Motive

In recent years, physically based and semi-distributed models have been frequently used to address the influence of land-use change and climate change on hydrology (Weiler et al., 2005). The Generalized Watershed Loading Functions (GWLF; Haith and Shoemaker, 1987; Haith et al., 1992; Wu and Haith, 1993) model is one of the stream flow models that incorporates the physical mechanisms and water balance relationship within a watershed. The most important advantage of using the GWLF model is that parameters can be adjusted according to the land-use types, soil characteristics and climate conditions of a watershed. For this reason, the GWLF model has been widely applied to estimate human, natural, and climate effects on hydrologic systems (Tung and Haith, 1995; Fan, 1998; Cheng et al., 2007; Markel, 2006).

In previous studies, the amount of ET in a watershed was calculated using evapotranspiration cover coefficient (CV) during the GWLF simulated procedures. The CV for each land-use type is defined as a ratio of actual ET to potential evapotranspiration (PET). However, the estimation of actual ET and CV of a large spatial scale is problematic. For example, CV can be determined from the published seasonal values based on crop types such as those given in the user’s manual of the

GWLF model (Haith et al., 1992), this approach often requires estimates of crop development (e.g., planting dates, time to maturity, etc.) which may not be available.

Moreover, a single set of consistent values is seldom available for all of a watershed’s land-use, and settling for a cursory CV value could greatly reduce the accuracy of stream flow simulations (Haith et al., 1992, Davis and Sorensen, 1969).

The increasing availability of remote sensing technology now produces satellite images that can easily and effectively provide large scale spatial and temporal surface information. For hydrology studies, Actual ET can be computed without quantifying other complex hydrological processes through remote sensing techniques (Morse et al., 2000). Thus, much previous research adopted remotely sensed data to calculate the energy balance parameters such as surface temperature, net radiance, sensible heat flux, soil heat flux, and then estimated the actual ET according to these parameters (Chen et al., 2006; Laymon et al., 1998; Mauser and Scha¨dlich, 1998; Menenti and Choudhury, 1993; Morse et al., 2000). Most of these earlier studies focused on the comparison of ET among various spatial scale and temporal stages. However, few researches have calculated CV parameters using remote-sensing-based ET for the purpose of stream flow simulation. Investigations regarding the effects of land-use types and spatial scales on ET are also seldom attempted.

In addition to the climatic factors, land-use change will influence the amount of ET,

and thereby affect the balance of the hydrologic cycle (Cheng et al., 2007). In the GWLF model, parameters such as curve number (CN) and CV are related to the land-use status of a watershed. Most previous studies assumed that catchment land-use remains consistent over long periods of time (Arnell and Reynard 1996). This assumption may reduce the accuracy of prediction. Many existing spatial simulation models have been applied in various fields (Muller and Middleton 1994; Turner 1993).

A Markov model is the most widely used approach. In the Markov model, area change is summarized by a series of transition probabilities from one state to another over a specified period of time. These probabilities can be subsequently used to predict the land-use properties at specific future time points (Burham 1973). Many researchers have applied the Markov model to monitor the land-use and landscape change (Cheng et al.

2005; Lindsay and Dunn 1979; Muller and Middleton 1994; Turner 1993), but few integrate Markov predictions into hydrological assessments under changing climate conditions.

During watershed ecosystem monitoring, we observed that ecosystems are nested and reside within each other. The boundaries of ecosystems are open to transfer energy and materials to or from other ecosystems, and this linkage among systems, energy exchange to occur at various spatial scales. A disturbance to a large system may also affect smaller component systems existing within it. Consequently, the relationship

between an ecosystem at one scale and ecosystems at smaller or larger scales must be examined to predict the effects of human disturbances (Bailey, 1996, Cheng et al., 2005).

Previous research has focused on the effects of global and regional scales on environmental parameters (Rao, 1990; Tokumaru and Kogan, 1993; Yu et al., 2002;

Chen et al., 2006), but studies evaluate the issues of ecosystems at various scales and their effects on environmental parameters. A further investigation of the multi-scale relationship of environmental characteristics under various ecosystem classification systems is needed.

The northern part of Taiwan is a region which includes several political (ex. the capital city of Taiwan: Taipei city), scientific (ex. Hsin-chu science park), and agricultural centers (Lan-Yang flat land). These scientific or agricultural centers play important roles on technology development and crop supplies of Taiwan. An overall hydrological analysis is important in this area. Moreover, to realize how the hydrologic system would be changed in the future is also necessary for the water resource management. Figure 2 is an illustration of the study purposes.

