloud processes, thermodynamics and larger scale patterns of atmospheric circulation is required for a complete understanding of orographic precipitation (Roe 25). This however, is extremely difficult given the spatial and temporal scales involves. The basis for any investigation into orographic precipitation is the study of moist air flow and an understanding of the nature of the terrain-induced ascent. Orographic precipitation can be described by examining two inherently linked mechanisms: airflow dynamics and cloud microphysics (Barstad et al. 27). Airflow dynamics are responsible for the lifting of air that can generate cloud formations within a moist environment (Barstad et al. 27) and can be influenced by condensation and evaporation, which can also affect the stability of the flow (Roe 25, Barstad et al 27). In addition, the water phase and evolution of hydrometeors and cloud droplets are controlled by the changes in the cloud microphysics (Barstad et al 27). The mechanisms that influence orographic precipitation are still poorly understood due to the complexity of the factors involved, including the interactions between airflow and topography, the complexity of cloud physics and the complex role of latent heating (Jiang 23). These mechanisms and processes are discussed in detail in the following sections. 2

12 Air Saturation and Temperature When considering orographic precipitation it is of fundamental importance that an air parcel has the capacity to carry water vapour (Roe 25). This is particularly important since cloud water on average equates to only 4% of atmospheric moisture (Barry et al. 1998). An air parcel is considered saturated, that is when its relative humidity reaches 1%, if the partial pressure of the water vapour within the air parcel, e, achieves a threshold value, e sat (Barry et al. 1998, Roe 25). The relative humidity of an air parcel is define by Barry et al (1998) as the actual moisture content of air as a percentage of that contained in the same volume of saturated air at the same temperature (Barry et al. 1998). The saturation vapour pressure is a function of temperature and is given by the Clausius-Clapeyron relationship (1) e sat T at 6.112exp (1) b T where e sat is measured in millibars, a = 17.67, b = C and T is the temperature measured in degrees Celsius and is accurate to within.3% when the temperature is between the range of - 35 C T 35 C (Barry et al. 1998, Smith & Barstad 24, Roe 25, Joshi et al 28). Another measure of humidity is the dew-point temperature, which is the temperature at which saturation occurs if the air is sufficiently cooled at a constant pressure and the vapour content remains constant (Barry et al. 1998). In addition, when the dew point temperature and the air temperature are equal then the relative humidity will be 1% (Barry et al. 1998). When a parcel of unsaturated air encounters topography and is forced to ascend, the parcel will expand adiabatically as the air pressure drops and will cool until the water vapour contained reaches saturation (Barry et al. 1998, Roe 25). At this point, the saturation-specific humidity, q sat, or the mass of saturated water vapour per unit mass of the air parcel is described by equation (2) q sat T z sat e T, z.622 (2) p where p (z) is atmospheric pressure and z is height (Wallace & Hobbs 1977, Roe 25). The mass of water vapour per unit volume in a saturated air parcel where the density of air is ρ, is described by equation (3) 3

13 sat sat z q z q exp (3) H m where b H m (4) a where H m is the characteristic e-folding scale height for atmospheric moisture (Roe 25) and is the temperature lapse rate, which comes from the approximation that the atmospheric temperature varies linearly with height given by z (5) where z is height and T is the temperature at z =. Vapour pressure within the atmosphere will rarely exceed the saturation value of an air parcel by more than 1% (Houze 1993). Hence, the rate of condensation of water vapour of an ascending saturated air parcel, with a vertical velocity of w, is considered to have approximately the same rate of change as the saturated moisture content of an air parcel (Houze 1993 and Roe 25). This process is described by equation (6) sat sat sat q q dz q d C w (6) dt dz dt dz where C is condensation (Roe 25). It is clear from equations (1) to (6) that the temperature and elevation of a parcel of air can influence the amount of water vapour within a parcel of air (Barry et al 1998, Smith & Barstad 24, Roe 25) which in turn influences the amount of precipitation that can be generated. Thus, it is evident that orography influences condensation by affecting the airflow within the atmosphere, where the airflow response may be either stable or unstable depending on convection (Barry et al 1998, Smith & Barstad 24, Roe 25)). 4

14 Adiabatic Lapse Rate and Atmospheric Stability In order to understand how precipitation occurs, the mechanisms that cause air to rise (Barry et al. 1998) and the movement of air within the atmosphere that causes atmospheric stability or instability must be considered (Hemond & Fechner-Levy 2). In the atmosphere, the vertical temperature profile may either enhance or suppress the vertical mixing of air. Due to the compressibility of air, neutral stability occurs in the atmosphere when the actual vertical temperature gradient or the actual lapse rate is equal to the adiabatic lapse rate (Hemond & Fechner-Levy 2). Such changes in temperature that involve no addition or subtraction of heat are referred to as adiabatic (Barry et al. 1998). The adiabatic lapse rate is define by Hemond & Fechner-Levy (2) as the rate at which the temperature of an air parcel changes in response to the compression or expansion associated with a change in height, under the assumption that the process is adiabatic; that is, there is no energy exchange occurring between the given air parcel and its surroundings. There are three main types of lapse rates, environmental (or static) lapse rate, dry adiabatic lapse rate and saturated adiabatic lapse rate (Barry et al. 1998). The environmental lapse rate refers to the actual temperature decrease with height on any given occasion; as such, it is not actually adiabatic (Barry et al. 1998). The dry adiabatic lapse rate is the rate at which a rising parcel of unsaturated (dry) air changes temperature as it increases with height (Barry et al 1998, Hemond & Fechner-Levy 2). Under these conditions as an air parcel rises it expands, losing internal energy, resulting in a decrease in temperature (Krolak 21). The reverse occurs when an air parcel decreases in height. This situation is typically stable since when an air parcel rises it becomes warmer and less dense than its surroundings, which are typically cooler and denser, so the air parcel will fall back down, as shown in Figure 1 (Hemond & Fechner-Levy 2). 5

15 Figure 1 Adiabatic Lapse Rate in the context of orography (Krolak 21) When an air parcel is saturated with water vapour and rises, undergoing adiabatic cooling, its temperature will decrease, causing its relative humidity to exceed 1% resulting in the condensation of water (Hemond & Fechner-Levy 2). As water condenses heat is released which diminishes the effect of adiabatic cooling, resulting in the air parcel being warmer than an unsaturated air parcel under the same conditions (Barry et al 1998, Hemond & Fechner-Levy 2). The movement of air (up and down) will affect its stability. For instance, if an air parcel is moved upwards, it will become denser than the surrounding air and will fall to its original position (Krolak 21); this process is considered stable. However, if the air becomes less dense than the surrounding air as it rises it will continue to rise and is considered unstable (Barry et al. 1998, Krolak 21). Similarly, if the air has the same density as its surroundings it will have no tendency to rise or fall, and is considered neutrally stable (Barry et al. 1998, Krolak 21). The lapse rate of an air parcel is used to determine the stability of the air parcel (Barry et al 1998). If the actual lapse rate of unsaturated air is less than the dry adiabatic lapse rate of the same air column it is considered to be stable (Krolak 21). The characteristics of unstable air is that is has a tendency to continue moving (Barry et al. 1998), therefore if the actual lapse rate of air is greater than the dry adiabatic lapse rate it is considered to be in a state of absolute instability (this applies whether the air is saturated or unsaturated) (Krolak 21). Although, typically air is considered unstable if the actual lapse rate of air is greater than the saturated adiabatic lapse rate (Krolak 21). 6

16 The orographic lifting of air will typically involves both dry and saturated adiabatic lapse rate (Hemond & Fechner-Levy 2). As an air parcel is lifted it follows the dry adiabat until water vapour condenses where it will continue to rise, now following the wet adiabat, assuming that supercooling does not occur (Barry et al. 1998). Supercooling describes a non-equilibrium situation where air cools below the dew point without condensation (Hemond & Fechner-Levy 2). This process is referred to as conditional instability, where the environmental lapse rate is between the dry and saturated adiabatic lapse rates, and is a very common atmospheric state (Barry et al. 1998) Airflow Dynamics An understanding of the basic thermodynamics of orographic precipitation has been firmly established by Wallace and Hobbs (1977), Smith (1979), Jiang (23), Smith et al. (23), Roe (25) and Smith (26). Additionally, airflow cannot be considered independently of precipitation (Roe 25). The dynamic response of airflow when in the presence of orography plays a crucial role in the mechanisms that control hydrometeor growth; such as advection, condensation, evaporation and precipitation (Roe 25). Typically, more than one mechanism will be involved in regards to the generation of precipitation and change in airflow (Roe 25). Moreover, the relationship between orographic precipitation and upslope flow, on a case-by-case basis is controlled by several factors, including ambient thermodynamic stratification, the dynamics of the atmospheric airflow over or around an orographic barrier, moisture availability, latent heat release and the timescales involved and efficiency of hydrometeor generation (Smith 1979, Neiman et al 21). However, it is useful to consider these mechanisms independently in order to understand all of the processes involved (Roe 25). The atmosphere is very sensitive to any vertical motion according to Smith (1979) for two key reasons. Firstly, strong stable stratification of the atmosphere acts as a resistance to vertical displacement where buoyancy forces will try to return the vertically displaced air parcels, however this restoration can often generate strong winds (Smith 1979). Secondly, the lower atmosphere is typically high in water vapour and even the slightest adiabatic ascents will lead to saturation of the air, which in turn will cause condensation and possible precipitation (Smith 1979). Higher topography is therefore likely to have a greater influence on the atmosphere. 7

