Effects of ENSO and Climate Change on Reference Evapotranspiration in Southern Vietnam

ENSO和气候变化对越南南部参考作物蒸发量的影响

+ Author Affiliations + Find other works by these authors

PDF

  • This study aims to evaluate the effects of climate change and El Niño–Southern Oscillation (ENSO) on reference evapotranspiration (ETo) in the climate sub-regions of South Vietnam (SVN) and the role of the related meteorologi-cal variables. The trend analysis shows that ETo has been increasing quite clearly, especially during the rainy season. By examining the contribution of meteorological variables to the increase of ETo, it is indicated that temperature is the main contributor. Among the meteorological variables that are related to ETo, temperature is also the factor that has gained the most significant increase. Analysis of variance reveals that there is no difference in the increase of ETo among the climate sub-regions of SVN. The effect of ENSO on ETo is assessed based on the difference in the mean value of ETo between the El Niño and La Niña phases. The results show that this difference is most obvious from October to May, and the main factor contributing to the increase is not temperature but sunshine hours. The difference in sunshine hours between the warm and cold phases in these months is around one hour per day, contributing about 58%–86% to the ETo difference. Further analysis of variance shows that ENSO has different levels of influences on ETo in the climate sub-regions. Compared to the increase in ETo due to climate change over the past 40 years, the ETo difference between El Niño and La Niña phases is many times higher. In addition, since the effect of ENSO on ETo is most obvious in the study area during the dry season, it is much stronger. In order to mitigate the effect of ENSO on drought in this area, monitoring and forecasting meteorological variables that have the main contribution to the variation of ETo, including the number of sunny hours, should be promoted.
    本文旨在评估气候变化和厄尔尼诺–南方涛动(ENSO)对越南南部区域参考作物蒸发量(ETo)的影响及相关气象因素的作用。趋势分析表明,ETo在越南南部不同的气候分区中均不断增长,特别是雨季。分析气象因子对ETo增长的贡献发现,温度的贡献最大,同时在所有气象因子中,温度也表现出最显著的增长。另外,比较了厄尔尼诺(ENSO暖位相)和拉尼娜(ENSO冷位相)阶段的ETo平均值,发现十月至次年五月其差别最大,并且贡献最大的不是温度而是日照时数。日照时数在ENSO暖位相和冷位相可相差每天约1小时,由此引起ETo差值的58%–86%。进一步分析发现ENSO对不同气候分区ETo的影响不同。与过去40年温度变化引起的ETo改变相比,ENSO冷暖位相导致的ETo改变高出了很多倍。由于ENSO对ETo的影响发生在研究区域的干季,因此这种影响更加强烈。为减轻ENSO加剧研究区域的干旱灾害,应大力提倡对影响ETo变化的气象因子(比如日照时数)进行监测和预报。
  • 加载中
  • Fig. 1.  Locations of the weather stations and topography of the climate sub-regions in SVN.

    Fig. 2.  Map of ETo trends in SVN in the period 1977−2018.

    Fig. 3.  (a) The mean value of ΔETo in stations during 1977−2018, (b) t-test on the average value of ETo between the El Niño and La Niña years, and (c) the average value of ETo. In (a) and (c), the error bars show standard errors of the mean corresponding to the 0.05 significance level.

    Fig. 4.  The mean value of ΔETo by stations corresponding to the El Niño and La Niña events, which had medium intensity or higher for the period 1977−2018. The error bars show standard errors of the mean corresponding to the 0.05 significance level.

    Fig. 5.  Correlation coefficients between ΔETo and (a) ΔT, (b) ΔH, (c) Δn, and (d) Δu10 in the weather stations.

    Fig. 6.  (a) Mean value of ΔTo during 1977−2018 and (b) influence of ENSO on ETo by temperature factor. The error bars show standard errors of the mean corresponding to the 0.05 significance level.

    Fig. 7.  Contribution proportion of T, H, n, and u10 to ΔETo.

    Fig. 8.  (a) Mean value of ΔH during 1977−2018 and (b) influence of ENSO on ΔETo by relative humidity. Error bars show standard errors of the mean corresponding to the 0.05 significance level.

    Fig. 9.  (a) Statistical values of sunshine hours, (b) mean value of Δn during 1977−2018, and (c) effect of ENSO on ΔETo by sunshine hours. In (a), the horizontal line within the box indicates the median. The upper and lower edges of the box represent the 75% and 25% percentiles, respectively. The upper and lower ends of the whiskers represent the maximum and minimum values. In (b, c), the error bars show the mean standard errors.

    Fig. 10.  The ratio of ΔETo by climate sub-regions to ΔETo average in the SVN when considering only the variation of sunshine hours. The vertical axis shows the ratio of ΔETo by climate sub-regions to ΔETo average in the SVN. Results are categorized by three climate sub-regions S1, S2, and S3.

    Fig. 11.  (a) Mean value of Δu10 during 1977−2018 and (b) effect of ENSO on ΔETo by wind speed at 10 m. The error bars show the mean standard errors corresponding to the 0.05 significance level.

    Table 1.  Results of determining ETo trend at the weather stations during 1977–2018

    Station IDSZSlope (mm yr−1)Cl (%)Station IDSZSlope (mm yr−1)Cl (%)
    1−69−0.740.26 21−7−0.070.18
    2−281−3.04−1.2899222192.361.4295
    31902.051.6195233894.212.48 99.9
    4−43−0.460.07 241081.160.2270
    5470.500.55 25−21−0.220.04
    6−77−0.82−0.26 26−123−1.32−0.4480
    790.090.07 27−138−1.49−0.2280
    81401.511.178028−101−1.08−0.3770
    91651.781.209029−95−1.02−0.26
    102983.222.2399302342.532.1298
    111291.390.998031270.280.33
    12−39−0.41−0.04 322562.761.9799
    131181.270.7370331591.711.3190
    143573.862.45 99.9341461.571.1780
    15320.340.00 352953.191.7299
    16600.640.62 36−207−2.23−1.5395
    171071.150.697037900.970.62
    18−3−0.020.22 382182.351.5795
    19810.870.51 391091.170.9570
    20−148−1.59−1.318040−28−0.29−0.26
    Download: Download as CSV

    Table 2.  Results of the ETo trend among climate sub-regions

    Source of variationSSDFMSFP-valueFcrit
    Between groups 0.072191 2 0.0360950.0348090.9658213.251924
    Within groups38.36716371.03695
    Total38.4393539
    Note: SS, sums of squares; DF, degrees of freedom; and MS, mean square.
    Download: Download as CSV

    Table 3.  Trend of ETo in the rainy and dry seasons for the period 1977−2018

    SeasonNumber of stations showing a
    clear trend with α = 0.01
    Sen’s slope (mm yr−1)
    MeanError
    Dry season15−0.540.39
    Rainy season25 1.290.31
    Download: Download as CSV

    Table 4.  Result of determining temperature trend at the weather stations in the period 1977−2018

    Station IDSZSlope (°C yr−1)Cl (%)Station IDSZSlope (°C yr−1)Cl (%)
    13894.210.02399.9212853.080.01299
    23133.380.01799.8224514.880.02299.9
    34795.180.02699.9235726.190.03399.9
    43313.580.01499.9243093.340.00299.8
    52682.890.01199255205.630.02599.9
    63183.440.01699.8265796.260.03599.9
    74635.010.02199.9274574.940.02599.9
    83533.820.01899.9283724.020.01499.9
    94915.310.02499.9293263.520.01299.8
    104064.390.02099.9305545.990.02699.9
    114725.110.01999.9314835.220.02199.9
    123974.290.01699.9323673.970.01299.9
    133333.600.01799.9333724.020.01399.9
    145315.740.03299.9345626.080.02699.9
    153443.720.01899.9351731.860.00790
    163433.710.01999.9363393.660.01499.9
    174094.420.02399.9373533.820.01499.9
    184735.120.02799.9384224.560.02299.9
    195806.280.03499.9395395.830.02999.9
    201391.50−0.00580403193.450.01299.8
    Download: Download as CSV

    Table 5.  The role of meteorological variables in the ETo trend

    Meteorological
    variable
    ETo trend caused by the changes in each meteorological variable (mm yr−1) Contribution (%)
    S1S2S3S1S2S3Mean
    T0.590.590.62103.2103.293.498.8
    H0.060.030.1510.85.822.714.9
    n−0.03−0.07−0.08−5.7−11.5−11.6−9.8
    u10−0.050.02−0.03−8.32.6−4.4−3.8
    Download: Download as CSV