Reduction

watershed) Impact on steam flow Impact on steam flow of watershed

Figure 2. Illustration of the study purposes 2.2. Objective

Based on above purposes, the objectives of this study are as follows:

(1) To evaluate the ET difference among various land-use types

Land-use maps generated by a hybrid classification approach (Hoffer and Fleming, 1978, Lo and Choi, 2004) and daily actual ET obtained from Surface Energy Balance Algorithm for Land (SEBAL; Bastiaanssen et al. 1998a) are integrated to investigate the effects of land-use types on ET.

(2) To analyze the effect of ecosystem classification systems at various spatial scales on environmental parameters

Two spatial scales (regional scale and local) and ecosystem classification systems (geographic climate method and watershed division method) were adopted to assess their effect on environmental parameters.

(3) To investigate the effect of future land-use status and ET change on stream flow simulation under climate change conditions

Compared with the traditional stream flow simulation, which calculates a CV using the published reference values and without evaluating land-use change, our present efforts presents an approach to estimate CV by Markov model and SEBAL model, which includes future land-use status and ET change. Our study was motivated by the following three questions. Is the accuracy of stream flow simulation improved by using the CV estimated from remote sensing? Is the integration of SEBAL model and Markov model a feasible scheme to predict the future land-use and ET parameters for stream flow simulations? Does the consideration of land-use change and ET change affect the results of hydrologic assessment under climate change conditions in north Taiwan?

(4) To assess the future impact on hydrological cycle of north Taiwan

Flow series from 1995 to 2002 were adopted to represent the current hydrological condition, and then compared with the future flows to investigate how land-use change, ET change, and climate change affected river flows and the hydrologic cycle of north Taiwan.

3. LITERATURE REVIEW

3.1. Land-Use Classification using Remote Sensing

Classification of land-use and land cover using satellite images is considered an essential task in modeling the earth as a system. Traditionally, supervised and unsupervised classifications are two common image classification approaches, each with advantages and disadvantages (Lillesand and Kiefer, 2004; Lang et al., 2008). The supervised approach involves a training stage, which allows the input of analyst’s experience into image classifications. However, this approach has been regarded as overly subjective and difficult to correctly implement, because user-defined training data may not be normally distributed. The unsupervised approach can automatically generate almost unlimited number of spectral classes, which are solid spectral foundations for generating information classes, but it requires the analyst manually label the resultant spectral classes into information classes (Lang et al., 2008).

To improve the accuracy of image classification, an integrated algorithm called a hybrid classification approach that takes advantage of both classification approaches has been developed (Hoffer and Fleming, 1978, Lo and Choi, 2004). In this hybrid approach, cluster analysis was first used to acquire the spectral signatures objectively, and then the signature file was imported into the supervised classification to generate the land-use

map. Hybrid classification has been widely applied in ecosystem monitoring studies, and the results from previous studies demonstrated that the integrated algorithm could provide an accurate and consistent classification of land use mapping. For example, Lo and Choi (2004) adopted the hybrid classification method to map the land use/cover of the Atlanta metropolitan area using Landsat 7 Enhanced Thematic Mapper Plus (ETMz) data; Lang et al. (2008) applied the hybrid approach to generate a land-use map of Indiana, USA. Lillesand and Kiefer (2004) indicated that the hybrid approach indeed increased the repeatability and accuracy of land-use classification.

3.2. Estimation of Environmental Parameters and ET based on the SEBAL Methodology

SEBAL is an image processing model that calculates the actual ET and other energy exchanges at the earth’s surface using digital image data collected by Landsat or other remote sensing satellites measuring visible, near infrared, and thermal infrared radiation (Bastiaanssen et al., 1998a). The major concept of this model is that ET flux is calculated as a residual of the surface energy budget equation and is expressed as the energy consumed by the evaporation process:

H G Rn

LE   ₀  (1)

where, LE is the latent heat flux (W/m²); Rn is the net radiation flux at the surface (W/m²); G0 is the soil heat flux (W/m²); H is the sensible heat flux to the air (W/m²). LE is converted into ET, expressed as a depth of water per time, by dividing by the latent heat of vaporization.