17 To summarise, air is forced to rise over sloping terrain which results in the expansion and cooling of air parcels, as the air temperature drops the relative humidity increases which eventually leads to the condensation of water vapour (Smith 26). Condensation will usually occur on dust particles in the atmosphere to form small cloud droplets; these cloud droplets then transform into larger hydrometeors such as rain or snow and will fall under the influence of gravity which forms precipitation (Wallace & Hobbs 1977, Smith 26), as further discussed in Section In a stably stratified atmosphere, an air parcel will always seek a neutral level of buoyancy (Roe 25). As such, the forced vertical displacement of air can create restoring forces that can result in the generation of atmospheric gravity waves (Smith 1979, Roe 25). The lifting of low-level air by orographic barriers and from the excitation of gravity waves will enhance the vertical motion of air typically leading to the formation of clouds and precipitation (Clark & Hall 1994). According to Clark & Hall (1994), gravity waves of buoyancy waves can often penetrate into deep layers of the atmosphere, which can significantly influence the intensity, location, and the microphysics of precipitation. 8

18 Figure 2 Orographic Precipitation Mechanisms Illustration (a) Stable Upslope Ascent, (b) Partial blocking of the approaching air mass, (c) Down-valley flow induced by evaporative cooling, (d) Lee-side Convergence, (e) Convection triggered by solar heating, (f) Convection due to mechanical lifting above the level of free convection, and (g) Seeder-feeder Mechanism (Roe 25) The most obvious mechanism of orographic precipitation is stable upslope ascent (Figure 2a). Stable upslope ascent involves the forced mechanical lifting of air on the windward slope, resulting in condensation and precipitation (Barry et al 1998, Roe 25). Airflow can often be blocked or unable to ascend over topography if the flow is not strong enough or if the atmosphere is too stable (Roe 25). In this case, airflow is diverted around the topography or it may stagnate, as shown by Figure 2b. Orographic precipitation is enhanced when blocked air downwind of topography increase lifting (Barry et al 1998, Jiang 23, Medina & Houze 23, Roe 25). Figure 2c illustrates a similar effect, where the incoming air is forced to rise over blocked air. However, in this case, blocked air is caused by melting and the evaporation of precipitation that cools the surrounding air mass as it falls resulting in down-valley airflow (Roe 25). A lesser- 9

19 known mechanism is leeside convergence, as shown by Figure 2d, where atmospheric flow splits around the topography and the air converges and ascends on the leeside (Roe 25). Another mechanism of orographic precipitation is from solar heating of the slopes leading to convection, as illustrated by Figure 2e (Barry et al 1998, Roe 25). The triggering of unstable convection is another mechanism of orography precipitation (Banta 199, Roe 25), as shown by Figure 2f. Unstable convection occurs when orography lifts air above the level of its free convection; that is, the level where the air becomes less dense than the surrounding air (Roe 25). Unstable ascent can produce a greater amount of super-cooled cloud water droplets, as well as enhancing the local condensation rate, which increases the efficiency in terms of fall-out of precipitation (Smith 1979, Roe 25). The final mechanism is seeder-feeder, as illustrated by Figure 2g (Roe 25). In this case, there is typically a large-scale precipitating cloud, called the seeder, which is at a high level undisturbed by the topography, and a lower level cloud forced to ascent over the topography, referred to as the feeder (Smith 1979, Roe 25). In this scenario precipitation, falling from the seeder cloud accretes additional moisture into the feeder cloud as it falls through, either by coalescence or riming, which enhances the amount of precipitation over the topography (Barry et al 1998, Roe 25). The mechanism was originally used to explain the additional precipitation observed over small hills that were considered unable to generate precipitation by themselves, as the airflow is too quick for hydrometeors to form through coalescence and condensation (Roe 25). However, an issue with this mechanism is that this same scenario can arise from upslope accent where cloud water is washed-out by falling hydrometeors. Additionally, according to Barry et al. (1998), orography can influence precipitation by retarding the movement of cyclonic systems and increasing the amount of frontal precipitation. The common link between all of these mechanisms is the generation of condensation; however, how the condensation reaches the ground depends on the dominant precipitation conversion mechanisms at the time (Roe 25). As such, when considering orographic precipitation, precipitation conversion mechanisms need to be considered Cloud Microphysics The effects of microphysical processes are of significant importance in determining how condensation is dispersed over topography as precipitation (Smith 26). This is despite the fact 1

20 that these processes usually operate at scales which are several orders of magnitude less than those typically of interest when investigating orographic precipitation (Roe 25, Smith 26). There are numerous ways in which cloud water and cloud ice particles can interact, combine and grow to form precipitation (Roe 25) and the behaviour of these different categories of condensed water is another key element in the understanding of orographic precipitation. In addition, the transfer of latent heat with the surrounding atmosphere from the exchange of cloud particle growth and evaporation can have a significant impact on airflow (Wallace & Hobbs 1977, Houze 1993, Jiang 23 and Roe 25). Evaporation provides the initial moisture input into the atmosphere (Barry et al. 1998). When the moisture in air exceeds its ability to carry water as a vapour, clouds will form (Roe 25). Since supersaturation levels of the vapour phase are typically required for homogeneous nucleation almost all condensation is considered to occur through heterogeneous nucleation (Jiang & Smith 23, Roe 25). Heterogeneous nucleation describes the process where condensation occurs on air particles which are called cloud condensation nuclei (CCN) (Roe 25). The concentration of cloud condensation nuclei and ice nuclei (IN) can affect the formation and evolution of cloud particles (Jiang & Smith 23). Fewer nuclei (CCN and IN) results in condensation being taken up by fewer cloud particles which may grow more rapidly and potentially have faster fall speeds. Whereas more nuclei (CNN and IN) may lead to smaller and more numerous cloud particles depending on the condensation rate (Roe 25). In cold clouds, where temperatures are less than C, water droplets that are supercooled can coexist, in theory, with ice particles to temperatures of approximately - 4 C (Roe 25). The main mechanism involved in cold clouds is riming, where supercooled droplets freeze directly onto ice particles (Barry et al 1998, Roe 25). Ice particles can also grow through the diffusion of water vapour, referred to as deposition, and through the collision of ice particles, referred to as aggregation (Roe 25). In warm clouds, where temperatures are above C, cloud droplets can grow via collisions between water droplets, referred to as coalescence, or by the diffusion of water vapour onto droplets, referred to as condensation (Barry et al 1998, Roe 25). As expected, these processes are highly dependent on the ambient in-cloud conditions (Roe 25) such as temperature, turbulence and aerosol content (Smith 26). Cloud droplets are typically small, with radii of the order of 2 μm, and are abundant with concentrations between 1 and 1 cm -3, and will follow the motion of air (Barstad et al 27). Raindrops and drizzle form as a result of coalescence and collisions between cloud 11

21 droplets with diameters between 1 μm to millimeters that have appreciable terminal velocities (Barstad et al 27). Once formed, a hydrometeor will fall to the ground under the influence of gravity at some terminal velocity (V t ) (typically rain and snow have a terminal velocity of 6 m/s and 2 m/s respectively) (Smith 26). The fallout time τ f, that is the time taken for the hydrometeor to reach the ground, is dependent on height and can be expressed as H f (7) V t where H is the height from which the hydrometeor has fallen (Smith 26). These fallout and conversion time scales are of significant importance since the lifting of air caused by topography is temporary (Smith 26). As such any condensed water that has not been converted into precipitation once the air has past the peak will begin to descend and the remaining water may be subject to evaporation (Smith 26). The rainshadow effect describes the sharp decrease in precipitation experienced on the leeside of a mountain range (Roe 25); however further understanding and discussion of the mechanisms responsible for this effect are beyond the scope of this research. The transformations of cloud droplets can be represented by the mixing ratios for cloud water, q c, water vapour, q v, and rain water, q r using the following conservation laws (Grabowski & Smolarkiewicz 22, Barstad et al 27) q t v U qv Cd E p (8a) q t c U qc Cd Ap C p (8b) q t r U q r qrvt z A p C p E p (8c) where U is the air velocity. The terms on the left hand side of the equations (8a-8c) denote advection terms and local tendency for each of the water species. The right hand side of the equations are the source and sink terms where; C d is the condensation rate of the source water cloud; A p is the autoconversion rate of cloud water into rain (i.e. the initial source of rain); C p is 12

