Table 35 A-5.1 Weights for different levels of water deficit
Related Figures (88)
Source: EM-DAT: The OFDA/CRED International Disaster Database; www.em-dat.net - Université Catholique de Louvain - Brussels - Belgium, 2014. Table 1-1 Number of natural disasters between 1900 and 2014. ‘loods and droughts are the two main meteorological catastrophes. Flooding occurs he most (33%), whereas droughts are less frequent (5%- Table 1-1), but cause more atal impacts in terms of the number of the death toll. Combined, the hazards account or 80% of the total number of people affected by natural disasters (Table 1-2). furthermore, they have created enormous environmental, social, and economic mpacts, such as degraded wetland areas, diminished natural biodiversity, destroyed vgricultural land and industrial facilities, and high economic costs. It is expected that ‘limate change effects (i.e., more intense rainfall, increasing seawater levels, etc.) will exacerbate flood occurrences and impacts in the future (Jones, 1999; quoted Lehner, DOI, Alcamo, Henrichs, & Kaspar, 2006). In addition to increased climate variability, oopulation growth and societal interference are the other critical factors that exacerbate adverse consequences (Changnon, Pielke Jr, Changnon, Sylves, & -ulwarty, 2000; Douben, 2006; Kuntiyawichai, 2012; Pielke Jr & Downton, 2000). source: EM-DAT: The OFDA/CRED International Disaster Database; www.em-dat.net - Université Catholique de Louvain - Brussels - Belgium, 2014. able 1-3 Number and percentage (in the parentheses) of victims and damage from flooding between 1900 and 2014. Source: EM-DAT: The OFDA/CRED International Disaster Database; www.em-dat.net - Université Catholique de Louvain - Brussels - Belgium, 2014. Source: EM-DAT: The OFDA/CRED International Disaster Database; www.em-dat.net - Université Catholique de Louvain - Brussels - Belgium, 2014. Table 1-4 Number and percentage (in the parentheses) of victims and damage from droughts between 1900 and 2014. There are considerable spatial and temporal variations of precipitation quantities across the basin. The provinces in the east, i.e., Sisaket, Yasothon, Amnat Charoen and Ubon Ratchathani, receive additional rainwater from tropical depressions from the east. Thus, on average, the eastern provinces collect more rain than the western areas (The Meteorological Department of Thailand, n.d.). Table 2-1 Irrigated rice fields by province in the Mun River Basin. Source: Derived from the land use map obtained from the Land Department, Thailand. Figure 2-3 Rice cultivation area of the year 2013. Provinces are listed from West to East. Source: Land Development Department, Thailand. Source: Department of Groundwater Resources, Thailand. Figure 2-8 Expected well yield (m3/hr.) and TDS (mg/1) of the year 2011-2012. Source: Department of Provincial Administration, Thailand. Source: National Statistical Office, Thailand. 2.9 People depending on agriculture Figure 2-11 People depending on agriculture by province of the year 2016. Source: Office of the Education Council, Thailand. The average years of educational attainment were classified into four classes: < 6 Elementary), 6-9 (Junior High), 9-12 (Senior High) and > 12 and they were subsequently scored as 4, 3, 2 and 1, respectively. This indicates that provinces with high average year of schooling are less susceptible to droughts, whereas the provinces with lower average year of schooling are relatively more vulnerable to drought events Figure 2-12). Figure 2-12 Education level by province of the year 2016. Table 3-1 Dry/wet conditions corresponding with the seven SPEI categories During the months of the rice-growing season, all dry and wet conditions occurred (Table 3-2). Approximately 35% of the total number of months experienced anomalous dry or wet conditions. The proportion of wet months was slightly higher than the dry ones. The number of extreme wet and dry months increased considerably when the time scale increased from 1 month to 3 months (i.e. SPEI-1 to SPEI-3). This shows that extreme events may not always be detected effectively with the 1-month SPEI because impacts accumulate and become more visible over time. Therefore, the SPEI-1 in combination with SPEI-3 provide a more complete description of dry/wet conditions. combination with SPEI-3 provide a more complete description of dry/wet conditions The provinces are presented from west to east according to their geographical locations. The numbers at the top of the table indicate the months (July to November). The climatic events are classified into two main groups: the wetness (blue) and dryness (orange). The colour levels display the degree of conditions ranging from moderate to extreme. Table 3-3 Spatial and temporal variations of climatic characteristics of rice-growing months that are classified based on SPEI-1 values. *** p-value < 0.001, ** p-value < 0.01, * p-value < 0.05, ‘ p-value < 0.1 The red and blue colors represent negative and positive values respectively. Provinces are listed from west to east. Table 3-5 Number of months with positive and negative significant correlations at 90% and 95% significant levels. 