Climate change impacts on hydrology and water resources

Get Complete Project Material File(s) Now! »

CHAPTER 3 Study sites and data

This chapter gives a brief description of the study region and its data. Different sub-catchments are highlighted because this research aims to test the hypotheses with different catchment characteristics which may reflect the remarks on the findings. The first section presents information on the study region and its sub-catchments. The next section describes the main characteristics of the selected catchments. Data types and sources are explained in the following sections. The chapter summary is presented in the final section, emphasizing the differentiation of data usage for certain sub-catchments.

Study region

North Island of New Zealand

The preliminary study region of this study is the North Island of New Zealand. The total area of the North Island is 117,713 km2 making it the world’s 14th largest island. It had a population of 3,422,000 by June 2013, accounting for approximately 77% of New Zealand’s population. The geographic location of the North Island is between 350S and 420S in latitude and 1720E, 1790E in longitude and covers the areas of nine administrative regions, namely Northland, Auckland, Waikato, Bay of Plenty, Gisborne, Hawke’s Bay, Taranaki, Manawatu-Wanganui and Wellington (Figure 3.1).
The climate of the North Island is characterized as warm subtropical. Climate in the region is dominated by westerly winds and dramatically driven by topographic effects. However, the region can further be divided into different climate zones such as Northern New Zealand (Northland, Auckland), Central North Island, South-West North Island and Eastern North Island.
Most areas of New Zealand have between 600 mm and 1600 mm of annual precipitation spread throughout the year with a dry period during the summer. Over the North Island areas, more precipitation falls in winter than in summer. Mean annual temperatures range from 10°C in the south to 16°C in the north of New Zealand. The coldest month is usually July and the warmest month is usually January or February. Temperatures also drop about 0.7°C for every 100 m of altitude. Being consistent with temperature variation, more evaporation is recorded in summer and less evaporative water occurs in winter. Average daily evaporation varies between 0.3 mm/day to 5.1 mm/day for winter and summer days, respectively (Sturman, 2001; NIWA, 2005). The information on spatial variability of mean annual precipitation and potential evaporation along with mean annual temperature over New Zealand is presented in Figures 3.2 -3.4.

Waikato catchment

Waikato catchment is the largest catchment in the North Island of New Zealand. The total area of Waikato catchment is about 14,500 km2 making up about 12% of the North Island (Figure 3.5). The Waikato River is the longest river in New Zealand with a total length of 425 km. The river originates from Mount Ruapehu, the largest active volcano in New Zealand, at 2,797 m above sea level and drains into Lake Taupo, the largest lake in New Zealand with a surface area of 616 km2. The Waikato River flows through the North Island passing eight dams and nine hydroelectric powerhouses northward into the Tasman Sea at Port Waikato from Lake Taupo.
Mean annual precipitation varies between 1,100 mm to 2,000 mm per year in the lower and upper catchment, respectively. Mean annual evapotranspiration is approximately 650 mm per year accounting about half the annual precipitation (Waikato Regional Council, 2008). Evaporation and transpiration are mainly from pasture and forest (Sturman, 2001).
The annual sunshine total ranges between 1,800 hours and 2,100 hours, wind speed is from 2 m/s to 10 m/s depending on the change in topography (Figures 3.6-3.7).
The Waikato catchment plays an important role in both regional and national development since hydroelectricity and farming are the main activities in the catchment region. For example, the hydroelectric schemes in the Waikato catchment provide approximately 13% of New Zealand’s current electricity requirements (Dravitzki, 2009). However, these activities are heavily dependent on precipitation. This may create pressures on the water resource management of the catchment in which the major concern is about evaporative water loss. In addition, soil water deficit also plays an important role in the surface water balance because soils in many areas of New Zealand have highly variable available water capacities (NIWA, 2005). Flooding that is a major consequence of heavy precipitation has become a big problem in the region. Approximately 67% of annual maximum floods occur in the winter-spring period (July-October), and only 14% of annual maximum floods have occurred in the summer months. Although impacts of flood and inundation are mitigated through river control, flood plains are still most vulnerable to flooding. This is because there is rapidly increasing residential and farming activities within the region (Dravitzki, 2009).