22 the accretion rate of cloud water by rain; and E p is the evaporation rate of rain (i.e. when rain falls into subsaturated air) (Grabowski & Smolarkiewicz 22, Barstad et al 27)). There are several nonlinear processes occurring within clouds such as the collisions and coalescence of liquid water, the riming of snowflakes by supercooled droplets and the accretion of small ice crystals by larger ice crystals (Smith 26). Since these three processes involve the collision between two or more types of particles, the rate of conversion of the smaller particles into the larger particles is proportional to the product of the two original concentrations (Jiang & Smith 23, Smith 26). Condensation is the direct cause of all of the different forms of precipitation and is always associated with a change in air volume, humidity, pressure or temperature (Barry et al. 1998, Roe 25). According to Barry et al (1998), condensation will occur as a result of one of the following scenarios. Firstly, if the air temperature is reduced but the air volume remains constant so that the air is cooled to dew point. Secondly, if the volume of air is increased without the addition of heat, cooling will occur since the adiabatic expansion of air will cause the energy to be consumed. Thirdly, a change in both air volume and air temperature will reduce the moisture holding capacity below the existing moisture content. Lastly, by evaporation adding moisture to the air. Precipitation refers to any liquid or frozen form of water although typically only rain and snow are considered to make significant contributions to totals of precipitation (Barry et al. 1998) Knowledge Gaps Over the past three decades extensive research has been conducted in Europe and America on orographic precipitation (Barstad & Smith 25). This has confirmed the influence of orographic precipitation within relatively steep and high mountainous terrain such as the Alps, the Rocky Mountains, the Appalachian Mountains and the Andes (Ascencio et al 23, Bosquet & Smull 23, Doyle & Smith 23, Medina & Houze 23, Rotunno & Ferretti 23, Smith 23, Smith et al 23, Falvey & Garreaud 27). The elevation along the Darling Escarpment is, however, significantly lower than these regions with an average elevation of between 2-3m. Precipitation enhancement has been observed over similarly low terrain such as in the lowlands of Norway, central Pennsylvania, southern New England and the south of Wales, with scales of 13

23 less than 1 m and reaching up to 5m (Bergeron 1961, Passarelli & Boehme 1983, Roe 2, Colle & Yuter 27). It is still unknown, however, whether the same mechanisms involved in orographic precipitation at larger scales are applicable to low terrain (Kitchen & Blackall 1992, Cosma et al 22, Colle & Yuter 27), particularly in relation to the degree of influence of small scale terrain on the atmosphere (Cosma et al 22). Additionally, it is yet to be known whether one of these mechanisms will emerge as dominant at smaller spatial scales (Roe 25). Research into orographic precipitation within Australia has been limited to the eastern seaboard Alps in New South Wales which have elevations of 2m (Basist et al 1994). The influence of the Darling Escarpment on precipitation has been briefly mentioned by Wright (1974) but not in terms of the mechanisms involved or the degree of this influence. 14

24 2.2. SOUTH WEST OF WESTERN AUSTRALIA The SWWA is defined as the region bound by the south of latitude 3 S and west of longitude 12 E shown by Figure 3 (Samuel et al 26, Bates et al 28). The Darling Escarpment is a dominant topographical feature of the SWWA. Figure 3 Southwest region of Western Australia defined as the region southwest of the line General Climate This region experiences a Mediterranean climate that is characterised by cool wet winters and hot dry summers (Charles et al 27). The majority of the annual rainfall, approximately 8%, falls in the winter months (May to October) of the year (Wright 1974, IOCI 22, Charles et al 15

25 24, Hope 26, Bates et el 28). Precipitation tends to be lower in late winter; that is from August to October, than in the beginning of winter, May to July (IOCI 22). The area also exhibits a marked gradient in winter rainfall decreasing from west to east, and a temperature gradient decreasing from north to south (IOCI 22). Typically, the western boundary of a landmass receives a low amount of rainfall in part due to the ocean currents flowing along the eastern boundary towards the equator reducing the availability of atmospheric moisture (IOCI 22, Bates et al 28). However, off the coast of the SWWA there is a lack of a cool off-shore current and a high availability of atmospheric moisture, which coupled with the force of the predominant south west winter winds provides the region with relatively abundant rainfall, approximately double that of any similarly exposed localities of other continents (Gentilli 1972, IOCI 22, Bates et el 28). An annual rainfall cycle exists where low rainfall is experienced in summer and heavy rainfall is experienced in winter because of the annual cycle of the planetary winds (Gentilli 1972, IOCI 22). Low rainfall in the summer months is caused by the subtropical belt of high pressure that extends across the SWWA where the southern most extension is reached in late summer (Gentilli 1972, Bates et al 28). The subtropical high pressure belt moves gradually north during autumn until it lies almost completely outside of the SWWA region during winter, only to gradually return south during spring (Gentilli 1972). The subtropical high pressure belt is made up of successive anticyclonic cells that are located over the Australian continent and the Indian Ocean (IOCI 22). Midlatitude cyclonic frontal systems generate the majority of winter precipitation (Gentilli 1972, Wright 1974, Charles et al 24). Winter rainfall, according to Wright (1974), is mainly associated with moist unstable westerly winds that have progressive troughs within the westerly airflow. This unstable westerly airflow generates heavy precipitation that is enhanced by convergence induced by the friction on the air as it reaches the coast and the Darling Escarpment (Wright 1974). A cold front describes the movement of a cooler air mass towards a warmer air mass that will displace a warmer air mass upwards (Krolak 21). A key feature of this frontal precipitation, over the SWWA, is the rapid increase in intensity and northward spread at the beginning of winter and the slower retreat and decrease at the end of winter (Charles et al 24). Additionally, northwest cloudbands that interact with these approaching cold fronts associated with mid- 16

26 latitude synoptic depressions that form pre-frontal rainbands contribute significantly to the total precipitation associated with a frontal passage (IOCI 22). Indeed, Wright (1974) describes a second fundamental winter rainfall type that affect the SWWA as generally continuous rain associated with an upper tropospheric jet stream from the northwest and widespread ascent in the mid troposphere, usually associated with a surface wind between west and north. This situation is commonly associated with the approach of a cold front, where relatively warm, humid air is drawn southwards from the Indian Ocean. The quantity of precipitation generated by this type is relatively consistent from north to south, experiences little coastal enhancement and decreases as the airstream progresses inland (IOCI 22). Whilst the former type, however, is associated with convection in a moist unstable airstream, usually during and after the passage of a cold front and that is probably the more frequent with west to southwest winds, (Wright 1974) that is generally enhanced by coastal friction and orography (IOCI 22). Bates et al (21) identified that there are six key rainfall States observed in the SWWA. Of these states the two most important were State 3 and 5; where State 3 produced widespread rainfall over the region but concentrated on the west coast and escarpment whilst state 5 generated dry conditions everywhere over the region (Bates et al 21) Climate Variability and Change Climate variability typically refers to any variation in the mean, extremes, standard deviation and other statistics of climate features at larger scales than those experienced for a single event (Bureau of Meteorology 27). Additionally, climate variability can also refer to cyclic climatic events that do not alter the long-term climate statistics (IPCC 27). Climate change, however, is typically associated with anthropogenically forced processes (Bureau of Meteorology 27, IPCC 27). Within this document climate variability refers to natural variability and climate change refers to anthropogenic forcing (global warming). Over much of the SWWA, winter rainfall has decreased substantially since the mid-2 th century (IOCI 22, Charles et al 24, Bates et al 28). Large areas of the SWWA experienced a sharp and sudden decrease in winter rainfall around the mid-197s by up to 15 2% (IOCI 22, Charles et al 24), better illustrated by Figure 4. Another major feature of this step change is the absence of very high rainfall years that were a relatively common feature of throughout the 2 th 17