3.3.4 Climate impacts on rice yield 3.3.4.1 Regression with climatic variables Table 3-8 R? values of the multiple linear regression models. The regression results for a particular month gives more explanatory power (R?) than for the average over the entire growing season. Those particular months may coincide with crucial crop development stages that directly influence yields, such as, tillering from the second half of August to mid-September and panicle differentiation from mid-September to October. Further, regressions with Tmin and Tmax separately gave more satisfactory outcomes (higher R?) than using the mean temperature (Tave). e 3-3 Estimated impacts of climate trends from the multiple linear regression for 1984-2013 on rice yield (kg/ha/ decade). Only the months with the highest R? for the given provinces are presented. Most other months are statistically insignificant. The orange, red, and blue bars express the yield impacts due to trends in Tmin, Tmax and precipitation, respectively. The grey bars present the total yield impacts due to trends in all climatic variables. The error bars indicate 90% confident interval. 4 Estimated impacts of precipitation and 1-month SPEI trends from simple linear regression of rice yield with precipitation and with 1-month SPEI for 1984-2013 on rice yield (kg/ha/ decade). Only the months with the highest R2 for the given provinces are presented, most other months are statistically insignificant. The blue and pink bars indicate the impacts due to trends in precipitation and 1-month SPEI, respectively. The error bars present 90% confident interval. Rice yield are stronger correlated with the SPEI-1 than with precipitation and SPEI-3 both in statistical significance and value). We conclude that, the 1-month SPEI is better capable of detecting soil moisture deficiency and crop stress in rice than precipitation und SPEI-3. We also examined the correlations between rice yields with 2, 6, 9 and 12- nonth SPEIs (Table 3-9) and the results show inferior to those between the yields and SPEI-1 and SPEI-3. The correlations with the SPEI-1 follows a similar pattern to those Nith precipitation (mostly positively correlated, except in September). In some orovinces, the yield decreases with increases of precipitation in the month September, sven though the monthly precipitation is lower than crop water requirement. This is Jue to extremes in precipitation intensity, which according to the precipitation records used in this study) could be as high as 250 mm per day. he trends of all climatic variables and SPEI-1 are predominantly increasing. In the fun River basin, Tmax is rising faster than Tmin, unlike in China (Zhou et al., 2004) nd India (Padma Kumari, Londhe, Daniel, & Jadhav, 2007), the first and second rgest rice producers, where Tmin has been increasing faster than Tmax. The upward ends in Tmin and Tmax are relatively pronounced in October and November and are lore apparent in provinces in the East than in the West. The records also show that in yme years, daily Tmax in October in some provinces exceeded 35 °C. If the observed end continues in the future, the negative effect of Tmax on rice yield is likely to be ore visible. The increasing trends in precipitation and SPEI-1 varied considerably. legative impacts of precipitation on yields mostly occur in September, which is sually the wettest month. In absence of proper drainage, larger volumes of rainwater ill likely negatively impact rice development. Table 4-1 Data description and their sources used in SWAT, rainfall frequency analysis, and HEC-RAS. Figure 4-1 Summarized process of the study fable 4-2 Fourteen SWAT model parameters, methods, initial values, and their calibrated ranges and values. The model performance and efficiency were subsequently assessed using three indicators: coefficient of determination (R2), Coefficient o f efficiency (or Nash-Sutcliffe (NS) Nash and Sutcliffe (1970)), and the percent bias (PBIAS). The R* and NS are the most widely used statistic documented for calibration and validation (Arnold, Moriasi, et al., 2012). R? defines how