    Table 6.  Trend of H, n, and u10 during 1977−2018

    Meteorological variableNumber of stations showing
    a clear trend with α = 0.01
    Average trend
    H14−0.012 % yr−1
    n 3−0.0022 h yr−1
    u1015−0.0003 m s−1 yr−1
    Download: Download as CSV

    Table 7.  Occurrence frequency of El Niño and La Niña phases during 1977−2018 and the standard deviation of SST in the Niño3.4 region

    Month123456789101112
    Occurrence frequency (%)Warm phase10.99.66.85.77.07.06.56.58.710.010.910.9
    Cold phase9.88.47.95.67.06.57.98.48.49.310.310.3
    S (°C)1.040.880.700.580.570.580.660.750.870.981.081.10
    Download: Download as CSV

    Table 8.  Analysis of variance and t-test on the ETo trends by climate sub-regions

    Anova
    Source of variationSSDFMSFP-valueFcrit
    Between groups0.04612320.0230616.9490450.002743.251924
    Within groups0.122790370.003319
    Total0.16891239
    t-test
    Sub-region pairDFttcrit
    S1S2200.7211.725
    S1S3281.7251.687
    S2S3261.6731.689
    Download: Download as CSV
  • [1]

    Allen, R. G., L. S. Pereira, D. Raes, et al., 1998: Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements, FAO Irrigation and Drainage Paper 56. Food and Agric, Organ, Rome, Italy.
    [2]

    Cao, L. G., and Z. C. Zhou, 2019: Variations of the reference evapotranspiration and aridity index over Northeast China: Changing properties and possible causes. Adv. Meteor., 7692871. doi: 10.1155/2019/7692871.
    [3]

    Dadaser-Celik, F., E. Cengiz, and O. Guzel, 2016: Trends in reference evapotranspiration in Turkey: 1975-2006. Int. J. Climatol., 36, 1733–1743. doi: 10.1002/joc.4455.
    [4]

    Darshana, A. Pandey, and R. P. Pandey, 2013: Analysing trends in reference evapotranspiration and weather variables in the Tons River Basin in Central India. Stoch. Environ. Res. Risk Assess., 27, 1407–1421. doi: 10.1007/s00477-012-0677-7.
    [5]

    Du, C. Y., J. J. Yu, P. Wang, et al., 2016: Reference evapotranspiration changes: Sensitivities to and contributions of meteorological factors in the Heihe River Basin of Northwestern China (1961–2014). Adv. Meteor., 4143580. doi: 10.1155/2016/4143580.
    [6]

    Espadafor, M., I. J. Lorite, P. Gavilán, et al., 2011: An analysis of the tendency of reference evapotranspiration estimates and other climate variables during the last 45 years in Southern Spain. Agric. Water Manage., 98, 1045–1061. doi: 10.1016/j.agwat.2011.01.015.
    [7]

    Halpert, M. S., and C. F. Ropelewski, 1992: Surface temperature patterns associated with the Southern Oscillation. J. Climate, 5, 577–593. doi: 10.1175/1520-0442(1992)005<0577:STPAWT>2.0.CO;2.
    [8]

    Heydari, M. M., R. Aghamajidi, G. Beygipoor, et al., 2014: Comparison and evaluation of 38 equations for estimating reference evapotranspiration in an arid region. Fresen. Environ. Bull., 23, 1985–1996.
    [9]

    Hidalgo, H. G., D. R. Cayan, and M. D. Dettinger, 2005: Sources of variability of evapotranspiration in California. J. Hydrometeorol., 6, 3–19. doi: 10.1175/JHM-398.1.
    [10]

    Irmak, S., I. Kabenge, K. E. Skaggs, et al., 2012: Trend and magnitude of changes in climate variables and reference evapotranspiration over 116-yr period in the Platte River Basin, central Nebraska-USA. J. Hydrol., 420-421, 228–244. doi: 10.1016/j.jhydrol.2011.12.006.
    [11]

    Ju, J. H., and J. Slingo, 1995: The Asian summer monsoon and ENSO. Quart. J. Roy. Meteor. Soc., 121, 1133–1168. doi: 10.1002/qj.49712152509.
    [12]

    Kendall, M. G., 1975: Rank Correlation Methods. Charles Griffin, London, 272 pp.
    [13]

    Kousari, M. R., M. A. A. Zarch, H. Ahani, et al., 2013: A survey of temporal and spatial reference crop evapotranspiration trends in Iran from 1960 to 2005. Climatic Change, 120, 277–298. doi: 10.1007/s10584-013-0821-5.
    [14]

    Lam, H. C. Y., A. Haines, G. McGregor, et al., 2019: Time-series study of associations between rates of people affected by disasters and the El Niño Southern Oscillation (ENSO) Cycle. Int. J. Environ. Res. Public Health, 16, 3146. doi: 10.3390/ijerph16173146.
    [15]

    Li, B., F. Chen, and H. D. Guo, 2015: Regional complexity in trends of potential evapotranspiration and its driving factors in the Upper Mekong River Basin. Quater. Int., 380–381, 83–94. doi: 10.1016/j.quaint.2014.12.052.
    [16]

    Liu, X. M., and D. Zhang, 2013: Trend analysis of reference evapotranspiration in Northwest China: The roles of changing wind speed and surface air temperature. Hydrol. Process., 27, 3941–3948. doi: 10.1002/hyp.9527.
    [17]

    Luong, V. V., 2021: Development of a new ENSO index to assess the effects of ENSO on temperature over southern Vietnam. Theor. Appl. Climatol., 144, 1119–1129. doi: 10.1007/s00704-021-03591-3.
    [18]

    Mann, H. B., 1945: Nonparametric tests against trend. Econometrica, 13, 245–259. doi: 10.2307/1907187.
    [19]

    Martens, B., W. Waegeman, W. A. Dorigo, et al., 2018: Terrestrial evaporation response to modes of climate variability. npj Climate Atmos. Sci., 1, 43. doi: 10.1038/s41612-018-0053-5.
    [20]

    Meza, F. J., 2005: Variability of reference evapotranspiration and water demands. Association to ENSO in the Maipo River Basin, Chile. Global Planet. Change, 47, 212–220. doi: 10.1016/j.gloplacha.2004.10.013.
    [21]

    Miralles, D. G., M. J. Van Den Berg, J. H. Gash, et al., 2014: El Niño–La Niña cycle and recent trends in continental evaporation. Nat. Climate Change, 4, 122–126. doi: 10.1038/nclimate2068.
    [22]

    Möller, M., J. Tanny, Y. Li, et al., 2004: Measuring and predicting evapotranspiration in an insect-proof screenhouse. Agric. For. Meteor., 127, 35–51. doi: 10.1016/j.agrformet.2004.08.002.
    [23]

    Nguyen, D.-Q., J. Renwick, and J. McGregor, 2014: Variations of surface temperature and rainfall in Vietnam from 1971 to 2010. Int. J. Climatol., 34, 249–264. doi: 10.1002/joc.3684.
    [24]

    Nouri, M., M. Homaee, and M. Bannayan., 2018: Spatiotemporal reference evapotranspiration changes in humid and semi-arid regions of Iran: Past trends and future projections. Theor. Appl. Climatol., 133, 361–375. doi: 10.1007/s00704-017-2176-8.
    [25]

    Phan, V. T., T. Ngo-Duc, and T.-M.-H. Ho, 2009: Seasonal and interannual variations of surface climate elements over Vietnam. Climate Res., 40, 49–60. doi: 10.3354/cr00824.
    [26]

    Raghavan, S. V., M. T. Vu, and S. Y. Liong, 2016: Regional climate simulations over Vietnam using the WRF model. Theor. Appl. Climatol., 126, 161–182. doi: 10.1007/s00704-015-1557-0.
    [27]

    Ropelewski, C. F., and M. S. Halpert, 1996: Quantifying Southern Oscillation–precipitation relationships. J. Climate, 9, 1043–1059. doi: 10.1175/1520-0442(1996)009<1043:QSOPR>2.0.CO;2.
    [28]