27 century (Bates et al 28). This decline was not gradual, but more of a switching into an alternative rainfall regime (Charles et al 24). Figure 4 Average May-October rainfall over the period as a percentage of the average May- October over the period (IOCI 22) Changes in the region s winter precipitation are related to variations in atmospheric circulation (IOCI 22). Its has been found in Perth that variations in year to year precipitation are closely related to variations in Perth s Mean Sea Level Pressure (MSLP) (Smith et al 2). Typically, if the MSLP increases then precipitation usually decreases as shown in Figure 5. As such, years of high precipitation correspond with low pressures over the greater Australian region with the strongest correlations (between -.8 to -1.) over the SWWA (IOCI 22). 18

28 Figure 5 Relationship between MSLP and SWWA Rainfall (IOCI 22). Note the inverted pressure axis. The El Niño Southern Oscillation influences pressure across the Australiasian region thus relates to both the climate and air temperature of the SWWA (Jones & Trewin 2). The relationship between the El Nino Southern Oscillation and Australian rainfall has been found by Nicholls et al (1996) to change on multi decadal time scales. However, the El Nino Southern Oscillation relationship with the SWWA has changed since the mid-195s (Nicholls et al 1999) and rainfall is now lower than in the past for any value of the Southern Oscillation Index (SOI). The SOI index refers to the standardised pressure difference between Darwin and Tahiti and is commonly used as an index of the El Nino South Oscillation (Nicholls et al 1996). This indicates that other factors may be involved in the change in rainfall. It is thought that changes in sea surface temperature in the Indian Ocean are a better indicator to variations in SWWA rainfall (IOCI 22). Smith et al (2) found that southern Indian Ocean sea surface temperatures appear to be correlated with the south west rainfall where dry conditions are typically associated with warm sea surface temperatures. However, this relationship mostly reflects the fact that the Indian Ocean sea surface temperature has warmed over the last few decades as the rainfall of the southwest has decreased; additionally this correlation has weakened in recent years (Smith et al 2). As such, it is thought that this correlation simply reflects the current trends experienced by sea surface temperature and rainfall 19

29 rather than a link to the cause of these trends (Smith et al 2). Additionally, the correlations between rainfall and sea surface temperature were not consistent over time or seasonally; however this may indicate a more complex relationship than the assumed linear trend (IOCI 22). Furthermore, Bates et al (21) found that the occurrence of weather state 3 has decreased since the mid-197s, as a result of a decrease in the moisture content of the lower atmosphere and the sudden increase in mean atmospheric pressure; whilst the occurrence of weather state 5 has increased. The similar sudden drop in both atmospheric pressure and the SOI in conjunction with a decline in the number and intensity of synoptic depressions affecting the south west (Simmonds & Keay 2) indicates that large scale processes may be responsible for this decline (IOCI 22). Additionally, changes in the frequency and intensity of cold fronts and the observed decrease in the number and intensity of mid-latitude depressions alter the frequency of weather patterns (Simmonds & Keay 2, IOCI 22). Therefore, the IOCI (22) has suggested that this sudden large-scale change in the atmospheric circulation of the Indian Ocean sector can be regarded as a partial cause of the decrease in rainfall in the SWWA. Bates et al (28) has suggested that the recent decline in the SWWA rainfall is a returning of rainfall, levels to those observed at the end of the 19 th century and that the decline is simply part of a naturally variable cycle. However, IPPC found that there is strong evidence that the majority of global warming observed over the past 5 years is attributable to human activities (Houghton et al 21). Surface temperatures have risen globally by.76 o C in the last 1 years, compared to the 185 to 1899 mean (Bureau of Meteorology 27, IPCC 27). The majority of this temperature increase has occurred since the 197s (Bureau of Meteorology 27) where annual mean temperatures have increased at a rate of.15 C per decade in all seasons except summer (Bates et al 28). Typically, both global day and night-time temperatures have increased since the mid 2 th century, which is reflected in temperature trends of the SWWA (IOCI 22, Bureau of Meteorology 27). Furthermore, the most significant temperature increase has been the minimum and maximum daily temperature through autumn and winter (IOCI 22). The IPCC have attributed this global warming trend to the enhanced greenhouse effect (Houghton et al 2

30 21). Regardless of whether greenhouse gas emissions continue to increase it is projected that SWWA will continue to become drier and warmer through this century (IOCI 22). Temperature has considerable influence on the hydrologic cycle, since the atmosphere s water holding capacity will increase as surface temperature rise increasing evaporation (Bureau of Meteorology 27). Change in evaporative balance may have flow on effects to precipitation, with some areas receiving more rainfall than others due to the increased atmospheric water content (Hope & Foster 25). Ongoing work to better understand the mechanisms driving this step decrease in rainfall have included the analysis of climate model simulations (Bates et al 28). Hope (26) and Bates et al (28) found that when coupled models are forced by varying greenhouse gas concentrations (such as those observed over the 2 th century) that they respond in a similar manner that leads to the decrease rainfall over the SWWA. Additionally, Hope (26) found that analysis of 2th century rainfall using eight different climate models resulted in every model experiencing drier conditions over the SWWA after However, in most cases the simulated decline in rainfall only equated to half of the observed decline (Bates et al 28). Therefore, the observed decline could simply reflect a naturally occurring dry period with a small contribution of the decline attributable to anthropogenic forcing (Bates et al 28). This has lead the IOCI (22) to conclude that the rainfall decline in the SWWA is most likely due to both the enhanced greenhouse effect and natural variability. In the SWWA, average annual streamflow typically ranges from 3 to 2% of the total rainfall (Bureau of Meteorology 27, Hope & Foster 27). A 15 2% decline in annual rainfall, translates to a 4% decrease in annual stream flow and runoff into dams, which is disproportionately large (Charles et al 24, Hope & Foster 25) compared to historical percentage contributions. This has serious implications for the water supply to SWWA s catchments as well as the reliability of water resources in the region (Charles et al 24). This is especially significant for natural ecosystems, agriculture and urban water supply considering that historically the SWWA was considered one of the most reliable in Australia (IOCI 22, Hope & Foster 27) 21

31 3. APPROACH 3.1. STUDY SITE DESCRIPTION The study region focussed on in this study, shown by Figure 6, includes the Swan Coastal Plain, Darling Escarpment and the leeside of the Darling Escarpment. From this point onwards the study region will be referred to as the greater region of the Darling Escarpment (GRDE). Figure 6 Topographic Map of the Darling Escarpment and the Swan Coastal Plain 22

32 The Darling Escarpment runs roughly parallel to the southwest coast of Western Australia and extends 322 km from Moore River in the north to Brunswick Junction in the south (Macquarie University 24). Figure 6 shows the topography of the GRDE, where the Darling Escarpment, shown in shades of green and yellow, rises sharply from the Swan Coastal Plain, which is shown in shades of blue. The Darling Escarpment has an average elevation of between 2 to 3 m with its highest peak near Mt Cooke at 582 m (Macquarie University 24) GAUGING STATION DETAILS Precipitation data was obtained from the National Climate Centre at the Australian Government Bureau of Meteorology (Bureau of Meteorology 28) from twenty-one gauging stations located in the GRDE. The locations and details of the rain gauging stations are shown in Figure 7 and Table 1 respectively. Figure 7 Precipitation Gauging Stations Locations Schematic 23

33 Table 1 Precipitation Gauging Station Specifications No. Station Latitude Longitude Operational Elevation Site Name Number (South) (East) Period (m) 1 97 Chidlow Gingin Perth Airport Jarrahdale Mundaring Weir Perth Regional Office Serpentine , Wungong Dam Kalamunda , Kwinana BP Refinery Gosnells City Karnett Subiaco Wastewater Treatment Plant Perth Metropolitan Bickley Dwellingup Mandurah Park Northam Bakers Hill Beverley Pingelly These gauging stations were selected based on their proximity to the Darling Escarpment, quality and consistency of data, operational time period, and elevation. It was assumed from the accompanying metadata that accompanied these records were satisfactorily homogenous and consistent for analysis. This is in accordance with the data selection procedure used by Srikanthan & Stewart (1991), where homogeneity refers to the invariability of the site and its surrounding environmental conditions, whilst consistency refers to the measuring technique, sampling interval and the method of data processing. Additionally, the study site was restricted to include only gauging stations that were located within the 2 by 2 km study area. This was done in accordance with Basist et al (1994) to minimise the variation of macroclimatic conditions throughout the study region. Based on these criteria precipitation data was obtained and analysed over the period from 19 until 27. The Perth Regional Office gauging station closed in 1992 and was moved to a new location in 1993 and from that point onwards was called the Perth Metropolitan gauging station (Bureau of 24

