Climate Change Scenarios Analysis in Coastal Region of Thailand

Abstract

The impact of climate change is estimated to be particularly severe in many developing countries, including the coastal zones prone to flooding and drought. As a coastal region, Thailand has been affected by climate change significantly in terms of temperature and rainfall distribution change. In this study, a downscaling study was carried out to identify some climate-related planning parameters for the representative cities in Thailand using the regional climate model. Results indicated that in the study area around the Hat Yai, the near future atmosphere temperature will increase about 1 degrees Celsius to 1.5 degrees Celsius compared to the present condition. The minimum annual temperature will increase about 0.8 degree from 23 degrees Celsius to 23.8 degrees Celsius. It is ranging between 20.8 degrees Celsius-25 degrees Celsius. The maximum annual temperature in Hat Yai will increase about 1 degree from 32.6 degrees Celsius to 33.6 degrees Celsius. The mean annual temperature will increase about 1 degree from 27.8 degrees Celsius to 28.8 degrees Celsius.

Headnote

ABSTRACT

Wang, Y.; He, B.; Herath, S.; Basnayake, S. and Huang, W., 2014. Climate change scenarios analysis in coastal region of Thailand. In: Huang, W. and Hagen S.C. (eds.), Climate Change Impacts on Surface Water Systems. Journal of Coastal Research, Special Issue, No. 68, pp. 160-167. C°Conut Creek (Florida), ISSN 0749-0208.

The impact of climate change is estimated to be particularly severe in many developing countries, including the coastal zones prone to flooding and drought. As a coastal region, Thailand has been affected by climate change significantly in terms of temperature and rainfall distribution change. In this study, a downscaling study was carried out to identify some climate-related planning parameters for the representative cities in Thailand using the regional climate model. Results indicated that in the study area around the Hat Yai the near future atmosphere temperature will increase about 1 °C to 1.5 °C compared to the present condition. The minimum annual temperature will increase about 0.8 degree from 23 °C to 23.8 °C. It is ranging between 20.8 °C- 25 °C. The maximum annual temperature in Hat Yai will increase about 1 degree from 32.6 °C to 33.6 °C. The mean annual temperature will increase about 1 degree from 27.8 °C to 28.8 °C. The temperature will increase largely in summer and rainy season in the future. The near future rainfall is projected to increase in most of seasons, especially during the rainy season which will bring more risk for flood disaster. Furthermore, rainfall pattern and distribution will be also changed in the near future in Songkhla, with more rain to be expected in rainy season.

ADDITIONAL INDEX WORDS: Climate change, temperature, rainfall, coastal region, Thailand.

(ProQuest: ... denotes formulae omitted.)

INTRODUCTION

Adaptation to climate change has become one of the focal points of curremnt development discussion. The impact of Climate change are estimated to be particulary severe in many developing countries, including the coastal zones and the areas prone to flooding and drought (UNDP, 2006; IPCC, 2007; Michel and Pandya, 2009; World Water Assessment Programme, 2009; Zhang et al., 2013). Southeast Asia is expected to be seriously affected by the impacts of climate change due to the high dependency of economy on agriculture and water resources in the region. Thaliand is one of developing countries in this region that has been affected by climate change signficantly in term of temperature and rainfall distribution change. The occurrence of flood and drought phenomena tend to be more higher severity and frequency in the country (Koontanakulvong, 2011). Meanwhile, with the limited information and adaptive capcity about future climate condition under the changing environment , the people in this region are expected to be severely threatened by additional influence of climate.

Generation of climate change scenarios can play vital role in prevention and mitigation of meteorological disasters (Mendes and Marengo, 2010). Outputs of the climate models are the major candidates to be the inputs of physical- hydrological- and agricultural-models (He et al., 2011). General circulation models (GCMs) representing physical processes in the atmosphere, ocean, cryosphere, and land surface are the most advanced numerical tools currently available for simulating the response of the global climate system to increasing greenhouse gas concentrations (Mendes and Marengo, 2010). While they demonstrate significant skill at the continental and hemispherical scales and incorporate a large proportion of the complexity of the global system, they are generally not designed for local or regional climate change impact studies and are inherently unable to present local subgrid-scale features and dynamics owing to their coarse spatial resolution (Coulibaly, 2004). In this connection, this study was done using the regional climate model called PRECIS (Providing Regional Climate for Impact Studies) for Thailand. Regional climate models can generate necessary parameters as accurate as possible compared to Global Climate Models (GCMs). Therefore, adaptation of the regional climate model is essential for the assessment of climate change impacts to be occurred in near future to the country. As soon as we can utilize the outputs of climate models, the Government as well as the people of the country will get benefit from the scientific community.