    Sabziparvar, A. A., S. H. Mirmasoudi, H. Tabari, et al., 2011: ENSO teleconnection impacts on reference evapotranspiration variability in some warm climates of Iran. Int. J. Climatol., 31, 1710–1723. doi: 10.1002/joc.2187.
    [29]

    Singh, R. M., and P. Shukla, 2020: Drought characterization using drought indices and El Niño effects. Natl. Acad. Sci. Lett., 43, 339–342. doi: 10.1007/s40009-019-00870-6.
    [30]

    Stojanovic, M., M. L. R. Liberato, R. Sorí, et al., 2020: Trends and extremes of drought episodes in Vietnam sub-regions during 1980–2017 at different timescales. Water, 12, 813. doi: 10.3390/w12030813.
    [31]

    Talaee, P. H., B. S. Some’e, and S. S. Ardakani, 2014: Time trend and change point of reference evapotranspiration over Iran. Theor. Appl. Climatol., 116, 639–647. doi: 10.1007/s00704-013-0978-x.
    [32]

    Valipour, M., 2015: Importance of solar radiation, temperature, relative humidity, and wind speed for calculation of reference evapotranspiration. Arch. Agron. Soil Sci., 61, 239–255. doi: 10.1080/03650340.2014.925107.
    [33]

    Vu-Thanh, H., T. Ngo-Duc, and T. Phan-Van, 2014: Evolution of meteorological drought characteristics in Vietnam during the 1961–2007 period. Theor. Appl. Climatol., 118, 367–375. doi: 10.1007/s00704-013-1073-z.
    [34]

    Wang, X.-M., H.-J. Liu, L.-W. Zhang, et al., 2014: Climate change trend and its effects on reference evapotranspiration at Linhe station, Hetao irrigation district. Water Sci. Eng., 7, 250–266. doi: 10.3882/j.issn.1674-2370.2014.03.002.
    [35]

    Webster, P. J., and S. Yang, 1992: Monsoon and ENSO: Selectively interactive systems. Quart. J. Roy. Meteor. Soc., 118, 877–926. doi: 10.1002/qj.49711850705.
    [36]

    Yassen, A. N., W.-H. Nam, and E.-M. Hong, 2020: Impact of climate change on reference evapotranspiration in Egypt. CATENA, 194, 104711. doi: 10.1016/j.catena.2020.104711.
    [37]

    Yu, L.-L., Z.-Q. Xia, J.-K. Li, et al., 2013: Climate change characteristics of Amur River. Water Sci. Eng., 6, 131–144. doi: 10.3882/j.issn.1674-2370.2013.02.002.
    [38]

    Zhang, D., X. M. Liu, and H. Y. Hong, 2013: Assessing the effect of climate change on reference evapotranspiration in China. Stoch. Environ. Res. Risk Assess., 27, 1871–1881. doi: 10.1007/s00477-013-0723-0.
    [39]

    Zhang, X. Y., S. Y. Chen, H. Y. Sun, et al., 2011: Changes in evapotranspiration over irrigated winter wheat and maize in North China Plain over three decades. Agric. Water Manage., 98, 1097–1104. doi: 10.1016/j.agwat.2011.02.003.
    [40]

    Zhang, Y. Q., J. L. Peña-Arancibia, T. R. McVicar, et al., 2016: Multi-decadal trends in global terrestrial evapotranspiration and its components. Sci. Rep., 6, 19124. doi: 10.1038/srep19124.
    [41]

    Zhou, W., and J. C. L. Chan, 2007: ENSO and the South China Sea summer monsoon onset. Int. J. Climatol., 27, 157–167. doi: 10.1002/joc.1380.
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Effects of ENSO and Climate Change on Reference Evapotranspiration in Southern Vietnam

    Corresponding author: Van Viet LUONG, vietvanluong@gmail.com
  • Industrial University of Ho Chi Minh City, Ho Chi Minh City 71408, Vietnam

Abstract: This study aims to evaluate the effects of climate change and El Niño–Southern Oscillation (ENSO) on reference evapotranspiration (ETo) in the climate sub-regions of South Vietnam (SVN) and the role of the related meteorologi-cal variables. The trend analysis shows that ETo has been increasing quite clearly, especially during the rainy season. By examining the contribution of meteorological variables to the increase of ETo, it is indicated that temperature is the main contributor. Among the meteorological variables that are related to ETo, temperature is also the factor that has gained the most significant increase. Analysis of variance reveals that there is no difference in the increase of ETo among the climate sub-regions of SVN. The effect of ENSO on ETo is assessed based on the difference in the mean value of ETo between the El Niño and La Niña phases. The results show that this difference is most obvious from October to May, and the main factor contributing to the increase is not temperature but sunshine hours. The difference in sunshine hours between the warm and cold phases in these months is around one hour per day, contributing about 58%–86% to the ETo difference. Further analysis of variance shows that ENSO has different levels of influences on ETo in the climate sub-regions. Compared to the increase in ETo due to climate change over the past 40 years, the ETo difference between El Niño and La Niña phases is many times higher. In addition, since the effect of ENSO on ETo is most obvious in the study area during the dry season, it is much stronger. In order to mitigate the effect of ENSO on drought in this area, monitoring and forecasting meteorological variables that have the main contribution to the variation of ETo, including the number of sunny hours, should be promoted.

ENSO和气候变化对越南南部参考作物蒸发量的影响

本文旨在评估气候变化和厄尔尼诺–南方涛动(ENSO)对越南南部区域参考作物蒸发量(ETo)的影响及相关气象因素的作用。趋势分析表明,ETo在越南南部不同的气候分区中均不断增长,特别是雨季。分析气象因子对ETo增长的贡献发现,温度的贡献最大,同时在所有气象因子中,温度也表现出最显著的增长。另外,比较了厄尔尼诺(ENSO暖位相)和拉尼娜(ENSO冷位相)阶段的ETo平均值,发现十月至次年五月其差别最大,并且贡献最大的不是温度而是日照时数。日照时数在ENSO暖位相和冷位相可相差每天约1小时,由此引起ETo差值的58%–86%。进一步分析发现ENSO对不同气候分区ETo的影响不同。与过去40年温度变化引起的ETo改变相比,ENSO冷暖位相导致的ETo改变高出了很多倍。由于ENSO对ETo的影响发生在研究区域的干季,因此这种影响更加强烈。为减轻ENSO加剧研究区域的干旱灾害,应大力提倡对影响ETo变化的气象因子(比如日照时数)进行监测和预报。
    • Vietnam is located in the tropical monsoon region, where the climate is clearly divided into two seasons, namely the dry and rainy seasons. Due to the complex topography and the differences in rainfall and temperature, Vietnam is classified into seven climate sub-regions (Phan et al., 2009; Nguyen et al., 2014; Raghavan et al., 2016). South Vietnam (SVN), which is the selected study area, consists of three sub-regions: (1) southern central Coast (S1), (2) central highlands (S2), and (3) southern area (S3) including the southeast and the lower Mekong Delta (Fig. 1). During the dry season, especially from January to April, a great amount of evaporation caused by high temperature, low humidity, and lack of rainfall often leads to severe drought in this area (Vu-Thanh et al., 2014; Stojanovic et al., 2020). SVN is also considered to be heavily affected by climate change as well as ENSO activities (Nguyen et al., 2014; Luong, 2021).

      Figure 1.  Locations of the weather stations and topography of the climate sub-regions in SVN.

      Evapotranspiration (ETo) is used for a variety of purposes such as combining with crop coefficients (Kc) for the calculation of irrigation requirements and water resource management or combining with precipitation for the assessment of drought levels (Allen et al., 1998). It is also one of the factors used for evaluating the effects of climate change (Darshana et al., 2013; Zhang et al., 2013; Yassen et al., 2020).

      The increase in temperature and the changes in humidity, sunshine hours, and wind speed caused by climate change are changing ETo and affecting water balance (Zhang et al., 2011; Yu et al., 2013). Many studies have been done to assess the impact of climate change on ETo. Their results showed that the magnitude and trend of the rise or fall of the ETo varied from place to place (Li et al., 2015; Yassen et al., 2020).