Representatives of forest and grass sub-catchments

The Waikato catchment is mainly covered by forest land and grassland accounting for more than 80% of the total catchment area (Figure 3.8). The three representative forest and grass land sub-catchments are selected as pilot case studies which are shown in Figure 3.9 and Figures 3.10-3.12. They are Mangatawhiri, Mangaonua and Whangamarino sub-catchments with a total area of about 182 km2, 166 km2 and 131 km2, respectively.
Mean areal annual rainfall in these catchments varies between 1,200 mm and 1,600 mm. Mean annual evaporation varies between 800 mm and 1,100 mm. Recorded mean annual streamflow fluctuates from 540 m3/s to 760 m3/s. The similar patterns in precipitation, evaporation and flow in these sub-catchments are shown in Figures 3.13 and 3.14. The elevation of these sub-catchments ranges between 40 m and 300 m, stretching from upstream to downstream of the sub-catchments.
The evaporation and transpiration processes in these sub-catchments may reflect the impact of vegetation cover on evaporative water loss. Landuse maps are presented in Figures 15-18. Along with this, groundwater and surface runoff could also be controlled by different soil types that differ from sub-catchment to sub-catchment (Figures 3.17-3.19). The dominant soils in the Mangatawhiri and Mangaonua sub-catchment are recent and allophamic soil respectively, while mixed ultic, gley, recent and brown soils are found in the Whangamarino sub-catchment. According to Newsome et al. (2000), recent soils have variable soil texture with high plant-available water capacity. Allophamic soils have low bulk density, little resistance to root growth and their top soils are stable and resist the impact of machinery and grazing animals in wet weather. Gley soils have high ground water table, shallow potential rooting depth and relatively high bulk density. Ultic soils have dispersible surface horizons susceptible to livestock treading damage, and are prone to erosion.

Data collection and processing

Data collection and analysis of different data types were implemented. These serve as input for extreme precipitation prediction, potential evapotranspiration estimation and water balance modelling as follows:
• Observed meteo-hydrological data
• Global Climate Model (GCM) and Regional Climate Model (RCM) raw data
• Topographic maps
• Soil and landuse maps
Traditionally, meteo-hydro data is measured at stations. Its daily time scale is appropriate because this properly explains trend in precipitation, potential evapotranspiration and streamflow.
In this research, there are numerous software and programmes were used to process the data. In particular, ArcGIS software version 10 was used to incorporate with spatial data types, such as to build maps, to visualise gridded data and to analyse spatial data.
The HDF View and the Unidat Net CDF packages were used to extract the large-scale data sets from GCMs/RCM at the station locations and for integrating the results over the study period. The two packages were originally developed by National Centre for Supercomputing Application (NCSA) and by North American Regional Climate Change Assessment Program (NARCCAP), respectively.
This research also developed several programmes based on Visual basic and FoxPro languages to deal with time series data pre-processing, including data extraction, combination and analysis.

 Observed meteo-hydrological data

Meteorological data that is used in this research includes precipitation and evaporation at daily time scale. Observed precipitation data was selected at 36 different stations across the North Island region including Waikato catchment. This selection may accommodate the study on variability of daily precipitation that is apparently very complex in comparison to other climatic variables’. Observed evaporation data from eight stations in the Waikato catchment comprising three representative sub-catchments was assembled to serve the purpose of detecting the trend in evaporation rate from the past to the present, especially over typical forested and grassed surfaces. Data was obtained from the National Institute of Water and Atmospheric Research of New Zealand (NIWA) database website (http://cliflo.niwa.co.nz/), which are available to download online. Likewise, daily discharge at three stations in the representative sub-catchments was collected for hydrological model calibrating and validating. This data was obtained from the Environment Waikato Department, New Zealand. In general, the observation period is between 1945 and 2010 for daily precipitation, 1970 and 2006 for daily evaporation. The summary on selected data is presented in Tables 3.1 and 3.2.