well the simulated values explain the observed dispersion. Its range varies from 0 to 1, with 0 indicating no correla tion at all and 1 representing the perfect correlation between simulated and observed values (Krause, Boyle, & Base, 2005). Typically, for values of R* 2 0.5, the simulated results are considered The stations are listed by from west to east. The stations are listed by from west to east. We also developed a SWAT model incorporating the dam operation; however, the results are not significantly different from the one without the dams, particularly at the outlets after station M.184 further downstream. There are three possible reasons for this. The first is that the outflows (in m3/s) from the reservoirs used in the SWAT model were derived from the total daily flow (in million m°) reported by the Royal Irrigation Department of Thailand (2018) due to lack of accurate flow-rate data. The calculation assumed equal release of water during the day. As a result, some mismatches in amounts and timing are observed. The second reason is that the areas along the course of the river, and its tributaries downstream of the dams, are highly utilized for agriculture, due to which slight differences in the flow may not be detected. The third reason could be that after station M.184, no influence from the dams is observed on the flow of the main river. The model surprisingly well simulated the flow dynamics at M.164 (R? = 0.76, NS = 0.74 and PBIAS = 15.9% for calibration and R? = 0.52, NS = 0.43 and PBIAS = 46.2% for validation) and M.185 (R? = 0.76, NS = 0.61 and PBIAS = 36.1% for calibration and R2 = 0.89, NS = 0.89 and PBIAS = -4.9% for validation). This is in spite of the fact that the stations are situated slightly downstream of the Lam Takhong Dam, and Lam Nang Rong Dam, respectively. of the Lam Takhong Dam, and Lam Nang Rong Dam, respectively. Table 4-5 Manning’s n value along the Mun River Figure 4-6 Flood map for 10 (a), 25 (b), 50 (c), and 100-year (d) return periods The information offers a better understanding of flood extent and characteristics, such as water depths and flow velocities in flood-prone areas, which is the first step towards flood risk assessment and management. It is useful and essential information for stakeholders and authorities in contemplating management strategies and policies for flood damage reduction. Such policies and strategies i nclude land use plans, floodplain building regulations, emergency preparedness and evacuation plans, as well as the development of structural mitigation measures to cope with flooding such as reservoirs and levee projects (Sanders, 2007; Sarhadi et al., 2012). For example, in the case of the Mun River Basin, floodplain management and regulations regarding agricultural, industrial and urban expansion should be developed and enforced, particularly at the upstream and the central parts of the basin, while levees are needed for protecting the exposed cities. The height of levees for d ifferent scenarios can be identified from the flood maps. This provides useful information for the decision- making process in relation to flood protection standards, w. hich often depend on the growth and yields. Many cities, for example, Phibun Mangsahan, Mueang, Warin Chamrap districts in Ubon Ratchathani province, Rasi Salai district in Si Sa Ket province, Rattanaburi, Tha Tum, and Chumphon Buri districts in Surin province, and Sateuk district in Buri Ram province are affected. These cities are situated in flood- prone areas which have been preferred for settlement since ancient times, owing to favorable geographic environments which facilitate economic growth, such as easy accessibility and nutrient-rich soils (Douben, 2006). The areas have a high degree of urbanization with high values of assets and infrastructure, thus flooding in this area can cause significant damage. Table 5-1 Weights for different levels of water deficit Finally, the drought hazard index is calculated by integrating the product of the weight of water deficit at each level (Wj) and its relevant frequency (Fj). The minimum value of DHI is 0, meaning there is enough rainwater for rice cultivation every year (no water deficit). The maximum amount is 4 = 4 x 1, meaning there is insufficient water for rice every year. The equation takes the form relevant rating. The double rating steps (to both severity and frequency) thus reduce the sensitivity of the DHI. Table 5-2 List of data used in drought risk analysis (with year and source) Assigning weights to the different factors and indicators are the key to developing a composite vulnerability index. There is no