34 Meteorology 28). For the purpose of this research, the Perth Regional Office and Perth Metropolitan gauging station are considered to be one gauging station that will hereafter be referred to as the Perth gauging station with an averaged elevation of 22 m. In addition, it should be noted that all precipitation records are only an estimate of precipitation. This is because precipitation data from gauging stations contain a number of inherent errors due to evaporation, gauge height, site location, splash in and small and large-scale turbulence in the airflow (Barry et al 1998) PRECIPITATION ANALYSIS Precipitation data was analysed temporally and spatially to identity annual and seasonal trends. Where precipitation data was incomplete from the records the missing data was interpolated using precipitation data from nearby stations in accordance with the methodology used by the Bureau of Meteorology (28) and Srikanthan and Stewart (1991) Orographic Effect To identify whether the orographic effect was occurring along the Darling Escarpment precipitation data was compared between all of the gauging stations obtained based on their elevation and proximity to the Darling Escarpment. Precipitation data from all gauging stations was also averaged over the analysis period and compared Regression Analysis between Precipitation and Elevation The rate of orographic precipitation along the Darling Escarpment was determined by identifying the relationship between elevation and precipitation through regression analysis. As such, a linear regression of precipitation against elevation was determined using following the equations (Snedecor & Cochran 1989, Basist et al 1994) Y 1 b b X (9) where X is elevation (m), Y is precipitation (mm), b is given by equation (1), and b 1 is given by equation (11) 25

35 b 1 Y b X (1) and b 1 n xi yi i 1 (11) n x i 1 2 i where n is the total number of data. In this case, the rate of orographic precipitation is the slope of the linear regression (b 1 ) (Basist et al 1994). Precipitation data from all gauging stations on the leeside and leeside slopes of the Darling Escarpment were not used to determine the rate of orographic precipitation. This was because these gauging stations would not be influenced by the orographic effect, rather the rain shadow effect. A rate of orographic precipitation was determined for each year over the analysis period however due to the different operational periods of each gauging station it was not always possible to include the same stations for each calculation. Additionally, some gauging stations at lower elevations were combined for years when there was a high proportion of gauging stations on the Swan Coastal Plain in the dataset. This was done to prevent the lower elevations from skewing the relationship between elevation and precipitation. This was applied to six gauging stations in total, which were grouped into pairs based on there proximity to one another. These pairs were then combined by averaging the two stations to give one value. This averaging was considered acceptable since the gauging stations were no more than 3 kilometres apart, and no more than 9 m difference in elevation, which was not considered significantly different. In order to determine the significance of the relationship between precipitation and elevation, for each linear regression the correlation coefficient (also called Pearson s product moment of correlation) was calculated. The correlation coefficient is a measure of the significance between two variables and is given by equation (12) (Snedecor & Cochran 1989) 26

36 r n i 1 n i 1 x x y y n 2 2 x x y y i i i i 1 i (12) The correlation coefficient can be any value between negative one and one (Snedecor & Cochran 1989). Taking the square of the correlation coefficient is a statistical measure of the goodness of fit of the real data points to the linear regression. These values can be between zero and one, where a value of one indicates that the data perfectly fits the linear regression (Snedecor & Cochran 1989). Typically, higher r 2 values have stronger correlation. For this study, r 2 values are considered statistically appropriate since a linear relationship was anticipated between precipitation and elevation Annual Time-series Analysis Time-series graphs of annual precipitation data at each gauging station were utilised to visually establish any apparent trends. In accordance with Srikanthan & Stewart (1991), eleven year moving averages were used to smooth data and identify variations or trends in the data. Several studies into the climate of the SWWA have suggested that there has been a stepchange downward in annual precipitation in the mid-197s (IOCI 22, Charles et al 24, Bates et al 28). Hence, gauging stations with longer time-series from the GRDE were analysis to determine where this step change occurred. This was done by identifying where the eleven year moving average appeared to deviate significantly from the established trend into another. From this each time-series was then separated into two epochs, before and after the step change. Comparison of the mean and standard deviation were then carried out for each epoch. This analysis focussed primarily on temporal trends between gauging stations on the Darling Escarpment compared to those on the Swan Coastal Plain and on the leeside of the Darling Escarpment. However, this analysis also investigated any spatial variability within the GRDE and at varying elevations. 27

37 Spectral Analysis Spectral analysis of the data involved the use of Fourier analysis to identify key frequencies and periods of the precipitation data. Fourier analysis is particularly useful in frequency analysis as it provides an insight into the periodicities of data by filtering out the noise and representing the data as a linear combination of sinusoidal components with different frequencies (Korner 1988, Oran Brigham & Brigham 1988). The following equations were used in the Fourier analysis X ( k) N n 1 x( n)exp i2 n 1 k 1 N (13) 1 k N where x( n) 1 N N k 1 2 a( k)cos k 1 n 1 2 k 1 n 1 N b( k)sin N (14) where a( k) real( X ( k)), b( k) imag ( X ( k)) 1 n N where n is the input data and N is the number of input data (Oran Brigham & Brigham 1988, Korner 1988). The mathematical software program Matlab was used to carry out the Fourier analysis. This analysis focussed primarily on identifying key temporal frequencies between gauging stations on the Darling Escarpment compared to those on the Swan Coastal Plain and on the leeside of the Darling Escarpment. Additionally, this analysis also investigated any spatial variability within the GRDE and at varying elevations Seasonal Analysis Precipitation data at each gauging station was analysed seasonally. This was done by assigning a season to each month of the year. Precipitation data from months with the same season were then 28

38 combined to give a total precipitation for each season. December of the previous year, January and February were considered summer months, March, April and May were considered to be Autumn months, June, July and August were considered to be winter months and September, October and November were considered to be spring months. The rate of orographic precipitation was calculated every year for each season using the same technique described in Section These seasonal rates and their corresponding r squared values were then compared to identify if the rate of orographic precipitation is seasonally influenced. Time-series plots of seasonal data were then compared with each other and to annual data to identify the influence of different seasons on the observed annual trends. 29

39 Mean Precipitation (mm) 4. RESULTS AND DISCUSSION 4.1. OROGRAPHIC EFFECT AND PRECIPITATION ALONG THE DARLING ESCARPMENT The analysis reveals that orographic precipitation is occurring along the Darling Escarpment despite the relatively low elevations along the Escarpment. The complete time-series of all of the precipitation data from within the GRDE is provided in Appendix A. Figure 8 and Figure 9 are simplified versions of the precipitation data from Appendix A. Figure 8 illustrate the effect of the Darling Escarpment on precipitation; where shades of green represent lower elevation on the Swan Coastal Plain, shades of blue represent elevations within the Darling Escarpment and shades of orange represent elevations on the leeside of the Darling Escarpment Kwinana (4m) Mandurah (15m) Subiaco Station (2m) Perth (22m) Perth Airport (15m) Gosnells (1m) Gingin (92m) Serpentine (12m) Mundaring (19m) Wungong (2m) Kalamunda (21m) Jarrahdale (24m) Dwellingup (267m) Karnett (286m) Bickley (384m) Chidlow (33m) Bakers Hill (33m) Pingelly (297m) Beverley (19m) Northam (17m) Figure 8 Precipitation in the GRDE (from west to east) illustrating the enhancement of precipitation around the Darling Escarpment The Bureau of Meteorology (28) has identified that the topographic features of the SWWA do exert a significant influence on precipitation. This influence is seen as a rapid increase in precipitation from the Swan Coastal Plain to the top of the Darling Escarpment (IOCI 22, Bureau of Meteorology 28). This is followed by a marked decrease in precipitation on the 3

40 Annual Precipitation (mm) leeside of the Darling Escarpment to the east (IOCI 22, Bureau of Meteorology 28) commonly referred to as the rain shadow effect Year Perth (22m) Jarrahdale (24m) Northam (17m) Figure 9 Characteristic Precipitation of the Swan Coastal Plain, Darling Escarpment and leeside of the Darling Escarpment From these Figures (8 & 9), it appears that precipitation is relatively consistent east west along the Swan Coastal Plain and increases significantly at higher elevations within the Darling Escarpment. In particular, the enhancement of precipitation is characterised by Figure 9, where the Perth time-series represents the Swan Coastal Plain, Jarrahdale represents the Darling Escarpment and Northam represents the leeside of the Darling Escarpment. It is apparent that precipitation is enhanced within the Darling Escarpment and decreases drastically at elevations on the leeside of the Darling Escarpment despite the fact that they are often equivalent to those within the Darling Escarpment and higher than those on the Swan Coastal Plain. 31