      Darshana et al. (2013) reported a significant downward trend of ETo in the Tons basin in India, which is from −1.75 to −8.98 mm yr−1. According to Irmak et al. (2012), the reduction of ETo in the Platte River basin in the US is 0.36 mm yr−1. In Iran, Egypt, and the upper Mekong, while ETo in some regions tended to increase, in others it tended to decrease (Kousari et al., 2013; Talaee et al., 2014; Li et al., 2015; Yassen et al., 2020). In contrast, Espadafor et al. (2011) showed that ETo tends to increase in the south of Spain. The trend of ETo also depends during the period. As reported by Zhang et al. (2013), the average ETo of the whole of China decreased by 1.44 mm yr−1 between 1960 and 1992 but increased by 2.24 mm yr−1 from 1993 to 2011.

      In addition to the studies of the trend of ETo, there are also several studies that aimed at analyzing the contribution of meteorological variables to the change of ETo. The results also showed that the contribution of meteorological variables varied from place to place. According to Irmak et al. (2012), the reduction of ETo in the Platte River basin was attributed to the radiation reduction, which was caused by the increase in clouds. Li et al. (2015) indicated that the increase or decrease of sunshine hours was the main cause of the rise or fall of ETo in some areas in the upper Mekong. Nouri et al. (2018) and Liu and Zhang (2013) also revealed that ETo change was mainly caused by the increase in air temperature and decrease in wind speed. Wang et al. (2014) showed that due to the increase in air temperature and the decrease in humidity in the Hetao area of China, ETo increased slightly. The remaining meteorological variables, i.e., sunshine hours and wind speed, had a rather small change and did not affect the ETo change. According to Espadafor et al. (2011), in southern Spain, ETo increased by 3.5 mm yr−1 because the air temperature and solar radiation rose and the relative humidity fell. Dadaser-Celik et al. (2016) evaluated the ETo trend during 1975−2006 in Turkey, and their results showed that the increase in air temperature and the decreases in wind speed and relative humidity led to the increase of ETo.

      The ETo formula can be divided into two parts: the radiative component and the aerodynamic component. Usually, the sensitivity of meteorological variables is evaluated based on the change of these two components when the value of the meteorological variable changes. Möller et al. (2004) and Wang et al. (2014) evaluated the sensitivity and role of each meteorological variable in the ETo trend by dividing the data by periods and conducting the following steps: (1) calculating the value of the ETo for all meteorological variables taken in the first period, (2) repeating step 1 but using the value of the meteorological variable whose sensitivity need to be assessed taken at the last period, and (3) comparing the difference in average ETo value between the two steps above. This study also applied a similar approach with some minor adjustments, which were shown in the methodology section.

      The effect of ENSO on climate variability is known through the analysis of Halpert and Ropelewski (1992) and Ropelewski and Halpert (1996). These analyzes gave that the effects of El Niño and La Niña depended on the place and time of the year. Lam et al. (2019) indicated that the effects of natural disasters such as droughts and floods were often related to the operation of ENSO. The effects of El Niño and La Niña on the climate are most evident in the tropics and subtropics, especially in the monsoon areas (Webster and Yang, 1992; Ju and Slingo, 1995; Zhou and Chan, 2007). In Vietnam, according to Stojanovic et al. (2020), the sub-regions of SVN are more sensitive to El Niño than the northern sub-regions.

      There were many studies about the effect of ENSO on evaporation on a regional scale (Hidalgo et al., 2005; Meza, 2005; Sabziparvar et al., 2011) as well as on a global scale (Miralles et al., 2014; Zhang et al., 2016; Martens et al., 2018). According to Martens et al. (2018), ENSO and other large-scale fluctuations affected the weather and evaporation dynamics on the continents, in which ENSO had a clear influence on ETo in the tropics. Many research results showed that there was a significant difference in ETo between El Niño and La Niña years, especially in winter and spring (Meza, 2005; Sabziparvar et al., 2011), and the correlation coefficient between ENSO monitoring indices and ETo was statistically significant (Cao and Zhou, 2019; Singh and Shukla, 2020).

      Due to the important role of ETo, its variability and trend are evaluated by many researchers as mentioned above. In addition, the determination of meteorological variables that have major contributions to ETo’s trend and variability also attracted many researchers. El Niño was the cause of severe droughts in SVN (Luong, 2021). Furthermore, some drought indicators in this region have increased (Stojanovic et al., 2020). Therefore, it is necessary to assess the impact of ENSO and climate change on ETo, which is related to water requirements for crops. The assessment of the impacts of ENSO and climate change on ETo in this study will be done separately. In addition, the study will also determine the contribution of meteorological variables to the change of ETo.

    2.   Data and methodology
    • The main studied data are meteorological observations in the South of Vietnam and the Oceanic Niño Index (ONI). The location of the weather stations, the boundaries of the climate sub-regions, and the study area topography are shown in Fig. 1. For brevity, this research used the station ID to identify the station. There are a total of 40 weather stations included in the analysis, consisting of 12 stations (IDs 1–12) in the sub-region S1, 10 stations (IDs 13–22) in the S2, and 18 stations (IDs 23–40) in the S3. The data are collected in the period of 1977−2018 (42-yr). This period is chosen because the meteorological monitoring data of the stations in SVN are quite sufficient. The used data consist of the weather variables relating to the calculation of ETo, such as temperature, wind, relative humidity, and sunshine hours. The meteorological variables in these weather stations are observed 8 times per day or extracted from the records. These stations are managed by the regional hydrometeorological centers and belong to the hydrometeorological monitoring network of the Vietnam National Hydro-Meteorology Service. The selected stations have negligible missing data, accounting for only 1.4%, and filled by multivariate linear regression. By t and F tests at the 0.05 significance level, it shows that the regression equations are reliable.

      ONI is an ENSO index developed by the United States Climate Prediction Center (CPC) and the International Research Institute for Climate and Society (IRI) based on the 3-month moving average of sea surface temperature (SST) in the Niño 3.4 region. For 5 consecutive months, if the ONI is equal to or higher than +0.5°C, an El Niño event is determined. In the contrast, when it is equal to or lower than −0.5°C, a La Niña event is defined. ONI data are collected from the website at https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/.

    • The main contents of this study include the calculation of ETo value by different conditions of meteorological variables, the analysis of the trend and variability of ETo, and the assessment of the role of related meteorological variables. The methods used are detailed below.

    • ETo can be calculated from meteorological data. The major methods of calculating ETo include FAO Penman-Monteith, Priestley-Taylor, Hargreaves, Makkink, Blaney-Criddle, and Samani-Hargreaves (Valipour, 2015). Therein, the FAO Penman-Monteith performs better than the others due to its clear physical significance (Allen et al., 1998; Heydari et al., 2014; Du et al., 2016).

      According to the results of an expert consultation, the FAO Penman-Monteith is considered as the best method for ETo calculation (Allen et al., 1998). This method requires data on radiation, air temperature, air humidity, and wind speed. Therefore, in this study, the FAO Penman-Monteith method is used to calculate ETo. The ETo formula is as follows (Allen et al., 1998):

      $${\rm{ET}}_{\rm{o}} = \frac{{0.48\varDelta ({R_{\rm{n}}} - G) + \gamma \dfrac{{900}}{{T + 273}}{u_2}({e_{\rm{s}}} - {e_{\rm{a}}})}}{{\varDelta + \gamma (1 + 0.3{u_2})}},$$ (1)

      where Rn (MJ m−2 day−1) is the net radiation at the crop surface, G (MJ m−2 day−1) is the soil heat flux density, T (°C) is the mean daily air temperature at 2-m height, u2 (m s−1) is the wind speed at 2-m height, es (kPa) is the saturation vapor pressure, ea (kPa) is actual vapor pressure, (kPa °C−1) is the slope vapor pressure curve, and γ (kPa °C−1) is the psychrometric constant.

      In Eq. (1), Δ, es, and ea are given by:

      $$\hspace{40pt} \varDelta = \frac{{4098{e_{\rm{s}}}}}{{{{(T + 273)}^2}}},$$ (2)
      $$\hspace{40pt} {e_{\rm{s}}} = 0.6108\exp \left({\frac{{17.27T}}{{T + 273}}} \right),$$ (3)
      $$\hspace{40pt} {e_{\rm{a}}} = {e_{\rm{s}}}\frac{H}{{100}}.$$ (4)

      In Eq. (4), H (%) is the relative humidity. The net radiation is the difference between the incoming net shortwave radiation (Rns) and the outgoing net longwave radiation (Rnl)

      $$\hspace{40pt} {{R}_{\rm{n}}}{{ = }}{{R}_{\rm{ns}}}{{ - }}{{R}_{\rm{nl}}},$$ (5)

      with Rns is given by:

      $$\hspace{40pt} {R_{\rm{ns}}} = 0.77\left({{a_{\rm{s}}} + {b_{\rm{s}}}\frac{n}{N}} \right){R_{\rm{a}}}.$$ (6)

      In Eq. (6), n (h) is the actual duration of sunshine, N (h) is the maximum possible duration of sunshine or daylight hours, Ra (MJ m−2 day−1) is the extraterrestrial radiation, and as and bs are regression constant. The default values for as and bs are 0.25 and 0.50 (Allen et al., 1998), respectively. For the study area, these coefficients are calculated based on n and Rn data using linear regression method.