DATA AND METHODS

Study Site Description

Hat Yai municipality, located in the downstream area of the Khong UTapha Basin, is the center of commercial trade and administration in southern Thailand. Its ideal geographical location also makes it a gateway to the major neighboring countries of Malaysia and Singapore, and is thus a city with high tourism potential for domestic and foreign tourists who visit year-round. Due to its geographical characteristics and to unplanned urbanization and deforestation in upstream areas, Hat Yai Municipality has become extremely vulnerable to flood disasters. With its population density and as a commercial center, the municipality has characteristics that can magnify the impact of flooding to which it is prone. The unprecedented flooding of 21 to 24 November 2000, triggered by torrential rains, has been described as one of the worst natural disasters in the history of urban Thailand. Future climate change is going to accelerate with the high likelihood of future flood disasters and the increased severity of flooding in the area. Figure 1-(a) shows the rain stations in Hat Yai province which was studied in this research. There are 57 rain stations in Hat Yai were used for the observation rainfall data in this study.

Model Set-up

The Regional Climate Model (RCM) used in this study is PRECIS (Providing Regional Climate for Impact Studies) developed by the Hadley Centre of the UK Meteorological Office. The PRECIS RCM is based on the atmospheric component of the HadCM3 climate model (Gordon et al., 2000) and is extensively described in Jones et al. (2004). The atmospheric dynamics module of PRECIS is a hydrostatic version of the full primitive equations and uses a regular longitude-latitude grid in the horizontal and a hybrid vertical coordinate. For this study, the PRECIS model domain for Southeast Asia has been set up with a horizontal resolution of 25x25 km. Our domain roughly stretches over latitudes 16.0?S to 40.0?N and longitudes 67.0 to 134.0?E. This domain covers Southeast Asian countries and South Asian countries as well (Fig. 1-(b)). It allows full development of internal mesoscale circulation (e.g. monsoon circulation) and includes relevant regional forcing.

The representation of topography is an important feature of climate models as it has a strong impact on the simulated climate fields, in particular spatial precipitation distribution. Where terrain is flat for hundreds of kilometers and away from coasts, the coarse resolution of a GCM may not matter, because higher resolution of PRECIS RCMs provides much better topographic details over the same region. The PRECIS RCM driven by ECHAM05 boundary data with A1B SRES moderate emission scenario for the period 1980-2039. ECHAM05 boundary data is produced by the Max Planck Institute for Meteorology in Germany. It covers the period 1940-2099 and In this study, the time periods for the model executing were 1980-2009 (baseline period) and 2010-2039 (future period) respectively. The first year in the run was considered as a spinup period and these data are not used in any analysis. After model result calibration procedure, the time series of temperature and precipitation for selected city was generated.

Bias Correction /Validation for Temperature

Model validation using historical station data is a crucial step in downscaling. In this study, the biases of temperature were calculated at the selected stations using three different bias estimation/correction approaches respectively against the observations, and the one that responded best to the observed variations was applied. After bias correction, the simulated temperature data for 2010 to 2039 are extracted at the coordinates of the observed meteorological stations for the study area. The extracted data are processed on monthly, seasonally and annual scales, and objectively compared with the observed data on the same scale. The correction of temperature involves shifting and scaling to adjust the mean and variance (Leander and Buishand, 2007; Terink et al., 2010) because temperature is assume to be normally distributed. The following three approaches were applied to the data in period 1981-2005 and verified for the data in period 2005-2009.

Approach 1

...

Where, T* is the value to be obtained after calibration. Tobs and Tmod are the observed and model monthly temperature respectively. In this equation over bar denotes the average over the considered period and SD the standard deviation.

Approach 2

...

Where the regression coefficients (α (slope) and β (intercept)) are obtained using a scatter plot of temperature from both model estimation and observation for considered period.

Approach 3

...

Bias Correction for Rainfall

Understanding of rainfall variability and trends in that variability is needed to help policymakers address current climate variation, future climate change and explore the policy implications for climate adaptation. Unlike temperature, rainfall varies in small scale, and climate change is likely to increase that variability in many of these regions. Thus, in this study, major research effort has been made on correcting the bias of rainfall in each grid point in the study area in order to get the more fine and accurate result to understand the rainfall distribution properties.