generally accepted weighting for factors and indicators: Wilhelmi and Wilhite (2002) assigned the weights based on the relative contribution of each factor to overall vulnerability, Chen, Cutter, Emrich, and Shi (2013) applied a statistic-based method (Principal Component Analysis: PCA) to evaluate the weights of the factors in social vulnerability assessment, and Yuan et al. (2015) assigned the weights based on expert opinions and employed accelerating genetic algorithm-based analytic hierarchy process (AGA-AHP) to quantify the opinion. In our study, as described above when we obtained the data, some data were already classified into different classes, thus we followed Wilhelmi and Wilhite (2002), because their method is widely accepted in vulnerability assessment where factors contain several classes (S. Pandey et al., 2012; Safavi, Esfahani, & Zamani, 2014; Shahid & Behrawan, 2008; J. Wu et al., 2011). Consequently, each class of all factors was given a weight between 0 and 4 (0 represents non-affected, 1 represents least significant and 4 represents the most significant). For example, soil types with less than 150 mm water- holding capacity are ranked 4, showing that they are less able to withstand a deficiency of precipitation than ones with higher water-holding capacity. Dummy weights were assigned to some classes for masking purposes, and they were excluded from the analysis (Table 5-3 and Table 5-4). Table 5-3 Weighting scheme for drought exposure analysis Table 5-4 Weighting scheme for drought vulnerability analysis We investigated statistical correlations between the factors in exposure and vulnerability indices to verify whether a multicollinearity problem exists by using Kendall’s Tau correlation coefficient (Tp). This measures associations between ranked data and the results range from -1 and 1. The correlation values were then examined Figure 5-5 presents drought exposure. People and rice fields in the central part of the basin are more exposed to drought than those in other parts. High (27%) and moderate (31%) levels of exposures are mostly observed at the central part of the basin due to nigh population density, and rice production, whereas low (36%) and very low (6%) 2xposures are mostly found in Nakhon Ratchasima and the eastern region. In terms of physical susceptibility, the upper and the western parts of the basin are more susceptible to droughts than the lower and the eastern ones (Figure 5-6a). About 13% of the total rice cultivation area is highly susceptible to water scarcity, mostly around the borders between upper and lower provinces and in the upper part of Nakhon Ratchasima. 34% and 41% of the area have moderate and low susceptibility respectively, with patches scattered over the basin. 12% of the area has very low susceptibility, mostly in the lower and eastern parts. however, in some plots, susceptibility rises to moderate and high because of saline soils and saline groundwater with low expected yield. The eastern and lower parts are not affected by salt, and has a high yield and good quality groundwater, although the soil water-holding capacities in some plots are low. This leads to low and very low susceptibility in these areas. Crop insurance, a mitigation measure that compensates farmers for crop losses due to natural disasters (Wilhelmi & Wilhite, 2002), is another crucial factor that can reduce drought vulnerability and risk. However, it has not been successful in Thailand. A pilot crop insurance program for rice was initiated in Thailand in 2009. However, as of 2014, only 1369 farmers were participating in the program, and the total area of rice land insured was only 45.3 km2. Also, the collected insurance premiums were less than the indemnity payments (Sinha & Tripathi, 2016). Conclusion: No issue of multicollinearity detected Table A- 5.2 Correlation matrix of exposed rice Conclusion: No issue of multicollinearity detected. Table A- 5.3 Correlation matrix of physical susceptibility Table A- 5.4 Correlation matrix of social susceptibility. Table 6-1 Summary of all water development projects in the Mun River Basin, 2016. [The medium-scale projects being implemented and complemented the large-scal schemes have storage capacity between 2 and 100 million m°, or water surface area < [5 km2, or command area between 480 and 12,800 ha, and, lastly, the small-scal orojects are ones with construction period less than 1 year and no land compensatio1 schemes embedded into development plans. Both are single-purpose project orimarily for domestic water use or irrigation, and very few projects for fores -onservation. T yf the RID, and he vast majority of medium-scale projects are under the responsibilit: the rest are under the Department of Water Resources (Hydro anc Agro Informatics Institute, 2012), while about 16 government agencies involve i small-scale proj ects (Patamatamkul, 2001). The important features of medium- anc small-scale projects are present in Table 6-2. Figure 6-1 Water development projects at large, medium, and small scales, including electric pump, 2016; Source: The Royal Irrigation Department, Thailand. [able 6-3 Proposed and ongoing projects under the RID at the Mun River Basin, 2016. 