Global and Regional Climate Model (GCM/RCM) data

Daily large-scale climatic variables including precipitation, air temperature, wind speed, radiation, and relative humidity as the outputs of two GCMs and one RCM are used for this research.
Basically, GCM and RCM data contains daily values of 26 large-scale variables coupled with different scenarios developed by the Intergovernmental Panel on Climate Change (IPCC SRES scenarios). In this research, 26 large-scale variables are obtained from two GCMs, the Canadian Global Climate Model (CCGCM3.1/T47) which is presented in a grid form at 3.750×3.750 spatial resolution and the United Kingdom Hadley Centre Coupled Model (HadCM3) at 2.50×2.50 spatial resolution. The corresponding 26 daily observed large-scale predictor variables are originally from the National Centre of Environmental Prediction (NCEP) re-analysis with spatial resolution of 2.50×2.50. Both the GCM and NCEP data are obtained from same source in which the NCEP data is regridded to the horizontal grid resolution of the GCM.
The CCGCM3.1/T47 and HadCM3 data were obtained at the Data Access Integration (DAI, see http://quebec.ccsn.ca/DAI/) and the Canadian Climate Impacts and Scenarios project (CICS, http://www.cics.uvic.ca/scenarios/). Likewise, 26 dynamically downscaled variables from a very high 0.050×0.050 spatial resolution RCM HadCM3 model by the NIWA were used.
These GCM and RCM data sets contain both historical and future data. Data over two different periods, 1961-2000 and 2001-2090 were defined for present and future climates, respectively.
Future data from only one emissions scenario, IPCC SRES A2 scenario, was then used for all simulations. The A2 emissions scenario was selected because it was one of the ‘marker’ scenarios developed through the IPCC (IPCC, 2000, 2007). Furthermore, the A2 scenario is commonly used by the NIWA to study the climate change effects and impacts assessment for New Zealand local governments (MfE, 2008; Mullan et al, 2001a, b).
CGCM3.1/T47, GCM HadCM3 and RCM HadCM3 data are presented in a grid form with their sizes of 375 km x 375 km, 250 km x 250 km and 5 km x 5km, respectively (Figure 3.20). There are four grid cells in CCGCM3.1/T47 data and six grid cells in GCM HadCM3 data covering the whole North Island including the Waikato catchment. There are 3600 cells in RCM data covering the whole Waikato catchment. RCM data sets are extracted to the points that the selected stations and sub-catchments are located (Table 3.3).
These data were used for simulating the variation in precipitation and potential evapotranspiration from the past to the future. For the convenience of computation as well as the comprehensive assessment of future projections, the decadal 30-year that was the standard normal period defined by the World Meteorological Organization (WMO) is used.

Soil type and landuse maps

For soil-water balance modelling, soil type and landuse maps are required. These data are obtained from the Landcare Research Institute (LRIS), the Waikato Regional Council (WRC).

Summary

This chapter gave a brief overview of the study region and its data availability for the research. The description of the North Island region was first presented, followed by the Waikato catchment and its three sub-catchments. The selection was based on data availability in the study areas and surroundings. Moreover, this research thesis aims to investigate the precipitation variability across the North Island. Evapotranspiration variability is examined over the Waikato catchment while water balance is inspected from its three sub-catchments with different landuse and soils. This chapter also introduced different data types and sources coupled with several software and programs which are effective for both temporal and spatial data ranging from data pre-processing to analysis.