The detailed processes for rainfall bias correction are shown in the following:

(1). Extract PRECIS precipitation data in each grid in the boundary of study region;

(2). Collect rainfall from observed stations from Royal Irrigation Department (RID), Thai Meteorology Department (TMD);

(3). Clip the grid point of PRECIS data by using of boundary of the study area;

(4). Interpolate the observed data from observation stations (base time data) in monthly time series data in year 1980 to 2009 by using inverse Distance Weighting Method in the same grid point of PRECIS data;

(5). Correct the bias of PRECIS precipitation by using SD ratio method in each grid which inside the boundary of the study area;

(6). Validate the result of bias corrected rainfall with the observed data.

In this study, the SD ratio method is applied in this study to correct the bias of PRECIS precipitation, this bias correction technique was developed by Cheng et al. (2007), and the equation of SD ratio method is shown in the following:

...

where Phis-new and Phis-old are predictions of monthly data for a GCM historical run after and before correction. StdO and Stdhisold are standard deviations of observations and model predictions for the historical run, respectively. his old P - and O are overall averages of model predictions and observations.

...

where Pfuture-new and Pfuture-old are monthly model projections after and before correction. Stdhis-new and his old P - are the standard deviation and mean of corrected model projections of the historical run using Eq. (1).

RESULTS AND ANALYSIS

Temperature Data Validation

The bias correction methods for temperature were calibrated for the data in period 1981-2005, and validated for the data in period 2005-2009. Calibration of downscaled temperature during the year 1981 to 2005 and verification of downscaled temperature during the year 2005 to 2009 are shown in Table 1. In the calibration and verification for the bias correction method, Root Mean Square Error (RMSE) and Coefficient of Determination (R2 ) are used to check the performance of bias correction method. If RMSE has small value (approaching 'zero') and R2 has big value (approaching 'one'), it means bias correction has good performance to represent the observed data. Compare the raw PRECIS model output with different bias correction methods, it is shown that bias correction methods can improve the raw PRECIS output closer to the observed data, and among 3 bias correction methods, it can be found that Method 2 shows the best performance, for the smallest RMSE value and biggest R2 value. Thus method 2 is applied in this study for reduce the bias of raw PRECIS temperature data.

Temperature in Hat Yai

(1) Annual temperature change

The temperature is described as an averaged value, and as a daily fluctuating value reflected by a minimum temperature and maximum temperature. Thus the average minimum, mean and maximum annual temperature changes in Hat Yai, from 1981 to 2039 are shown in Figure 2. In the figure, red curve represent bias corrected temperature from model simulation and prediction, while black curve represent observed temperature at station. From the figure and analysis, it can be found that the annual temperature tends to increase in general. And in details, the monthly temperature from 1981 to 2039 in Hat Yai is ranging between 25.5°-30.5°C. The lowest mean temperature 25.5 °C occurs in December, and the highest temperature 30.5 °C occurs in March. The minimum annual temperature will increase about 0.8 degree from 23 °C to 23.8 °C. It is ranging between 20.8 °C- 25 °C. The maximum annual temperature in Hat Yai will increase about 1 degree from 32.6 °C to 33.6 °C. The mean annual temperature will increase about 1 degree from 27.8 °C to 28.8 °C. The temperature will increase largely in summer and rainy season in the future.

(2) Seasonal temperature Change

The seasonal temperatures change in Hat Yai is shown in Figure 3. It can be seen that the temperature tends to increase in all seasons in Hat Yai in future, and the seasonal temperature will rise about 0.4 °C to 1.5 °C on the average.

Temperature Data Projection

From the analysis of bias corrected future projection of climate data, the trend of future temperature can be investigated by comparing with the present period-baseline (1980-2009) in Hat Yai city.

It was found that the future temperature tends to increase and varies by areas. the bias corrected downscaled temperature reveals that temperature will increase 0.5°C and 0.8°C in winter (Nov. Dec. Jan. and Feb.), increase about 0.8°Cin summer time and increase around 0.8°C and 1.0°C in rainy season comparing to the baseline. In General, the area around the Hat Yai temperature measuring station will have an increasing temperature about 1 °C to 1.5 °C compared to the present in near future. Especially, temperature will increase largely in summer and rainy season in the future.