      In Eq. (6), Ra and N are calculated as follows:

      $$\hspace{40pt} {R_{\rm{a}}} = 37.6{d_{\rm{r}}}\left( {{W_{\rm{s}}} {\rm{sin}}\psi {\rm{sin}}\delta + {\rm{cos}}\psi {\rm{sin}}{W_{\rm{s}}}} \right), $$ (7)
      $$\hspace{40pt} N = 7.64 {W_{\rm{s}}}, $$ (8)

      with

      $$\hspace{40pt} \begin{aligned}[b] &{W_{\rm{s}}} = {\rm{arccos}}\left( { - {\rm{tan}}\psi {\rm{tan}}\delta } \right),\\ &\delta = 0.409 {\rm{sin}}(0.0172J - 1.39),\\ &{d_{\rm{r}}} = 1 + 0.033 {\rm{cos}}\left( {0.0172J} \right). \end{aligned} $$ (9)

      In Eqs. (7)–(9), ψ (rad) is the geographical latitude, dr is the inverse relative distance between earth and sun, δ (rad) is the solar declination, and J is the Julian day.

      In Eq. (5), Rnl is given by:

      $$\hspace{40pt} \begin{aligned} & {R_{\rm nl}} = 118{\left({T + 273} \right)^4}\times{10^{ - 9}}\,\,\,\,\\ & \cdot \frac{{\left({0.34 - 0.044\sqrt {{e_{\rm{a}}}} } \right) \left({0.1 + 0.9\dfrac{n}{N}} \right)}}{{59.7 - 0.055T}}. \end{aligned}$$ (10)

      In Eq. (1), the soil heat flux density is given by:

      $$\hspace{40pt} G= 0.38({T_i} - {T_{i - 1}}), $$ (11)

      where Ti and Ti1 are the air temperature on days i and i − 1. If calculating G according to monthly average temperature, then:

      $$\hspace{40pt} G = 0.14({T_m} - {T_{m - 1}}), $$ (12)

      where Tm and Tm−1 are the average temperature in the m and m − 1 months.

      In Eq. (1), the psychrometric constant is given by:

      $$\hspace{40pt} \gamma = 0.00163\frac{P}{{{{2501 - 2361}}{{.10 - 3T}}}}, $$ (13)

      where P (kPa) is the atmospheric pressure.

      The wind speed at 2 m (u2) is calculated from the wind speed at 10 m (u10) as follows:

      $$\hspace{40pt} {u_2} = 0.77{u_{10}}. $$ (14)
    • To identify trends of a chronological data, Mann–Kendall (MK) non-parametric test (Mann, 1945; Kendall, 1975) was used. Compared to the linear trend method, this method can avoid false trend caused by some extreme values. With an ordered time series data (x1, x2,..., xn) where xi represents the data at time i, the MK test statistic is given by:

      $$\hspace{40pt} S = \sum\limits_{k = 1}^{n - 1} {\sum\limits_{j = k + 1}^n {\rm{sign}} } \left( {{x_j} - {x_k}} \right), $$ (15)

      where

      $$\hspace{40pt} {\rm{sign}} \left( {{x_j} - {x_k}} \right) = \left\{ {\begin{array}{*{20}{c}} {\;1\quad\;\;{{\rm if}}\;{x_j} - {x_k} > 0}\\ {\;0\quad\;\;{{\rm if}}\;{x_j} - {x_k} = 0}\\ { - 1\quad{{\rm if}}\;{x_j} - {x_k} < 0} \end{array}} \right., $$ (16)

      and

      $$\hspace{40pt} Z = \left\{ {\begin{array}{*{20}{l}} {\dfrac{{S - 1}}{{\sqrt {{\rm{Var}}(S)} }}\,\,\,\quad\quad{\rm{if}}\,S > 0} \\ {0\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\quad\quad\quad\quad{\rm{if}}\,S = 0} \\ {\dfrac{{S + 1}}{{\sqrt {{\rm{Var}}(S)} }}\,\,\,\quad\quad{\rm{if}}\,S < 0} \end{array}} \right.,$$ (17)

      where Var(S) is the variance of S, calculated by the following equation:

      $$\hspace{20pt} \begin{aligned} &{\rm{Var}}(S) = \frac{1}{{18}}[n(n - 1)(2n + 5) \\ & + \sum\limits_{p = 1}^g {{t_p}(1 - {t_p})(2{t_p} + 5)]}. \end{aligned}$$ (18)

      where tp is the number of data points in the pth tied group, and g is the number of tied groups in the dataset.

      In the test, the null hypothesis is rejected when the value of Z is less than the critical value of Z obtained from the normal distribution table.

      Nonparametric estimate of the slope is called Sen’s slope. Sen’s slope, β, is defined as follows:

      $$\hspace{40pt} \beta = {\rm{median}} \left({\frac{{{x_k} - {x_i}}}{{k - i}}} \right) \quad \quad \forall i < k, $$ (19)

      where xk and xi are two values of x, i = 1, 2, …, n − 1. In this study, the trend values are assessed at the 0.1 significance level.

    • In the analysis of the change of ETo caused by climate change or activities of ENSO, the role of a meteorological variable is represented by its contribution. In this study, the contribution of each meteorological variable to the change of ETo is calculated based on the ratio of the value of the change of ETo caused by this variable to the change caused by all variables.

      To assess the contribution of meteorological variables to the change and variability of ETo, this study uses an approach similar to that of Möller et al. (2004) and Wang et al. (2014), except that the data are not divided into periods.

      Accordingly, in this study, the determination of the contribution of a meteorological variable to the trend or variability of ETo is done by the following steps:

      (1) Determining the trend or the difference of ETo between the warm and cold phases, called ΔETo.

      (2) Replacing the value of the remaining variables by the corresponding monthly average value of the period 1977–2018, and recalculating ETo. Then, determining the trend or the difference of ETo between the warm and cold phases, called ΔET'o.

      (3) Calculating the contribution of the considered variable by the ratio ΔET'o/ΔETo.

    • The analysis of variance is used to test the difference in ETo trend among climate sub-regions of the SVN. This method is also used to test the difference of ETo between the warm and cold phases of each climate sub-region. The t-test is used to evaluate the reliability of the difference in mean ETo values between ENSO phases.

    3.   Results and Discussion
    • Based on the observed data of n and Rn, as and bs for the study area receive values of 0.21 and 0.41, respectively. From these coefficients, ETo is calculated. The results of determining ETo trends at the weather stations are presented in Table 1. There are 13/40 stations, which have a relatively clear trend of ETo at the 90% confidence level (CI). The majority of these 13 stations tend to increase, except that only 2 stations have a downward trend. The mean value of ETo increased for the stations is 0.59 ± 0.32 mm yr−1 at the 5% significance level, corresponding to a rise of about 11–36 mm over the past 40 years. It can be seen that this increase is not high. Since the proportion of stations with ETo increased and the average increase is not high, it can be concluded that the effects of climate change on the annual average ETo are not clear.

