water

Article Interdecadal Variability in Myanmar Rainfall in the Monsoon Season (May–October) Using Eigen Methods

Zin Mie Mie Sein 1, Irfan Ullah 2,* , Farhan Saleem 3,4, Xiefei Zhi 2,5,*, Sidra Syed 6 and Kamran Azam 7

1 College of International Students, Wuxi University, Wuxi 214105, China; [email protected] 2 School of Atmospheric Science, Nanjing University of Information Science and Technology, Nanjing 210044, China 3 College of Earth and Planetary Sciences, University of Chinese Academy of Sciences, Beijing 100049, China; [email protected] 4 International Centre for Climate and Environment Sciences, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100049, China 5 Weather Online Institute of Meteorological Applications, Wuxi 214000, China 6 Institute of Peace and Conflicts Studies, University of Peshawar, Peshawar 25000, Pakistan; [email protected] 7 Department of Management Sciences, University of Haripur, Khyber Pakhtunkhwa 22780, Pakistan; [email protected] * Correspondence: [email protected] (I.U.); [email protected] (X.Z.)

Abstract: In this study, we investigated the interdecadal variability in monsoon rainfall in the Myanmar region. The gauge-based gridded rainfall dataset of the Global Precipitation Climatology Centre (GPCC) and Climatic Research Unit version TS4.0 (CRU TS4.0) were used (1950–2019) to investigate the interdecadal variability in summer monsoon rainfall using empirical orthogonal



 function (EOF), singular value decomposition (SVD), and correlation approaches. The results reveal relatively negative rainfall anomalies during the 1980s, 1990s, and 2000s, whereas strong positive Citation: Mie Sein, Z.M.; Ullah, I.; rainfall anomalies were identified for the 1970s and 2010s. The dominant spatial variability mode Saleem, F.; Zhi, X.; Syed, S.; Azam, K. showed a dipole pattern with a total variance of 47%. The power spectra of the principal component Interdecadal Variability in Myanmar (PC) from EOF revealed a significant peak during decadal timescales (20–30 years). The Myanmar Rainfall in the Monsoon Season summer monsoon rainfall positively correlated with Atlantic multidecadal oscillation (AMO) and (May–October) Using Eigen Methods. Water 2021, 13, 729. https://doi.org/ negatively correlated with Pacific decadal oscillation (PDO). The results reveal that extreme monsoon 10.3390/w13050729 rainfall (flood) events occurred during the negative phase of the PDO and below-average rainfall (drought) occurred during the positive phase of the PDO. The cold phase (warm phase) of AMO Academic Editor: Scott Curtis was generally associated with negative (positive) decadal monsoon rainfall. The first SVD mode indicated the Myanmar rainfall pattern associated with the cold and warm phase of the PDO and Received: 26 January 2021 AMO, suggesting that enhanced rainfall for about 53% of the square covariance fraction was related Accepted: 3 March 2021 to heavy rain over the study region except for the central and eastern parts. The second SVD mode Published: 7 March 2021 demonstrated warm sea surface temperature (SST) in the eastern equatorial Pacific (El Niño pattern) and cold SST in the North Atlantic Ocean, implying a rainfall deficit of about 33% of the square Publisher’s Note: MDPI stays neutral covariance fraction, which could be associated with dry El Niño conditions (drought). The third with regard to jurisdictional claims in SVD revealed that cold SSTs in the central and eastern equatorial Pacific (La Niña pattern) caused published maps and institutional affil- enhance rainfall with a 6.7% square covariance fraction related to flood conditions. Thus, the extra- iations. subtropical phenomena may affect the average summer monsoon trends over Myanmar by enhancing the cross-equatorial moisture trajectories into the North Atlantic Ocean.

Keywords: summer monsoon rainfall; interdecadal variability; AMO; PDO; Myanmar Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and 1. Introduction conditions of the Creative Commons Attribution (CC BY) license (https:// Rainfall variability is among the basic indicators of climate and water cycle changes creativecommons.org/licenses/by/ in a region [1]. Anomalous changes in rainfall dynamics may cause hydro-meteorological 4.0/). hazards such as flood, drought, and storms [2,3], thus ultimately resulting in loss of human

Water 2021, 13, 729. https://doi.org/10.3390/w13050729 https://www.mdpi.com/journal/water Water 2021, 13, 729 2 of 27