6.4 Flood and drought hazards in the Mun River Basin Table 6-4 Flooded area, flooded rice field, and flooding volume of different scenarios. Table 6-5 entails water supply, water needs for rice, and net water requirements of the wet-season rice at all growing months (July-November), and over the growing season. The amount of annual precipitation sufficiently fulfills the rice water needs for both varieties, but that of over the growing season does not. A little more rainwater is demanded for the KDML105 than the RD6. The maximum amount of water is required in July to saturate the soil for land preparation. Thus, droughts in this month do not harm rice but delay the onset of the growing season. The most severe water shortage appears in October in which rice is most vulnerable because two crucial stages: flowering and grain-filling, develop in this month. Both stages need much water to maintain the water layer throughout the month; otherwise, it will result in considerable yield reduction (Bouman, Lampayan, & Tuong, 2007; Fukai, Basnayake, & Cooper, 2000). able 6-5 Water supply, water needs, and net water requirements for the KDML105, RD6 over the entire basin. 6.5 Discussion on existing flood and drought mitigation projects Figure 6-3 The governance structure of water resources development in Thailand. resources development in Thailand. It is mandated to facilitate, advise, monitor, and evaluate the implementation of all water-related agencies country-wide (Figure 6-3) (Office of the National Water Resources, 2018a). Afterwards, the first National Water Resource Act, B.E. 2561 was enacted in December 2018 (Office of the National Water Resources, 2018b) upon which the 20-year water resources management master plan (2018-2037) drafted by the ONWR and other related agencies, was approved by the government in June 2019 (National News Bureau of Thailand, 2019). It is hoped that this reform and policy formulation will bring about the unity in water resources management in Thailand, and can effectively and sustainably tackle the chronic flood, drought, and wastewater problems in the country. 5.7.1 Improvement of existing measures The construction of designed distributed canals and systems, repair or rehabilitatior of the deteriorated elements and schemes should be accomplished. Moreover, in som: areas, the distribution canals may need to be reconsidered because of the salinit problem. The lessons from the irrigation scheme in Sisaket province document that th irrigation canals are used every 3-4 years to supplement the wet-season crops durins dry years, and some canals are left unused (Shannon, 2005). This knowledge anc experiences of local farmers should thus take into account together with the financia support from the central government and technical assistance from local governmen agencies. Ine of the main justification for the promotion of the investment in water-related projects in northeast Thailand was dry-season rice agriculture; however, the large- 5 cale adoption of dry-season rice farming never happened, less than 5% across the asin (Floch et al., 2007). From our point of view, it is difficult for the large-scale dry- = > eason rice cultivation to be possible if other factors, ie., cropping pattern, field rrigation practices, etc. remain unchanged. It is because the total precipitation over >) he wet rice-growing season over the past 30 years have never met the rice water needs. Vioreover, although the annual rainfall did at some rain-gauge stations, particularly at he stations on the east part of the basin, it is still not sufficient for both wet- and dry- eason rice agriculture (Figure 6-5). Furthermore, many farmers are reluctant to plant he second rice because of low yields, fewer profits due to expenses of electricity (for rrigation pumps), chemical fertilizer, and pesticide. Also, the farmers in many areas do not use irrigated water because it is highly saline, and that significantly reduce yields (Kamkongsak & Law, 2001; Shannon, 2005). Figure 6-6 Components of reconnecting oxbow lakes.