Figure 3. General computational process for determining ET using SEBAL (Morse et al., 2000)

Figure 3 is the schematic of the general computational process for determining ET using SEBAL. During the model processes, actual ET is computed as a component using 15 energy balance parameters, including the cosine of solar incidence angle (cosθ;

unitless), twenty-four hour extraterrestrial radiation (Ra24; W/m²), surface albedo at the top of atmosphere (αtoa; unitless), surface albedo (α0; unitless), normalized difference vegetation index (NDVI; unitless), emissivity (ε0; unitless), surface temperature (T0; K), transmittance (τsw; unitless), air density (pair; kg/m³), aerodynamic resistance to heat transport (rah; s/m), estimating friction velocity (u*; m/s), surface roughness for momentum transport, (zom; m), net radiation (Rn; W/m²), soil heat flux (Go; W/m²), sensible heat flux (H; W/m²) (Morse et al., 2000). In SEBAL procedures, Rn was estimated based on the following relationship (Bastiaanssen et al., 1998; Oberg and Melesse, 2004):

S↓is the incoming direct and diffuse shortwave solar radiation that reaches the surface (W/m²); α is the surface albedo, the ratio of reflected radiation to the incident shortwave radiation; R

L↓is the incoming longwave thermal radiation flux from the atmosphere (W/m²); R

L↑is the outgoing longwave thermal radiation flux emitted from the surface to the atmosphere (W/m²); ε

o is the surface emissivity, the ratio of the radiant emittance from a gray body to the emittance of a blackbody.

Soil heat flux (G0) is the rate of heat storage to the ground from conduction. In the

SEBAL model, an empirical relationship for G0 was given as:

Rn

NDVI

G₀ 0.30(10.98 ⁴) (3)

where, NDVI is the normalized difference vegetation index.

Sensible heat flux (H) is the rate of heat loss to the air by convection and conduction due to a temperature difference. The calculated equation was as bellow:

ah the difference in temperature between the surface and the air (K); and r

ah is the aerodynamic resistance (s/m). To calculate dT, the inverse of equation (5) was considered:

Therefore, during the SEBAL process, the user calculated dT at two extreme

“anchor pixels” by assuming values for H at the reference pixels. The reference pixels were carefully chosen so that at these pixels one can assume that H approximate zero at a very wet pixel (i.e., all available energy (Rn - G0) is converted to ET), and that LE almost equals zero at a very dry pixel, so that H = Rn - G0. These assumptions from the selected pixels provided endpoints for values and locations for H so that a relationship for dT can be established.

Once the values of H and G0 were calculated, the latent heat flux (LE) can be calculated from equation (1). This LE represented the instantaneous evapotranspiration at the time of the Landsat overpass. Following the computation of the evaporative fraction at each pixel of the image, one can estimate the 24-hour evapotranspiration for the day of the image by assuming that the value for the evaporative fraction (　) is constant over the full 24-hour period (Bastiaanssen et al. 1998). The evaporative fraction is calculated for the instantaneous values in the image as:

0

where, the values for Rn, G0, and H are instantaneous values taken from processed images. The 24 hour actual evaporation is calculated by the following equation:



where, ET24 is daily actual evapotranspiration (cm/day); Rn24 is daily net radiation; G24

is daily soil heat flux; 86,400 is the number of seconds in a twenty-four hour period; and λ is the latent heat of vaporization (J/kg). The latent heat of vaporization allows

expression of ET24 in cm/day.

Many previous studies applied remotely sensed data to calculate the energy balance parameters, and then estimated the actual ET according to these parameters. For example, Chen et al. (2006) applied the SEBAL model and four seasons of moderate resolution imaging spectroradiometer (MODIS) satellite images to estimate ET for the entire island of Taiwan; Laymon et al. (1998) used Landsat thematic mapper (TM) images and experience functions to estimate energy fluxes and latent heat flux, and further to calculate the ET in a semidesert area of West America; Mauser and Scha¨dlich (1998) modeled the spatial distribution of ET on a different scale using remote sensing data; Menenti and Choudhury (1993) applied Landsat MSS data to develop the surface energy balance index (SEBI) model, and then to estimate the ET of the Libyan desert in West Africa by using surface albedo and aerodynamic roughness; Morse et al. (2000) applied the SEBAL model and satellite images to calculate the ET, and the results showed that the R² value between ET acquired from remote sensing and observed data

was 0.98.