Rainfall Data Projection

Since the SD ratio bias correction method is applied to correct the bias from raw PRECIS output precipitation and compare with the observed rainfall data in each observation station. The performance of SD ratio bias correction method is shown in Table 2. By comparing Root Mean Square Error (RMSE) and Coefficient of Determination (R2), it can be seen that SD ratio method can reduce error of raw PRECIS output obviously.

Moreover, in order to check the performance of SD ratio method, the bias corrected raw PRECIS output rainfall in each grid point in Hat Yai city are compared with observed rainfall respectively. The comparison is shown in Figure 4 From the comparison, it can be seen that bias corrected annual rainfall in both cities are more closed to the observed rainfall compare with raw model output rainfall.

Cumulative distribution function (CDF) plot is used to determine the probability of a continuous random variable occurs any (measurable) subset of that range. CDF can be used in hydrology to describe the probability of data set. In this study CDF curve of average observed rainfall from all the rainfall stations in the area, raw PRECIS model output rainfall and bias corrected rainfall were generated to check the performance of bias correction method and occurrence probability of the monthly rainfall amount. Figure 5(a) shows the comparison of CDF of bias corrected rainfall, raw PRECIS rainfall with observed rainfall in the present time from 1980 to 2007 in Hat Yai. It can be seen that CDF of bias corrected rainfall is very close to observed rainfall compare with raw PRECIS rainfall meaning that the bias correction method applied in this study shows good performance and can reduce the bias of raw PRECIS rainfall obviously. Furthermore, the occurrence probability of monthly rainfall amount in 1980 to 2009 in Hat Yai can be also seen in the figure. For example, from 1980 to 2007, for observed rainfall, the probability of 300mm rainfall in a month is (1-0.96)*100=4%; for bias corrected PRECIS rainfall is (1-0.95)*100=5%; for raw PRECIS rainfall is (1- 0.90)*100=10%.

Rainfall Change in Hat Yai

In order to combine urban disaster risk management and climate change adaptation practice and develop appropriate solutions to improve the climate resilience of urban communities, it is necessary to understand the future rainfall change and tendency. Therefore, in this study, efforts have been made to investigate the change of rainfall in near future in details and in different time scales. For this purpose, spatial rainfall distribution map were generated in this study. Average observation rainfall distribution map from all rainfall stations in the Songkhla provi- nce (Hat Yai is the central city) were compared with future bias corrected average province rainfall distribution map in the different time scale. The comparison of annual average rainfall distribution map between observed present rainfall and future bias corrected projected rainfall in Hat Yai are shown in following from Figure 4. From observed annual rainfall (left) and the projected annual rainfall (right) maps, it indicates that the average annual rainfall is higher in the future than the present, especially in the central and northern parts of Songkhla Province.

Furthermore, the analysis on the seasonal distribution of rainfall shows that in summer seasons, the rainfall is projected to increase slightly. In rainy season months, analysis has shown that the average future seasonal rainfall in Hat Yai from 2010- 2039 will increase largely in the central and east part. In details, the future average rainfall in Hat Yai will increase obviously from Oct. to Dec. , while from Jan. to Feb. the rainfall in the future in Hat Yai will increasing slightly in the central part of the area. It can be seen that rainfall is projected to decrease in the rainy season, especially in November to January. Major increases of rainfall are projected to happen in the central and north part of the province.

Statistical analysis from created CDF plot (Figure 5(b)) showed that the exceedance probability of 300 mm of monthly rainfall projected for the near future is about 6%. This is higher than the 4% exceedance probability of monthly rainfall according to observed rainfall data.

Furthermore, the rainfall distribution change in each decade in Hat Yai by comparing with 2000 and 2009 are shown in the Figure 6. From it, the change of rainfall distribution in the cities in every decade can be seen clearly.

As for the percentage change of rainfall, in comparison with the baseline which shown in Figure 7, Hat Yai city is projected to have increases in rainfall, more during the decade 2020-2029. In the lower set of maps in the figure, more blue in the maps indicates a greater percentage of increase in rainfall. More red means a greater decrease in rainfall. From this figure, it can be seen that the central parts of the study area has greater percentage of increase in rainfall. It also indicated that the importance of prevention and mitigation of flooding disasters in this flood prone city.