      Station IDSZSlope (mm yr−1)Cl (%)Station IDSZSlope (mm yr−1)Cl (%)
      1−69−0.740.26 21−7−0.070.18
      2−281−3.04−1.2899222192.361.4295
      31902.051.6195233894.212.48 99.9
      4−43−0.460.07 241081.160.2270
      5470.500.55 25−21−0.220.04
      6−77−0.82−0.26 26−123−1.32−0.4480
      790.090.07 27−138−1.49−0.2280
      81401.511.178028−101−1.08−0.3770
      91651.781.209029−95−1.02−0.26
      102983.222.2399302342.532.1298
      111291.390.998031270.280.33
      12−39−0.41−0.04 322562.761.9799
      131181.270.7370331591.711.3190
      143573.862.45 99.9341461.571.1780
      15320.340.00 352953.191.7299
      16600.640.62 36−207−2.23−1.5395
      171071.150.697037900.970.62
      18−3−0.020.22 382182.351.5795
      19810.870.51 391091.170.9570
      20−148−1.59−1.318040−28−0.29−0.26

      Table 1.  Results of determining ETo trend at the weather stations during 1977–2018

      Considering the three climate sub-regions (S1, S2, and S3), the trend of ETo gains the mean values of 0.55, 0.55, and 0.64 mm yr−1, respectively. The results in Table 2 reveal that the F value is lower than the critical value of F (Fcrit) corresponding to the 0.05 significance level; in other words, there is no difference in the mean value among climate sub-regions. Compared to the average values of ETo in the S1, S2, and S3 sub-regions, which are 1540, 1380, and 1570 mm yr−1, respectively, ETo has increased approximately 1.5%–1.6% in the past 40 years.

      Source of variationSSDFMSFP-valueFcrit
      Between groups 0.072191 2 0.0360950.0348090.9658213.251924
      Within groups38.36716371.03695
      Total38.4393539
      Note: SS, sums of squares; DF, degrees of freedom; and MS, mean square.

      Table 2.  Results of the ETo trend among climate sub-regions

      Based on the results of the determination of the trend at the weather stations, Fig. 2 shows the trend distribution of ETo in SVN. To reduce the unevenness and uncertainty of the ETo trend, the local polynomial interpolation is used. According to Fig. 2, ETo tends to increase, but the increase is not uniform. Some places experience an increase of more than 1 mm yr−1, meanwhile the change is not clear in other places. This also occurs in the S1 and S3 sub-regions separately. However, the magnitude of the difference was negligible.

      Figure 2.  Map of ETo trends in SVN in the period 1977−2018.

    • The dry season in SVN takes place in 5 months, from December to the end of April, and the rest of the year is the rainy season. The results of the ETo trend determination in the dry and rainy seasons are summarized in Table 3. In this table, the average trend of ETo by stations is estimated at the 0.05 significance level.

      SeasonNumber of stations showing a
      clear trend with α = 0.01
      Sen’s slope (mm yr−1)
      MeanError
      Dry season15−0.540.39
      Rainy season25 1.290.31

      Table 3.  Trend of ETo in the rainy and dry seasons for the period 1977−2018

      The results show that the trends of ETo between the rainy and dry seasons are opposite. While ETo of the rainy season tends to increase, that of the dry season tends to decrease. In the dry season, the average value of the trend of ETo by the stations is −0.54 ± 0.39 mm yr−1. This value is 1.29 ± 0.31 mm yr−1 in the rainy season. The error values in Table 3 also show that the trend of ETo in the dry season is uneven between stations when compared with the one in the rainy season.

      In the dry season, the number of stations in which ETo shows a clear trend is not much, corresponding to only 15/40 stations. In contrast, during the rainy season, this number is higher, corresponding to 25/40 stations. In conjunction with the rate of change of ETo in the rainy season, which is much higher than that in the dry season, it can be seen that the trend of ETo in the rainy season is clearer than the one in the dry season. In the dry season, the rainfall in southern Vietnam only accounts for about 10% of the annual rainfall, the water shortage is often quite serious. A decline in ETo during the dry season can be a good sign for agriculture.

    • The results of the temperature trend determination at the weather stations are presented in Table 4. According to this table, most of the stations show a clear trend. At the 0.01 significance level, the temperature at 38/40 weather stations tends to increase. Thus, the manifestation of climate change in temperature in SVN is clear. At the 5% significance level, the mean value of the temperature increased in SVN is 0.019 ± 0.003°C yr−1. Of the 40 stations, there are 12 stations with an increase of over 0.022°C yr−1. These stations are all located in areas with rapid urbanization rate. The highest increase is in Tan Son Hoa station of Ho Chi Minh City (ID = 26) with an increase of 0.035°C yr−1. This is the largest city and has the highest urbanization rate in Vietnam. Urbanization has increased the ability to absorb and trap solar radiation, increased surface roughness, reduced moisture in the soil, and caused urban heat island effect. Due to the characteristics of high humidity and often cloudiness at night, temperatures in this area in the evening are still high; these make average temperature in the urban areas higher than in the rural areas. If these 12 stations are not included, the average temperature increase is about 0.015 ± 0.004°C yr−1, corresponding to an increase of 0.6 ± 0.17°C in the past 40 years.

      Station IDSZSlope (°C yr−1)Cl (%)Station IDSZSlope (°C yr−1)Cl (%)
      13894.210.02399.9212853.080.01299
      23133.380.01799.8224514.880.02299.9
      34795.180.02699.9235726.190.03399.9
      43313.580.01499.9243093.340.00299.8
      52682.890.01199255205.630.02599.9
      63183.440.01699.8265796.260.03599.9
      74635.010.02199.9274574.940.02599.9
      83533.820.01899.9283724.020.01499.9
      94915.310.02499.9293263.520.01299.8
      104064.390.02099.9305545.990.02699.9
      114725.110.01999.9314835.220.02199.9
      123974.290.01699.9323673.970.01299.9
      133333.600.01799.9333724.020.01399.9
      145315.740.03299.9345626.080.02699.9
      153443.720.01899.9351731.860.00790
      163433.710.01999.9363393.660.01499.9
      174094.420.02399.9373533.820.01499.9
      184735.120.02799.9384224.560.02299.9
      195806.280.03499.9395395.830.02999.9
      201391.50−0.00580403193.450.01299.8

      Table 4.  Result of determining temperature trend at the weather stations in the period 1977−2018

      To assess the contribution of temperature increase to ETo trend, the following analysis evaluated the change in ETo when replacing the values of the relative humidity, sunshine hours, and wind speed by the corresponding monthly mean values. The statistical results are shown in Table 5, in which a rise in temperature leads to a rise of ETo, corresponding to an average increase of about 0.6 mm yr−1 and a contribution of about 98.8% to the increase. Thus, among the factors related to ETo, the increase in temperature is the main contributor to the increase of ETo. The results also show that there is no significant difference in the contribution of temperature to the increase of ETo among sub-regions.

      Meteorological
      variable
      ETo trend caused by the changes in each meteorological variable (mm yr−1) Contribution (%)
      S1S2S3S1S2S3Mean
      T0.590.590.62103.2103.293.498.8
      H0.060.030.1510.85.822.714.9
      n−0.03−0.07−0.08−5.7−11.5−11.6−9.8
      u10−0.050.02−0.03−8.32.6−4.4−3.8

      Table 5.  The role of meteorological variables in the ETo trend

    • The results of determining the trends of relative humidity, sunshine hours, and wind speed are summarized in Table 6. According to this table, their trend is not clear, only 3−15 out of 40 stations show a significant trend at the 0.01 significance level.

      Meteorological variableNumber of stations showing
      a clear trend with α = 0.01
      Average trend
      H14−0.012 % yr−1
      n 3−0.0022 h yr−1
      u1015−0.0003 m s−1 yr−1

      Table 6.  Trend of H, n, and u10 during 1977−2018

      The average relative humidity trend in SVN is −0.012% yr−1. Relative humidity decreases quite clearly in the S3 sub-region with 0.017% yr−1, and the other two sub-regions yield a negligible decrease. The decrease in relative humidity causes ETo to increase. According to Table 5, the fall in relative humidity contributes to the rise of ETo in sub-regions S2, S1, and S3 by 5.8%, 10.8%, and 22.7%, respectively, and the average increase is 14.9%.

      Sunshine hours do not reveal a clear trend. Only 3/40 analyzed stations show a trend at the 0.01 significance level. During 1977−2018 in SVN, the number of sunshine hours decreases by about 0.0022 h yr−1, which contributes about −10% to the change of ETo (Table 5).

      According to Table 6, there is no significant change in wind speed, with an average decrease of only 0.0003 m s−1 yr−1, corresponding to 0.01 m s−1 over the past 40 years. This reduction is only within the permissible error in the wind speed measurement, and it is not significant. Besides, Table 5 shows that the contribution of wind speed to the trend of ETo is negligible. On average for stations, it contributes −3.8% to the ETo’s raise. At the two stations where ETo tends to decrease at the 90% confidence level in Table 1, there are the decreases in the number of sunshine hours and the wind speed.