CHAPTER 4 Extraction of extreme precipitation

This paper includes partial reproduction of the paper:
Pham, H. X., Shamseldin, A. Y. and Melville, B. W. (2014). Statistical properties of partial duration series. J. Hydrol. Eng., 19, 807-815.
This chapter focuses on abstracting extreme precipitation events from observation time series. A novel peak-over-threshold (POT) method is used to define the most plausible Partial Duration Series (PDS) model for the whole North Island of New Zealand. The first section of the chapter highlights the POT and PDS background and objectives. Following this, the technical and computational details including various statistical tests are explained. Finally, the selection of the most plausible PDS model for the study region is discussed.

Introduction

Partial Duration Series (PDS) model is preferable for hydrological frequency analysis when dealing with values exceeding a certain threshold, and it is capable of capturing more information about extreme events than the rival Annual Maximum Series (AMS) model. However, the use of PDS in hydrological application encounters the difficulty of determining the exceedance threshold due to physical and statistical complexity. There is an inverse relationship between the threshold value and the number of peaks selected above a threshold and a fixed number of average peaks per year is an option to define the threshold value. Along with this, a PDS is generated under an assumption that it is best described by a Generalized Pareto (GP) distribution. To date, the question arisen is how many peaks should be appropriate for modelling the PDS. This research investigates the optimum number of peaks for integer values based on the variability of the GP/PDS parameters with an increase in the value of number of peaks. The research for the first time addresses the question by using a wide range of historical precipitation data from the North Island region of New Zealand.

Overview of Partial Duration Series by Peak-Over-Threshold (POT)

Characteristics of PDS

PDS is a series of peaks over a threshold (POT) which is characterized by the average number of peaks each year (λ) or a threshold value (u). Given a daily series of observations x = {x1, x2, .. xi .., xn} over N years, n = 365*N, the resulting PDS is the set containing M values, y = {y1, y2, .. yj .. , yM}, which exceed a chosen threshold u, yi > u with u = constant.
In the PDS, the total number of peaks (M) is a random quantity which can be chosen from the average number of peaks per year, λ = M/N. The peaks are assumed to be independent random variables described by the GP distribution and the peak arrival following the Poisson distribution (Cunnane, 1973). The special case of the PDS, λ = 1, is the Annual Maximum Series (AMS) where the N highest peaks within the N years of data are chosen (Gordon et al., 2004).
The number of elements in the PDS series, (λN) depends on either the average number of peaks per year or the threshold value. For this reason, the PDS use the information from the original series more efficiently because it includes on average more than one event each year, if they satisfy the conditions established in defining an extreme event.

Significance of threshold value

To date there is no agreement on the selection of the optimum threshold value. Principally, the threshold value (u) decreases with the increase in the value of the average numbers of peaks per year (λ). The threshold value must be set at a high value to ensure only true peaks are selected, with the peak arrival being governed by the Poisson distribution. In contrast, the threshold must be set low enough to ensure that sufficient peaks are selected to enable reliable estimation of the distribution parameters (Ashkar et al., 1983, 1987; Buishand, 1989; Cunnane, 1979, 1985; Langa et al., 1999; Palutikof et al., 1999; Onoz and Bayazit, 2001; Trefry et al., 2005).

GP-Poisson distribution and parameter estimators

The Poisson distribution of peak occurrences

The number of peaks occuring in t years N(t) follows a random process which can be described by the Poisson distribution. In the case of the Poisson distribution, the probability density function is given by:
where n is the numbers of peaks within time interval (0,t), and λ is the mean arrival rate (or the average number of peaks per year) which can be estimated from the sample. One of the properties of the Poisson distribution is that the mean and variance are equal.

The GP distribution of peak magnitudes

The PDS set contains all mutually independent peaks following the GP distribution. The probability density function of the GP distribution f(x) for a variable x is given by:
where α, k and ζ are the scale, the shape and the location parameters, respectively. These parameters can be estimated by the method of L-moment (L-MOM) according to the following equations given by Hosking and Wallis (1997):
where τ3 is the L-MOM coefficient of skewness while λ1 and λ2 are the first and second order L-MOM.