Physical parameters obtained from the SEBAL model also have special meanings on the description of temperature, vegetated, hydrological and energy characteristics of an ecosystem. For example, Ra24 is the daily incoming solar radiation unadjusted for atmospheric transmittance; αtoa and α0 indicate the ratio of reflected to incident solar radiation at the atmosphere and ground surface; ε0 and T0 are temperature indices which denote the thermal energy radiated by the surface and surface temperature conditions of the area; NDVI is a sensitive indicator of the amount and condition of green vegetation;

zom is defined as the height above the “zero-plane displacement” that the zero-origin for the wind profile just begins within the surface or vegetation cover; Rn is the net radiant energy that the land surface actually receives and loses from or to the atmosphere. The allotment of Rn represents the energy transmission process within the ecosystem. Rn is divided into three components; ET24 is the twenty-four hour actual evapotranspiration. It also indicates the energy that used to support the photosynthesis and evaporate soil water; H is the energy used to heat the air; Go is the rest of the net energy which is stored in the ground or water body. The above environmental parameters were computed by the SEBAL model based on an energy balance algorithm. However, the acquisition of surface reflectance would vary with different terrains and meteorological conditions. For instance, the instantaneous and 24-hour solar radiations on a south slope

are much higher than on a north slope in the Northern Hemisphere. Atmospheric humidity and soil moisture are two important factors for ground reflectance, and they might influence the calculation of environmental parameters (Cheng et al., 2008).

3.3. Climate Change Scenarios

Climate change is a very complex issue. Policymakers need an objective source of information about the causes of climate change, its potential environmental and socio-economic consequences, and the adaptation and mitigation options to respond to it. This is why World Meteorological Organization (WMO) and UNEP established the Intergovernmental Panel on Climate Change (IPCC) in 1988.

The IPCC is a scientific body. The information it provides with its reports is based on scientific evidence and reflects existing viewpoints within the scientific community.

The comprehensiveness of the scientific content is achieved through contributions from experts in all regions of the world and all relevant disciplines including, where appropriately documented, industry literature and traditional practices, and a two stage review process by experts and governments.

The IPCC currently has three Working Groups and the Task Force on National Greenhouse Gas Inventories. The Working Groups and the Task Force have clearly defined mandates as agreed by the Panel and their activities are guided by two Co-chairs

each. They are assisted by a Technical Support Unit and the Working Group or Task Force Bureau. Working Group (WG )Ⅰ Ⅰ deals with "The Physical Science Basis of Climate Change", Working Group Ⅱ (WG ) with "Climate Change Impact, Ⅱ Adaptation and Vulnerability" and Working Group (WG ) with "Mitigation of Ⅲ Ⅲ Climate Change". The main objective of the Task Force is to develop and refine a methodology for the calculation and reporting of national green house gas emissions and removals. In addition to the Working Groups and Task Force, further Task Groups and Steering Groups may be established for a limited or longer duration to consider a specific topic or question (IPCC, 2004).

At regular intervals, the IPCC provides assessment reports of the state of knowledge on climate change, which become standard works of reference, widely used by policymakers, experts, and students. The findings of the first IPCC Assessment Report of 1990 played a decisive role in leading to the United Nations Framework Convention on Climate Change (UNFCCC), which was opened for signature at the Rio de Janeiro Summit in 1992 and enacted in 1994. It provides the overall policy framework for addressing the climate change issue. The IPCC Second Assessment Report of 1995 provided key input for the negotiations of the Kyoto Protocol in 1997, and the Third Assessment Report of 2001, as well as Special and Methodology Reports, provided further information relevant for the development of the UNFCCC and the

Kyoto Protocol. The IPCC continues to be a major source of information for the negotiations under the UNFCCC. The latest one is "Climate Change 2007", the Fourth IPCC Assessment Report (IPCC, 2007).

In 2000 the IPCC published a new set of emission scenarios, which address changes in the understanding of driving forces and emissions and methodologies since the completion of the IPCC IS92 scenarios. The Special Report on Emissions Scenarios (SRES) are based on an extensive assessment of driving forces and emissions in the literature, alternative modeling approaches and an “open process” that solicited participation and feedback from scientist’s around the world. An important part of this report is the consideration of the contributions to future emissions, from demographic to technological and economic developments, but, as requested in the terms of reference, none of the scenarios included future policies that explicitly address climate change.

Four different storylines were developed to describe the relationship between emission driving forces, and their evolutions, and add context to the scenario quantification (IPCC, 2000).

3.4. Application of the GWLF Model on Hydrologic Monitoring

The GWLF model is a physical hydrological model, which simulates the water balance within an upstream watershed. In the GWLF model, the stream flow consists of

runoff and discharge from groundwater. The latter is obtained from a lumped parameter watershed water balance. Daily water balance is calculated for unsaturated and shallow saturated zones. Infiltration to the unsaturated and shallow saturated zones equals the excess, if any, of rainfall and snowmelt less runoff and ET. Percolation occurs when unsaturated zone water exceed field capacity. The shallow saturated zone is modeled as a linear groundwater reservoir (Haith et al., 1992). The model structure of GWLF is shown as Figure 4.

Figure 4. Water balance function of the GWLF model

Figure 4. Water balance function of the GWLF model