      Thus, climate change led to a rise in ETo in SVN over the past 40 years, of which the main contribution was from the increase in temperature. The decrease in humidity also caused ETo to rise, and therefore, it was the additional factor contributing to the rise of ETo. In addition, the slight decrease of the sunshine duration also contributed to change ETo in the opposite direction.

    • The influence of ENSO on ETo was assessed by the difference between the warm phase (El Niño) and cold phase (La Niña), and it was denoted by ΔETo. Based on the ENSO phases determined according to the ONI, the average value of ΔETo for the stations is shown in Fig. 3a. Due to the influence of ENSO, ΔETo increased from October to May with an average increase from 0.22 ± 0.03 mm day−1 to 0.48 ± 0.07 mm day−1. In contrast, from June to September, ΔETo slightly decreased. When comparing the mean value of ETo in the El Niño and La Niña years by the t-test at the 0.05 significance level (Fig. 3b), the results show that from June to September, the t values are all less than the critical t values. Thus, the difference of ΔETo in these months is not statistically significant.

      Figure 3.  (a) The mean value of ΔETo in stations during 1977−2018, (b) t-test on the average value of ETo between the El Niño and La Niña years, and (c) the average value of ETo. In (a) and (c), the error bars show standard errors of the mean corresponding to the 0.05 significance level.

      April is the month with the highest ΔETo (Fig. 3a), and it is also the last month of the dry season in SVN. As a result of very low rainfall and very high ETo (Fig. 3c), the water imbalance is often quite severe. In this month, severe droughts often occur in all three climate sub-regions; thus, the level of drought would be more severe in the El Niño years. The average value of ETo in April is 5.19 ± 0.13 mm day−1. Compared with this value, in the El Niño years, ETo is higher than that in the La Niña years by about 9.2 ± 1.1%.

      On average from October to May, ΔETo gains a value of 0.34 ± 0.02 mm day−1, corresponding to about 124 ± 7.3 mm yr−1. In comparison to the trends caused by climate change, the influence of ENSO on ETo is stronger. Thus, the combination of both El Niño and climate change will make the drought in these months even more severe.

      The values of ΔETo in Fig. 3a are for all of the El Niño and La Niña events. If we only consider El Niño and La Niña events that had moderate or higher intensity, the value of ΔETo increases significantly from March to May (Fig. 4). In comparison to Fig. 3a, the value of ΔETo in these months is 1.5−1.9 times higher.

      Figure 4.  The mean value of ΔETo by stations corresponding to the El Niño and La Niña events, which had medium intensity or higher for the period 1977−2018. The error bars show standard errors of the mean corresponding to the 0.05 significance level.

      The influence of ENSO on ETo of this region was only evident in a few months. During the mid-rainy months, from June to September, the low value of ΔETo may be since El Niño or La Niña episodes usually start in fall and end in spring. Statistical results on the frequency of occurrence of the warm and cold phases of ENSO are presented in Table 7. According to this table, El Niño and La Niña are less active in the period from April to July. The standard deviation (S) of SST in the Niño 3.4 region during this period is also small. According to Nguyen et al. (2014) and Van Viet (2021), the meteorological variables in SVN fluctuate later than SST in the Niño 3.4 region about 3 months, so this issue could lead to ETo in SVN to fluctuate later than SST in the Niño 3.4 region. The lag time between SST in the Niño 3.4 region and ETo in SVN is about a few months, which could be the reasons that ETo in SVN from June to September is less influenced by El Niño and La Niña.

      Month123456789101112
      Occurrence frequency (%)Warm phase10.99.66.85.77.07.06.56.58.710.010.910.9
      Cold phase9.88.47.95.67.06.57.98.48.49.310.310.3
      S (°C)1.040.880.700.580.570.580.660.750.870.981.081.10

      Table 7.  Occurrence frequency of El Niño and La Niña phases during 1977−2018 and the standard deviation of SST in the Niño3.4 region

      From October to May, the mean ΔETo in the S1, S2 and S3 sub-regions were 0.26, 0.28, and 0.34 mm day−1, respectively. The results of t-test and analysis of variance on these mean values between regions at the 0.05 significance level are shown in Table 8. The results of analysis of variance of ΔETo by sub-regions show that F value is greater than Fcrit, in other words, there are at least two sub-regions in which the mean values of ΔETo are different. According to the t-test results, there is no difference in the mean values of ΔETo between sub-regions S2 and S1 and between S2 and S3; however, there is a difference between sub-regions S3 and S1. Thus, according to t-test, ΔETo in the S3 sub-region is higher than that of the S1 sub-region with the 0.05 significance. This result indicates that the area of SVN is only about 163,000 km2 but it shows the difference in the influence of ENSO on ETo between the sub-regions. The influence of ENSO on ETo is most obvious in the S3 sub-region. This may be because this area is located at the lowest latitude among the three regions, from 8.5o to 12oN, as well as along the coast and strongly influenced by monsoons due to its flat topography.

      Anova
      Source of variationSSDFMSFP-valueFcrit
      Between groups0.04612320.0230616.9490450.002743.251924
      Within groups0.122790370.003319
      Total0.16891239
      t-test
      Sub-region pairDFttcrit
      S1S2200.7211.725
      S1S3281.7251.687
      S2S3261.6731.689

      Table 8.  Analysis of variance and t-test on the ETo trends by climate sub-regions

    • ΔETo had a significant change from October to May. The variables ΔT, ΔH, Δn, and Δu10 denote the average difference of T, n, H, and u10 between the warm and cold phases, respectively. Figure 5 shows the mean values of ΔT, ΔH, Δn, Δu10, and ΔETo by the weather stations and their correlation coefficients. These correlation coefficients are all positive except for the relationship between ΔH and ΔETo. This is consistent with the mechanism of evaporation. The values of the correlation coefficient with ΔETo decrease in the order from Δn, ΔH, ΔT to Δu10. The tests at the 0.01 significance level show that the correlation coefficient between ΔETo and Δu10 is less than that of the critical value of Pearson’s. This reveals that only Δn, ΔH, and ΔT have a clear relationship with ΔETo, of which the sunshine duration reveals the most obvious influence of ENSO on ETo.

      Figure 5.  Correlation coefficients between ΔETo and (a) ΔT, (b) ΔH, (c) Δn, and (d) Δu10 in the weather stations.

    • To evaluate the influence of ENSO on ETo by temperature, the following analysis is based on the ETo value, which is calculated when replacing the relative humidity, sunshine hours, and wind speed data by their corresponding monthly mean value. In this way, the values of ΔETo and their corresponding ΔT are shown in Fig. 6. It can be seen in Fig. 6a, from November to May, the mean ΔT ranges from 0.2 ± 0.12 to 0.84 ± 0.10°C, and the highest value occurs in April and May. In the months from June to August and October, the mean ΔT is negligible. In September, the mean value of ΔT is −0.23 ± 0.08°C. Because ΔT is only evident in April and May, ΔETo shows a similar trend in these two months, corresponding to the mean values being 0.09 ± 0.015 and 0.08 ± 0.01 mm day−1 (Fig. 6b).

      Figure 6.  (a) Mean value of ΔTo during 1977−2018 and (b) influence of ENSO on ETo by temperature factor. The error bars show standard errors of the mean corresponding to the 0.05 significance level.

      During the period from November to May, in SVN, because the temperature in the warm phase was higher than in the cold phase, ΔETo increased. However, when comparing with the increase in ΔETo caused by all meteorological variables, the temperature contributed only from about 6% to 20%, and the average was 12% (Fig. 7). The insignificant contribution of temperature to the value of ΔETo can be explained by the fact that the temperature difference between the warm and cold phases is not much.

      Figure 7.  Contribution proportion of T, H, n, and u10 to ΔETo.