Parameter estimators

The use of L-moment measure is very effective for estimating distribution parameters (Greenwood et al., 1979, Hosking, 1986, 1990). Firstly, L-MOM is less influenced by the effects of sampling variability than conventional moments. Also, in some cases it produces more efficient parameter estimates. L-MOM is robust to outliers and virtually unbiased for small samples, making them suitable for regional frequency analysis, including identification of distribution and parameter estimation (Adamowski, 2009). L-MOM is identified as linear function of PWM:
βr = E{X[F(X)]} (4.6) where, F(x) is the cumulative distribution of x
The first four L-MOM expressed as linear combination of PWM are:
λ1=β0
λ2=2β1–β0 (4.7)
λ3=6β2-6β1+β0
λ4 = 20β3 – 30β2 + 126β1 – β0
where, L-mean, λ1, is a measure of central tendency L-standard deviation, λ2, is a measure of dispersion L-coefficient of variation, τ = λ2/ λ1 L-coefficient of skewness, τ3 = λ3/ λ2 L-coefficient of kurtosis, τ4 = λ4/ λ3
Physically, τ will be smaller if any two values tend to be close together than if they are far apart. τ3 will be positive if the lower two values of any subsample of three tend to be closer. Conversely, if the upper two are closer to each other than τ3 will be negative. For symmetrical distribution, such as normal distribution, τ3 will be close to zero. τ4 will be positive if middle values in any subsample of four tend to be closer, or conversely.

Data

Daily rainfall data was assembled from 36 stations (refer to Figure 4.1) where the station record length varied between 31 and 105 years. About 50% of the selected stations have record lengths greater than 60 years (see Figure 4.2).
The partial duration series (PDS) set is extracted from the observed daily rainfall data at each station by setting different number of average peaks per year with λ equal to 1, 2, 3, 4 and 5. This is due to the fact that there is no clear predetermined physical threshold in the observed precipitation series available at each station. Although λ can be a real number, this research uses integer values to simplify the analysis. The resulting models henceforth are referred to as PDS1, PDS2, PDS3, PDS4 and PDS5.

List of Figures
List of Tables
Abbreviations
Co-Authorship forms
CHAPTER 1 Introduction
1.1. Background
1.2. Motivation and novelty of the research
1.3. Research Objectives
1.4. Thesis Outline
CHAPTER 2 Literature Review
2.1. Climate change
2.2. New Zealand and Waikato climate change and its impacts
2.3. Climate change impacts analysis
2.4. Climate change impacts on hydrology and water resources
2.5. Regional hydrologic and water resources impact assessment of climate change
2.6. Summary
CHAPTER 3 Study sites and data
3.1. Study region
3.2. Data collection and processing
3.3. Summary
CHAPTER 4 Extraction of extreme precipitation
4.1. Introduction
4.2. Data
4.3. Methods
4.4. Results and discussions
4.5. Summary
CHAPTER 5 Regional extreme precipitation analysis
5.1. Introduction
5.2. Methods and techniques
5.3. Results and discussions
5.4. Summary
CHAPTER 6 Statistical downscaling of daily precipitation
6.1 Introduction
6.2 Study region and data
6.3 Methods and Techniques
6.4 Results and Discussions
6.5 Summary
CHAPTER 7 Future projections on extreme precipitation
7.1 Introduction
7.2 Data sets and methods
7.3 Results and Discussions
7.4 Summary
CHAPTER 8 Future projections on potential evapotranspiration
8.1. Introduction
8.2. Data sets and methods
8.3. Results and Discussions
8.4. Summary
CHAPTER 9 Assessment of climate change impacts on catchment water balance
9.1. Introduction
9.2. Data and methods
9.3. Results and Discussions
9.4. Summary
CHAPTER 10 Conclusions and recommendations
10.1. Conclusion remarks
10.2. Contributions to knowledge
10.3. Possibilities for further research
GET THE COMPLETE PROJECT