41 4.2. RATE OF OROGRAPHIC PRECIPITATION Relationship between Precipitation and Elevation along the Darling Escarpment Regression analysis was carried out for every year of the analysis period to determine the relationship between elevation and precipitation; these results are included in Table 2. Over this period, it was found that the rate of orographic precipitation (precipitation (mm)/elevation (m)) ranged from.45 mm/m to 3.58 mm/m with a mean value of 1.6 mm/m. The standard deviation of the rates over this period was found to be.61 and 96% of the data fell within two standard deviations of the mean. The values of the square of the correlation coefficient (r 2 ) over this period were consistently strong with a mean of.64 and a standard deviation of.16. This indicates that the relationship between precipitation and elevation is significantly strong. 32

42 Table 2 Regression Analysis Results for the rate of precipitation in relation to elevation along the Darling Escarpment Year Linear Regression r 2 Year Linear Regression r 2 Year Linear Regression r 2 19 y = 1.928x y = x y = x y =.6561x y = x y = x y =.748x y = x y = x y = x y = 1.644x y = x y = 3.296x y = x y = 1.598x y = x y = x y = 2.66x y =.926x y = x y = x y = 1.92x y = x y = x y =.5481x y = x y = x y =.5491x y = x y = 2.737x y = x y =.9755x y = x y = x y = x y = 2.331x y = x y = x y = x y = 1.644x y = 2.232x y = x y =.852x y = 1.267x y = x

43 Year Linear Regression r 2 Year Linear Regression r 2 Year Linear Regression r y = x y = x y = 1.146x y = 2.867x y = x y = x y = x y = x y = x y = x y = 2.558x y = 1.478x y = x y = 3.997x y = x y = x y = x y = x y = x y = x y = 1.937x y = x y = x y =.9195x y = x y = x y = x y = x y =.7697x y = x y = 2.567x y = 1.293x y = x y = 1.499x y = 1.461x y = x y = 1.899x y =.9497x y = x y = x y = 1.954x y = 1.777x y = x y = x y =.9477x

44 Year Linear Regression r 2 Year Linear Regression r 2 Year Linear Regression r y = 1.494x y = 1.731x y = 1.571x y = x y = x y = x y = x y = x y = x y = x y =.8157x y = x y =.7152x y =.8321x y = x y = x y =.4542x y = x

45 Rate of Orogrpahic Precipitation (mm/m) Seasonal Variability Regression analysis was also carried out for each season; summer, autumn, winter and summer, of the analysis period to determine the seasonal relationship between elevation and precipitation; these results are included in Appendix B and Figure Year Summer Autumn Winter Spring Figure 1 Seasonal comparison of the rate of Orographic Precipitation along the Darling Escarpment Over the analysis period, the results reveal that the rate of orographic precipitation is strongest in the winter months, where the rate of orographic precipitation in winter was found to range from.1 to 2.41 with a mean of.84. The rate of orographic precipitation in autumn and spring, mean rate of.3 and.4 mm/m respectively, was found to be comparable but both were substantially less than the rate of orographic precipitation in winter. The rate of orographic precipitation in summer, mean rate of.7 mm/m, was found to be predominantly close to zero suggesting that orographic precipitation is non-existent in summer. The square of the correlation coefficients (r 2 ) were calculated for each year and for each season of the analysis period. During winter a mean r 2 value of.57 with a standard deviation of.21 was found, which suggests that there is a strong relationship between precipitation and elevation in winter months. During autumn a mean r 2 value of.42 with a standard deviation of.21 was 36

46 found, which indicates that the relationship between precipitation and elevation in autumn months is not as strong. In spring a mean r 2 value of.61 with a standard deviation of.19 was found, which indicates that a strong relationship exists between precipitation and elevation in spring months. However, in summer a mean r 2 value of.32 with a standard deviation of.22 was found, which indicates that the relationship between precipitation and elevation in summer months is weak. This type of difference in the seasonal rate of orographic precipitation was anticipated since the majority of rainfall (up to 8%) in the SWWA is received in winter (Wright 1974, IOCI 22, Charles et al 24, Hope 26, Bates et el 28). Additionally, it is suggested that autumn and spring precipitation that is linked to weather patterns typically associated with winter account for the observed higher rate of orographic annually than in winter alone. This suggests that orographic precipitation and the mechanisms involved along the Darling Escarpment are influenced by the prevailing weather conditions, predominantly associated with winter precipitation in the SWWA Comparisons with Previous Studies Several studies around the world have investigated the impact of elevation on precipitation in relation to issues such as seasonality, atmospheric circulation patterns, key weather patterns and climate change (Puvaneswaran & Smithson 1991, Basist et al 1994, Malby et al 27). The principal results from several of these studies are presented and discussed in relation to the rate of orographic precipitation along the Darling Escarpment. Malby et al (27) investigated the spatial and temporal variability, specifically in relation to seasonality and climate change, of precipitation in the Lakes District in Britain from 197 to 26. The Lake District region in northwest England contains many of England s highest mountains, such as Scafell Pike (978m), and experiences a wet maritime climate in the west/southwest that becomes drier further inland towards the east/northeast (Barker et al 24, Malby et al 27). A linear regression of mean winter precipitation in relation to elevation was found to be y = 1. x + 32 (Malby et al 27) (Figure 11). 37

47 Figure 11 The rate of orographic precipitation in winter in Britain s Lakes District (Malby et al 27) This relationship between precipitation and elevation is similar to those observed along the Darling Escarpment and certainly fits within both the annual and winter rate of orographic precipitation range. This suggests that similar mechanisms of orographic precipitation are occurring along the Daring Escarpment as in England s Lake District, despite the large difference in the height of the topography. Puvaneswaran & Smithson (1991) examined the relationship between elevation and precipitation under different key circulation types in Sri Lanka. Sri Lanka is an island that experiences a warm tropical climate with relatively high mountains that rise to 2524 m (Puvaneswaran & Smithson 1991). Sri Lanka s precipitation is driven by monsoons, depressions and tropical cyclones where the south-west monsoons and the northeast monsoons are the dominant circulation types (Puvaneswaran & Smithson 1991). 38

48 Figure 12 Cross section of mean annual rainfall across Sri Lanka illustrating orographic precipitation and the rain shadow effect (Puvaneswaran & Smithson 1991) Regression analysis was carried out at the 1 % confidence level and the annual linear relationship was found to be y = 1.86 x + 21 (Figure 12) and the linear relationship under the influence of the northeast monsoon was found to be y =.83 x + 17 (Figure 13) (Puvaneswaran & Smithson 1991). The relationship between precipitation and elevation observed by Puvaneswaran & Smithson (1991) is similar to that found in this study and similarly fits within both the annual and winter rate of orographic precipitation range. This suggests that similar mechanisms of orographic precipitation are occurring in Sri Lanka and along the Daring Escarpment, despite the difference in elevation and climate. 39

49 Figure 13 The rate of orographic precipitation under the influence of the NE monsoon in Sri Lanka (Puvaneswaran & Smithson 1991) Falvey & Garreaud (27) examined the characteristics of precipitation in central Chile particularly focussing on the identification of orographic influences associated with specific meteorological forcing. Central Chile, like the SWWA, experiences a Mediterranean climate and receives over 75% of its annual precipitation in winter (Falvey & Garreaud 27). Additionally, although they reach 65m, the Andes in central Chile run parallel to the western coast of South America with a north-south orientation (Falvey & Garreaud 27). Falvey & Garreaud (27) did not determine a linear regression between precipitation and elevation however, as can be seen from Figure 14 a similar linear relationship to those already presented does exist between precipitation and elevation. 4

50 Figure 14 The rate of orographic precipitation in winter in central Chile, where R w is winter rainfall (Falvey & Garreaud 26) Basist et al (1994) investigated the statistical relationships between topography and the spatial distribution of mean annual precipitation for several distinct mountainous regions using linear bivariate and multivariate analyses. Central Kenya is situated along the equator and the study site was located at the edge of the East Africa Plateau that extends up to the highlands of the Mau Escarpment reaching heights of 52m (Basist et al 1994). Kenya experiences a hot and humid tropical climate and receives the majority of precipitation between March and August (Basist et al 1994). The Australia Alps situated in south eastern New South Wales reaches elevations of 2m and experience a temperate climate where rainfall is most prevalent in winter but is relatively constant all year round (Basist et al 1994). Taiwan is a mountainous island that reaches elevations of 4m near the centre of the island (Basist et al 1994). Taiwan experiences a tropical marine climate that is strongly influenced by monsoons and receives 75 % of its annual precipitation between March and October (Basist et el 1994). 41