    • Similar to temperature, in order to assess the effect of ENSO on ETo caused by the relative humidity variable, the following analysis was based on the ETo value, which was calculated when replacing the temperature, sunshine hours, and wind speed data by their corresponding monthly mean value. In this way, the values of ΔETo and their corresponding ΔH are presented in Fig. 8. Figure 8a revealed that, on average for SVN, except for February and the months from June to September in which ΔH was negligible, ΔH was quite high in the rest months, ranging from −1.3 ± 0.3% to −3.9 ± 0.6%. Thus, in comparison to the El Niño years, the La Niña years had higher humidity. With these values of ΔH, the corresponding values for ΔETo in these months ranged from 0.03 ± 0.01 to 0.1 ± 0.02 mm day−1 (Fig. 8b). When calculating on the total change of ETo, the share of H in these months was from 5% to 20% with an average of 14% (Fig. 7). In comparison to T, H contributed more to ΔETo.

      Figure 8.  (a) Mean value of ΔH during 1977−2018 and (b) influence of ENSO on ΔETo by relative humidity. Error bars show standard errors of the mean corresponding to the 0.05 significance level.

    • The latitude of SVN is from 8.5oN to 16oN, therefore the sun passes through the zenith of this area in April and August. During the period from March to September, the sun’s altitude is highest. Solar altitude angle at midday is usually above 70o. March and April have the highest sunshine hours because they are the last months of the dry season, with little cloudiness and long daytime (Fig. 9a). In contrast, in August and September, despite having a long daytime, but due to cloudy weather, they are the months which have the lowest sunshine hours.

      Figure 9.  (a) Statistical values of sunshine hours, (b) mean value of Δn during 1977−2018, and (c) effect of ENSO on ΔETo by sunshine hours. In (a), the horizontal line within the box indicates the median. The upper and lower edges of the box represent the 75% and 25% percentiles, respectively. The upper and lower ends of the whiskers represent the maximum and minimum values. In (b, c), the error bars show the mean standard errors.

      When replacing the data of temperature, relative humidity, and wind speed by their corresponding monthly mean value, the values of ΔETo and their corresponding ΔH are shown in Fig. 9. It can be seen in Fig. 9b, from October to May, the effect of ENSO on sunshine hours is very clear. The mean values of Δn are from 0.77 ± 0.12 to 1.58 ± 0.09 h day−1. From June to August, Δn decreased slightly. In September, Δn is approximately zero. The significant increase in Δn in the period from October to May causes ΔETo to increase with an average value from 0.18 ± 0.02 to 0.32 ± 0.02 mm day−1 (Fig. 9c). January and December are months had large Δn but their ΔETo was not large. It is because these months have the lowest solar altitude, and the solar altitude angle at midday is usually less than 60°. When calculating on the total change of ΔETo, the share of sunshine duration in these months ranges from 58% to 86% with an average of 74% (Fig. 7). Sunshine hours contribute more to ΔETo than temperature and humidity.

      When only considering the role of sunshine hours, the ratio of ΔETo by sub-regions to the average ΔETo in SVN is shown in Fig. 10. The mean values of this ratio for the sub-regions S3, S1, and S2 are 1.2, 0.93, and 0.88, respectively. These values show that among the three sub-regions, S3 has a clearer change of ETo between the warm and cold phases caused by the change in sunshine hours. In addition, the result also indicates that there is a quite large difference in the influence of ENSO on ETo in April and May between the sub-regions 1 and 3. This difference may be due to the location and topography of these climate sub-regions.

      Figure 10.  The ratio of ΔETo by climate sub-regions to ΔETo average in the SVN when considering only the variation of sunshine hours. The vertical axis shows the ratio of ΔETo by climate sub-regions to ΔETo average in the SVN. Results are categorized by three climate sub-regions S1, S2, and S3.

    • The difference in wind speed at 10-m height between the warm and cold phases is shown Fig. 11a. When replacing the data of temperature, humidity, and sunshine hours by their corresponding monthly mean value, the ΔETo determination result is shown in Fig. 11b. Figure11a shows that the change in wind speed is not clear and have not the common shape as that of T, H, and n. Due to the small change in wind speed, ΔETo does not change significantly, of which the highest ΔETo is taken place in April with its mean value of 0.016 ± 0.011 mm day−1 (Fig. 11b). The contribution of the change of wind speed between the warm and cold phases to ΔETo is shown in Fig. 7. Accordingly, it can be concluded that the influence of ENSO on ETo caused by the change in wind speed is insignificant.

      Figure 11.  (a) Mean value of Δu10 during 1977−2018 and (b) effect of ENSO on ΔETo by wind speed at 10 m. The error bars show the mean standard errors corresponding to the 0.05 significance level.

      Based on the assessment results of the difference of these factors between the warm and cold phases, it is clear that ENSO has a significant influence on climate as well as on ETo in SVN. This result also indicates that the change in sunshine hours contributes mainly to the change of ETo between El Niño and La Niña years.

      By the analysis of Pearson correlation coefficients between ΔT, ΔH, Δn, Δu10 and ΔETo as well as their contribution to ΔETo, these results are quite consistent. The variables that have a good correlation with ΔETo are also the main contributors to the change rate of ΔETo. In other words, when analyzing the effects of ENSO on ETo in SVN, the sensitivity of the meteorological variables increases in the order of wind speed, temperature, humidity, and sunshine hours.

    4.   Conclusions
    • By the Mann−Kendall non-parametric test and the Sen’s slope, the results show that ETo tends to increase in the rainy season and decrease in the dry season in SVN. The trend of ETo in the rainy season is clearer than in the dry season. Annual average value of ETo tended to increase, with the mean value for the stations was 0.59 ± 0.32 mm yr−1. The t-test result shows that the average increase in ETo over the three climate sub-regions is not different. In comparison to the annual average value of ETo in the sub-climates, the increase during 1977−2018 ranges from about 1.5% to 1.8%. Since the ETo only increases in the rainy season, it will not significantly increase the requirement of crop irrigation. Among the meteorological variables relating to ETo, the temperature change is most obvious. The temperature at 38/40 stations tends to increase at the 99% confidence level or more, and its average increase is about 0.6 ± 0.17°C over the past 40 years. The trends of the remaining meteorological variables are not really clear, as shown in the slope of the trend lines as well as the percentage of stations, which have a clear trend.

      By examining the role of relevant meteorological variables, it can be seen that temperature plays the most important role in the trend of ETo. In the past 40 years, the increase in temperature contributes 98.8% to the increase in ETo. Relative humidity has a slight decrease, but it is very sensitive to ETo and contributes to the rise of ETo about 15%. The slight decrease in sunshine hours also results in a ETo fall by 10%. Wind speed, due to its insignificant change, only makes a minor contribution to the change of ETo.

      In SVN, the influence of ENSO on ETo is only evident from October to May. During this period, the difference of mean ETo between the El Niño and La Niña years about 0.34 mm day−1. The driest period in SVN is from January to May while ETo increases from October to May during the El Niño year. This could lead to a reduce in the soil moisture in the early dry season and a rise of water requirements in the dry season. In the El Niño years, the precipitation in the dry season in SVN is often in a serious shortage. In addition, ETo and ΔETo are the highest at the later months of the dry season. Therefore, the severe drought in SVN usually occurs at the end of the dry season. In comparison to the trend of ETo over the past 40 years, the ETo difference between the warm and cold phases is much larger. Among the sub-regions, the influence of ENSO on ETo in the sub-region S3 is clearer. The reason is because this sub-region has the lowest latitude and flat topography and is located in coastal area.

      Among the factors of temperature, relative humidity, sunshine hours, and wind speed, the results show that the sunshine hours plays the most important role in the difference of ETo between the cold and warm phases. From October to May, the sunshine hours in the El Niño years are higher than that in La Niña years from 0.77 to 1.58 h day−1. This increase in sunshine hours contributes from 58% to 86% to the average ETo difference between the warm and cold phases in the period. Because there is no significant change in wind speed, the remaining contribution to the ETo difference between the two phases comes from humidity and temperature. The contribution of temperature and relative humidity are highest in April and May, but less than 20% for each factor. Because the number of sunshine hours plays a major role in the ETo difference between the two phases, it is necessary to conduct further studies to clarify the reason why this difference is significant. Compared to the one of climate change, the influence of ENSO on ETo in SVN is stronger. In addition, this influence is strongest in the dry season, seriously affecting agricultural production. Therefore, monitoring and forecasting meteorological variables related to ETo, especially the number of hours of sunshine, should be paid more attention to.

      Acknowledgments. This research is supported by the Project 2.21 in the framework of the bilateral cooperation between Vietnam and the Wallonie-Bruxelles Government in the period 2019−2021. We sincerely thank the organizations related to this project.

Reference (41)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return