CONCLUSIONS

This study has run the PRECIS RCM model with a horizontal resolution of 25x25 km to downscale the future climate information in the selected flood prone city in Thailand. Different bias correction methods were applied to correct the raw PRECIS output climate data for temperature and rainfall in the study area. It is found that SD ratio method can reduce the bias of raw PRECIS rainfall output and gave the coefficient of determination (R2) and RMSE in acceptable range and keep the changing trend of the original GCM data. From the validation bias corrected methods of precipitation, it is found that the correlation values in both validation periods show the high goodness of fit in the study area. The analysis results have shown that there will be decreasing rainfall in dry seasons and early wet seasons in near future in the study city. However the rainfall will increase in wet season in most of time, which will bring more risk for flood disaster. Furthermore, rainfall pattern and distribution will be also changed in the near future in the study areas. The generated climate change scenarios in this study can play a vital role in prevention and mitigation of meteorological disasters in the flood prone city in Thailand.

ACKNOWLEDGEMENTS

This work is funded by "One Hundred Talents Program" of Chinese Academy of Sciences, National Natural Science Foundation of China (No. 41471460), and the Asia-Pacific Network for Global Change Research (APN) "CAF2014- RR06(ARCP)-NMY-Wang: Integrated Flood Modelling and Pre-disaster Loss Estimation in Asian Countries". The authors are grateful for their financial supports.

References

LITERATURE CITED

Cheng, C. S.; Campbell, M.; Li, Q.; Li, G.; Auld, H.; Day, N. and Yap, D., 2007. A synoptic climatological approach to assess climatic impact on air quality in south-central Canada. Part II: future estimates. Water, air, and soil pollution, 182(1- 4), 117-130.

Coulibaly, P., 2004. Downscaling daily extreme temperatures with genetic programming, Geophysical Research Letters, 31, L16203, 1-4.

Gordon, C.; Cooper C.; Senior, C. A.; Banks, H.; Gregory, J. M.; Johns, T. C.; Mitchell, J. F. B. and Wood, R. A., 2000. The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Clim. Dyn.,16:147-168.

He, B.; Takara, K.; Yamashiki, Y.; Kobayashi, K., and Luo, P., 2011. Statistical Analysis of Present and Future River Water Temperature in Cold Regions Using Downscaled GCMs Data. Disaster Prevention Research Institute Annuals. B, 54(B), 103-111.

IPCC, 2007. Climate Change 2007-Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the IPCC, Cambridge University Press.

Leander, R. and Buishand, T. A., 2007. Resampling of regional climate model output for the simulation of extreme river flows. J. Hydrol., 332, 487-496.

Jones R.G.; Noguer M.; Hassell D.C.; Hudson D.; Wilson S.S.; Jenkins G.J. and Mitchell J.F.B., 2004. Generating high resolution climate change scenarios using PRECIS, Met Office Hadley Centre, Exeter, UK, 40 p.

Mendes, D. and Marengo, J.A., 2010. Temporal downscaling: a comparison between artificial neural network and autocorrelation techniques over the Amazon Basin in present and future climate change scenarios, Theor Appl Climatol, 100, 413-421.

Michel, D. and Pandya, A. (eds.), 2009. Troubled Waters- Climate Change, Hydropolitics, and Transboundary Resources The Henry L. Stimson Center, Washington, DC, USA.Nazrul MI (2009) Rainfall and temperature scenario for Bangladesh. Open Atmos. Sci. J., 3, 93-103.

Koontanakulvong S. and Sriariyawat A., 2011. Thailand flood 2011-causes and future scenario proceedings of USMCA 2011, Chieng Mai, Thailand, October, 2011.

Terink, W.; Hurkmans, R. T. W. L.; Torfs, P. J. J. F. and Uijlenhoet, R. , 2010. Evaluation of a bias correction method applied to downscaled precipitation and temperature reanalysis data for the Rhine basin, Hydrol. Earth Syst. Sci., 14, 687-703, doi:10.5194/hess-14-687- 2010.

UNDP, 2006. Human Development Report 2006-Beyond Scarcity: Power, Poverty and the Global Water Crisis. The United Nations Development Programme (UNDP), New York, USA.

World Water Assessment Programme, 2009. The United Nations World Water Development Report 3: Water in a Changing World. UNESCO, Paris and Earthscan, London.

Zhang, C.; Takase,K.; Oue,H.; Ebisu,N., and Yan H., 2013. Effects of land use change on hydrological cycle from forest to upland field in a catchment, Japan. Frontiers of Structural and Civil Engineering, 7(4), 456-465.