51 Figure 15 The rate of orographic precipitation in Kenya, New South Wales, and Taiwan (Basist et al 1994) Using linear bivariate analysis Basist et al (1994) identified that elevation is a significant predictor of precipitation. Basist et al (1994) found that all three of the high-elevation sites of Kenya, New South Wales and Taiwan had approximately the same rate of orographic precipitation, as shown in Figure 15. It is evident from these examples that a linear relationship does exist between precipitation and elevation. Furthermore, it is clear that the rate of orographic precipitation is similar between mountain ranges with high elevations and those with relatively low elevations. Additionally, the rate is similar for different climatic areas and orientations of the topography in relation to the coast. This suggests that a common mechanism of orographic precipitation exists which outweighs the differences between the locations, particularly in relation to elevation. 42

52 4.3. MECHANISMS OF OROGRAPHIC PRECIPITATION ALONG THE DARLING ESCARPMENT There are several mechanisms responsible for generating orographic precipitation, however not all of these are applicable to the enhancement of precipitation observed along the Darling Escarpment. The most straightforward mechanism of orographic precipitation is stable upslope ascent, where in a stable atmosphere air on the windward slope is forced to rise generating condensation, hence precipitation (Barry et al 1998, Roe 25). It is thought that this mechanism is occurring to some extent along the Darling Escarpment. Another mechanism of orographic precipitation is the partial blocking of an approaching airmass, whereby the airflow is blocked or unable to ascend over the topography (Jiang 23). In this case, orographic precipitation is enhanced when blocked air downwind of the topography increases the amount of lifting of an airmass (Medina & Houze 23). This mechanism is typically associated with high mountain ranges that are able to produce this blocking effect (Jiang 23, Medina & Houze 23). Consequently, this mechanism is not applicable for the Darling Escarpment due to the relatively low elevations observed. A similar mechanism is down-valley flow induced by evaporative cooling (Roe 25). In this case, evaporation within an airmass results in precipitation, cooling the surrounding air blocking the airflow resulting in the airmass falling and down-valley airflow (Roe 25). This mechanism is typically associated with cooler climates and high mountain ranges like those found in the North America and Europe (the Alps) (Doyle & Smith 23, Smith et al 23, Roe 25). Therefore, this mechanism is not applicable for the Darling Escarpment due to its relatively low elevation and Mediterranean climate. An alternative mechanism of orographic precipitation is leeside convergence, which occurs when airflow splits around topography where the air then converges on the leeside and ascends (Roe 25). This mechanism typically occurs when the width of the mountain range is small (Barstad et al 27) and hence is not applicable to the Darling Escarpment which is over 3km wide. A further mechanism that produces orographic precipitation is convection triggered by solar heating (Roe 25). This mechanism is typically associated with mountain ranges with a high elevations and a large slope area (Gochis et al 24). As such, the low elevation and relatively small slope of the Darling Escarpment prohibits this mechanism. 43

53 Another mechanism of orographic precipitation is the triggering of unstable convection, which occurs when orography lifts an airmass above the level of free convection (Smith 1979, Roe 25). This mechanism is similar to stable upslope ascent but differs in that the enhancement of precipitation arises from an unstable atmosphere (Fuhrer & Schar 24, Kirshbaum & Durran 24). Classification of what constitutes a stable and unstable atmosphere was discussed in Section Typically, heavy precipitation associated with storms is generated from unstable convection (Neiman et al 22, Jiang 23, Kirshbaum & Durran 24). It is thought that this mechanism is occurring to some extent along the Darling Escarpment. The final mechanism of orographic precipitation is the seeder-feeder mechanism, whereby a large scale precipitating seeder cloud unaffected by topography, whilst the feeder cloud is a low level cloud that is forced to ascend over the topography (Smith 1979, Roe 25). In this mechanism precipitation falls from the seeder cloud into the feeder cloud, which adds additional moisture to the feeder, enhancing precipitation (Barry et al 1998). However, this same scenario can arise from the upslope ascent of an airmass; hence, in regards to the Darling Escarpment the seeder-feeder mechanism and upslope ascent are considered to be the same mechanism. Therefore, the driving mechanisms enhancing the generation of precipitation along the Darling Escarpment are stable upslope ascent and unstable convection. It has been shown for the Darling Escarpment that orographic effect is strongest during the winter months of the year. According to Wright (1974), winter rainfall is mainly associated with moist unstable westerly winds. As such, this would result in unstable atmospheric conditions. This suggests that unstable convection is the dominant mechanisms of orographic precipitation along the Darling Escarpment. To verify this as the dominant mechanism along the Darling Escarpment would require further in depth research into the atmospheric conditions accompanying precipitation events. This would involve the use of ground based or aerial radars that can observe convection, atmospheric wind and temperature profiles, atmospheric pressure, atmospheric moisture content and the generation of precipitation (White et al 23, Roe 25). 44

54 4.4. VARIABILITY AND CHANGE IN PRECIPITATION WITHIN THE GREATER REGION OF THE DARLING ESCARPMENT OVER TIME Annual Variability and Trends Time-series analysis was undertaken at every gauging station using annual precipitation data. Several studies into the climate of the SWWA have suggested that there was a step-change downward in annual precipitation in the mid-197s (IOCI 22, Charles et al 24, Bates et al 28). Eleven year moving averages were determined for all gauging stations (Appendix C) to visually determine when this step change occurred. Comparisons were then made before and after the identified step change for all gauging stations with 7 years or more worth of precipitation data. It was found that of the eleven suitable gauging stations that a step change was observed in ten over the periods shown in Table 3. However, three of the longer series examined experienced different step changes to the others. Table 3 Period of the Step Change in Annual Precipitation in the SWWA Gauging Station Period of the Step Change Mandurah Perth Gingin , Mundaring Weir Wungong Dam Jarrahdale Dwellingup Chidlow Pingelly Northam At the Gingin gauging station an earlier step change was observed from 1934 to Inspection of the time-series of other gauging stations indicates that eight of the eleven experienced a decline in the 11 year moving average around the 193s, particularly in the late 193s. The incomplete data sets of Kalamunda and Serpentine also experienced this decreasing trend. However, this decline and step change is strongest at Gingin whereas in the others it appears to be part of a naturally variable cycle, such as a period of drought. The Gingin gauging station is quite close to the northern boundary of the region defined as the SWWA and as such, it is 45

55 thought that the observed step change is the result of factors influencing the mid-west region of Western Australia. The Wungong Dam gauging station also experiences a step change sooner to the other stations, beginning at 195 to The cause of this difference is unknown and was unexpected based on the results of other gauging stations nearby. Wungong Dam therefore appears to be unique case, as was Beverley gauging station where no trend or step change was observed. Nonetheless, at all of the gauging stations where a step change was observed the step change consistently ended at This finding is consistent with (IOCI 22, Charles et al 24, Bates et al 28). Notably, in general it appears that this step change began one year earlier at gauging stations located along and on the leeside of the Darling Escarpment than at the locations on the Swan Coastal Plain. Each of the gauging stations shown in Table 3 were then broken into two epochs (excluding Gingin), before and after the step change. The averages and standard deviations were then compared for each epoch. It should be noted that the step change years were not included in this analysis. The results of this analysis are shown in Table 4. Gauging Station Average Annual Precipitation Before Step Change (mm) Table 4 Step change Analysis Results Average Annual Precipitation After Step Change (mm) Standard Deviation Before Step Change Standard Deviation After Step Change Mandurah Perth Mundaring Weir Wungong Dam Jarrahdale Dwellingup Chidlow Pingelly Northam Figure 16 shows the percentage reduction in average annual precipitation after the step change for each of the gauging stations in Table 4. Wungong Dam has been excluded due to the 46

56 % Reduction in Average Annual Precipitation After 1974 difference in the period of the step change to the other gauging stations; nevertheless, Wungong Dam experienced a 26 % reduction in average annual precipitation from 1975 onwards Gauging Station Mandurah Perth Mundaring Jarrahdale Dwellingup Chidlow Pingelly Northam Figure 16 Percentage Reduction in Average Annual Precipitation from (west to east) before the 1975 step change to after It is evident from Figure 16 that the reduction in average annual precipitation after the step change is greater at elevations within the Darling Escarpment, than those on the Swan Coastal Plain, with the exception of Dwellingup. It is thought that Dwellingup has experienced a lesser percentage reduction due to its shorter time-series that does not contain some of the higher annual precipitation values found in the other gauging stations with longer time-series. The standard deviation at each of the gauging stations has also decreased by at least 2 % indicating that the variability in of precipitation has decreased. These results concur with the decrease in the rate of orographic precipitation identified in Section Dominant Frequencies Spectral analysis was undertaken to identify dominant frequencies in the region and to determine if any particular frequencies were more prevalent at different elevations. Ten of the gauging 47

57 stations were used in this analysis; Mandurah, Perth, Gingin, Mundaring Weir, Wungong Dam, Jarrahdale, Chidlow, Pingelly, Beverley and Northam. These gauging stations were chosen because they had the longest complete data sets and provided an adequate range of elevations and locations around the Darling Escarpment. Dwellingup was excluded in this case because it was not considered to have a long enough data set. Detailed results of the spectral analysis and periodogram s are included in Appendix D. It was found that two dominant and four minor frequencies exist in the gauging stations analysed. The frequency of 2.6 to 2.7 years was found at all of the gauging stations and a frequency of 3.1 years was found at seven of the gauging stations; Mandurah, Perth, Gingin, Mundaring, Jarrahdale, Chidlow and Pingelly. A frequency of 2.5 years was found at Pingelly and Beverley, whilst a frequency of 3.6 years was found at Chidlow and Northam. A frequency of 8.33 years was found at Gingin and Chidlow, whilst a frequency of 9 years was found at Jarrahdale and Northam. England et al (26) identified three precipitation frequencies over the SWWA; 8 years, 3.9 years and 2.4 years. However, none of these frequencies corresponds to those observed at any of the gauging stations analysed. However, Ansell et al (2) identified that there is a strong correlation between the Perth MSLP and rainfall in SWWA at frequencies of less than 3 years and at approximately 8 years. The dominant frequency of 2.6 to 2.7 years was identified at all of the gauging stations indicating, according to Ansell et al (2), a relationship between the Perth MSLP and rainfall in the SWWA. However, none of the frequencies appears to specifically relate to any particular elevation or region; Swan Coastal Plain, Darling Escarpment or Leeside of the Escarpment. Nonetheless, the relationship between Perth MSLP and rainfall in the SWWA has been strengthened by these results Seasonal Variability and Trends As previously mentioned, several studies into the climate of the SWWA have suggested that a downward step-change in annual precipitation in the mid-197s associated with a reduction in winter rainfall (IOCI 22, Charles et al 24, Bates et al 28). Time-series analysis was undertaken at every gauging station comparing annual precipitation data to seasonal precipitation 48

58 Rate of Orographic Precipitation (mm/m) data. Eleven year moving averages were determined for all gauging stations to visually determine if and when this step change occurred. It was found that of all the seasons, only winter precipitation experienced a step change in the mid-197s. No trend was established in summer, autumn or spring. Comparisons between annual and winter precipitation data are included in Appendix E. Furthermore, the same step changes observed in the annual precipitation data was also observed in the winter precipitation data. This verifies that the step change observed in annual rainfall in the SWWA is associated with winter precipitation Rate of Orographic Precipitation To determine how orographic precipitation may have changed over time the rate of precipitation in relation to elevation was investigated with respect to time. Figure 17 shows how this rate has varied over the analysis period Year Annual Rate 11yr Moving Average Figure 17 The rate of orographic precipitation over time It is evident from this graph that there has been a step decrease in the rate of orographic precipitation approximately over the period This step decrease matches the step 49

59 Rate of Orographic Precipitation (mm/m) decrease in rainfall that has been observed in the south-west of Western Australia from the mid- 197s and more specifically from 1975 (IOCI 22, Charles et al 24, Bates et al 28). Prior to the step change, that is the period from 19 to 1968, the rate of orographic precipitation ranged from.55 mm/m to 3.58 mm/m with a mean of 1.8 mm/m and a standard deviation of.64. After the step change, that is the period from 1975 to 27, the rate of orographic precipitation ranged from.45 mm/m to 1.9 mm/m with a mean of 1.3 mm/m and a standard deviation of.34. After the step change, the range of the rate of orographic precipitation decreased by more than 5 %. Likewise, the standard deviation almost halved after the step change compared to before the step change. In addition, the mean of the rate of orographic precipitation has decreased by 38 % after the step change Year Winter 11yr Moving Average Figure 18 The rate of orographic precipitation in winter over time The rate of orographic precipitation in winter was then determined to establish if a similar step change was evident. Figure 18 shows that a step decrease in the rate of orographic precipitation occurred approximately over the period This step decrease matches the step decrease in winter rainfall that has been observed in the SWWA from the mid-197s (IOCI 22, Charles et al 24, Bates et al 28), and the step change observed in the annual rate of orographic precipitation. It is suggested that the relatively low rates of orographic precipitation 5

60 experienced in the beginning of the time-series might be attributable to a lack of data from higher elevations. For the annual rate of orographic precipitation prior to the step change the mean of the correlation coefficient squared (r 2 ) was.63 with a standard deviation of.16, while after the step change the mean r 2 value was.68 with a standard deviation of.16. For the winter rate of orographic precipitation prior to the step change the mean r 2 value was.55 with a standard deviation of.21, while after the step change the mean r 2 value was.62 with a standard deviation of.23. This shows that there has not been a significant change before and after the step change in terms of the strength of the relationship between precipitation and elevation. This suggests that the mechanisms of orographic precipitation have not change as a result of the step change but rather that the prevailing atmospheric conditions that lead to the generation of precipitation have altered. Since the rate of orographic precipitation is linear, the greater the rate of orographic precipitation the greater the amount of precipitation at higher elevations. Consequently, a step decrease in the rate of orographic precipitation (both annually and during winter) indicates that precipitation is decreasing at a greater rate at higher elevations. This result is interesting because most research (IOCI 22, Charles et al 24, Bates et al 28) indicates that rainfall has decreased at a similar rate across the SWWA. However, the decrease in the rate of orographic precipitation contradicts this, indicating that there may be significant spatial variations in this decrease Implications The cause of the step decrease in the rate of orographic precipitation and the greater decrease of precipitation within the Darling Escarpment are most likely linked to the causes responsible for the decrease in winter rainfall that has been observed over the SWWA. Smith et al (2) found that the observed decrease in SWWA winter rainfall is linked to the density of low pressure systems within the region and to changes in the MSLP over the southern Indian Ocean. Furthermore, Smith et al (2) suggests that changes in MSLP are part of the coupled air sea interactions resulting from broader changes in atmospheric circulation patterns. However, it is still unknown what is driving these changes (Ansell et al 2, Smith et al 2). 51

61 This decrease in winter rainfall may also be attributed to climate change and global warming associated with the greenhouse effect (Houghton et al 21, IOCI 22). It is thought that as a result of global warming the SWWA has increased in average temperature since the mid 2 th century (Bureau of Meteorology 27, Bates et al 28). This temperature increase may have changed the evaporative balance of the atmosphere and influenced its water holding capacity (Hope & Foster 27). Therefore, climate change may have altered atmospheric conditions in the SWWA. The consensus is that a combination of natural variability and climate change could have caused this step decline in rainfall (IOCI 22, Charles et al 24, Bates et al 28). It is well known that topography modifies precipitation based on pre-existing atmospheric conditions (Smith 1979). Based on this and the results of this study it is thought that the Darling Escarpment has amplified the step decrease in precipitation that has been observed over the rest of SWWA. Regardless of the causes of this step decrease in winter rainfall, the decline in precipitation and the associated drying conditions has seriously affected water resources within the SWWA (Power et al 25) and it is predicted that these drying conditions are expected to continue. The Integrated Water Supply System (IWSS) supplies water to approximately 1.6 million people across the greater south of Western Australia; including Perth, the South West, the Goldfields and the Wheatbelt (McFarlane 25, Water Corporation 28a). Of the IWSS, approximately 2 35% of this water is from surface water sources obtained from dams and storage reservoirs within the SWWA (Water Corporation 28a). Additionally, water resources from these surface water sources are also used for irrigation of agricultural land (McFarlane 25). As can be seen from Figure 19, the majority of dams within the SWWA are located within the Darling Escarpment. 52

62 Figure 19 Locations of SWWA water supply dams (Water Corporation 28b) Prior to the 197 s the average annual streamflow typically ranged from 3 to 2 % of rainfall in the SWWA (Hope & Foster 27). However, a 15 2% decline in annual rainfall, translates to a 4% decrease in annual stream flow and runoff into dams (Charles et al 24). This effect on surface water storage is apparent when observing the current dam s water levels compared to their capacity (Figure 2). 53

63 Figure 2 October 28 water storage in the main water supply dams of the SWWA (Water Corporation 28a) Consequently, a greater decrease in precipitation within the Darling Escarpment has serious implications for current and future water resources within the SWWA since the majority if surface water resources (dams) are located within the Darling Escarpment. Furthermore, since the SWWA previously had relatively consistent precipitation (IOCI 22, Hope & Foster 27), natural ecosystems within this region have been significantly impacted. Additionally, a greater decrease in precipitation within the Darling Escarpment will affect natural surface water flows, the majority of which originate from within the Darling Escarpment. 54