lives, and destruction of biodiversity and natural resources [4,5]. Variation in rainfall pattern may also occur on intra-seasonal and inter-annual scales, posing serious threats to agriculture, water resources, and the sustainable development of a region [5,6]. The Intergovernmental Panel on Climate Change (IPCC) stated that global -warming-induced climate change has intensified the variability in rainfall over space and time across the globe [7]. It is also projected that the spatiotemporal variability in rainfall and related extremes will intensify in the near future [7]. These projections can be used as feedback for the policy makers in water resource management, disaster risk reduction, and regional climate analysis [8]. The climate of Myanmar is dominated by the Southeast Asian (SEA) monsoon; about 70% of the total annual rainfall is received during monsoon season (June to September), with large spatiotemporal variance at the country level across latitude and longitude. With high exposure and low resilience, the SEA region is highly vulnerable to climate change influences, especially as the intensity and frequency of extreme events may increase in the near future [9]. Ge et al. [10] showed future projections in rainfall extremes over the SEA using CMIP6 multi-model ensemble. They reported significant changes in the number of heavy rainfall days and the intensity of daily rainfall, demonstrating that locally heavy rainfall is likely to occur over a short time and that more rainfall extremes over the SEA are probable in a warmer future. Such increasing trends in extreme rainfall have been scientifically observed during last few decades over Asia [10,11], and will likely increase further with global warming. Variation in the monthly distribution of monsoon rainfall and its amounts will have a substantial impact on the agricultural production, water resources, and overall economy. It is expected that rainfall intensity shifting and patterns will undergo certain changes, and extreme weather events such as drought and flood may occur more frequently [12]. In recent years, such changes have already been experienced over the SEA, resulting in heavy rainfall and loss of capital, infrastructure, and human lives during 2009 and 2014 in Myanmar [13–15] and the 2015 and 2017 floods in India and Bangladesh [16]. These changes, to some extent, were linked to large-scale anomalous atmospheric circulation patterns and the resultant flux, which caused local and remote damage on a vast scale [17]. The classical division of monsoon areas used at the beginning of the 20th century was entirely based on the annual reversal of winds and increase rainfall during a short period of time [18]. The classical definition binds the monsoon to the eastern hemisphere, including Asian, Australian, tropical African, and Indian Ocean monsoon systems. The simulation experiments and geological records suggest that the Asian monsoon is sensitive to land surface features such as Tibetan uplift and Himalayan terrain [19,20]. Regional variability in magnitude, onset, and withdrawal of monsoonal rainfall is linked to continental, oceanic, and atmospheric circulation patterns [21]. Changes in these features are usually used as an integrated component in the forecast for predicting the above features [22,23]. The IPCC indicated that the impact of climate change on regional monsoon rainfall intensity and variability is more complex and uncertain, and the near-future monsoon system may be more severe than in the present decade [11]. Numerous studies have been conducted in Myanmar to identify the spatiotemporal trends in monthly, annual, and seasonal rainfall [21,24]. These variations strongly affect the ecological, social, and economic aspects of the country, such as availability of water resources, agricultural production, and economic practices [25,26]. A recent study indi- cated that the seasonal monsoon rainfall exhibits an increasing trend in the northwest of Myanmar, but a decreasing trend in the southern coastal belt [27,28]. The monsoon rainfall showed an upward trend in southeastern Myanmar, but a downward trend in the central parts [29]. The northern and northeastern plateau of Myanmar experienced an increasing trend in monsoon rainfall [30]. It has been projected that the interannual variation in monsoon rainfall will continue in the future with significant intensity, which will adversely impact the climate and hydrometeorological aspects of Myanmar [4,31]. Chhin et al., (2019) proposed an area-averaged sea surface temperature (SST) and a combination of different Water 2021, 13, 729 3 of 27

variables at each corresponding location over an equatorial region, and found a good predictor for operational long-term statistical prediction of monthly rainfall in the southern Indochina Peninsula, particularly for the pre-monsoon season. The aforementioned studies provided detailed information about the spatiotemporal changes in monsoon rainfall over Myanmar; however, none of the studies considered the sub-seasonal rainfall variation. Furthermore, the perceived potential drivers of monsoon variability have also not been explored in detail in the region [17,32]. According to Zaw et al., (2020), extreme events are associated with the broader-scale atmospheric circulations of the Pacific and Indian Oceans, especially during the positive phase of the El Niño southern oscillation (ENSO). Long-term climate variation affects seasonal to interannual climate variability and is important in the climate research over the SEA region, mainly over Myanmar. In this study, the empirical orthogonal function (EOF) method was used to analyze the decadal timescales of the dominant mode of the monsoon rainfall pattern. The extreme behavior of summer monsoon rainfall variability, which is related to global climate phenomena such as the Pacific decadal oscillation (PDO) and the Atlantic multidecadal oscillation (AMO), was analyzed using the singular value decomposition (SVD) method. After this, we examined the correlation between summer monsoon rainfall and SST on the decadal time scale. We aimed to fill the gap in the literature by assessing the interdecadal variability in summer monsoon rainfall to identify the relevant large-scale phenomena of the atmospheric and oceanic circulation. The summer monsoon mostly occurs during the months of May to October in the study region. Therefore, we considered May to October in our investigation of the monsoon season influence over Myanmar from 1950 to 2010 [4,33,34]. The rest of this paper is structured as follows: Section2 describes our data and methodology; the results are provided in Section3, with the discussion and conclusions presented in Section4.

2. Study Area Myanmar is a tropical country situated in the SEA region at 9◦–28◦ N and 92◦–101◦ E (Figure1). SEA is located in Asia, comprising the south of China, the east of India, and the northwest of Australia. The region is located between the Indian Ocean and the Bay of Bengal (BoB) in the west, the Philippine Sea, the South China Sea, and the Pacific Ocean in the east [35]. Myanmar’s climate is mostly influenced by the Indian summer monsoon [36]. The country has a tropical to subtropical monsoon climate with three seasons: hot, dry inter-monsoonal (mid-February to mid-May); rainy southwest monsoon (mid-May to late October); and cool and dry northeast monsoon (late October to mid-February) (NAPA, 2012). Myanmar is an agricultural country and monsoon rice occupies 80% of the planted rice area; for summer rice, it occupies 20% [37]. The region is located in the Indian monsoon regions where weather and climate disasters frequently produce damage in the summer monsoon season (May–October). For example, in 2019 and 2020, extreme floods and landslides in Myanmar occurred due to the strong monsoon rainfall, resulting in considerable loss of life and property. Long-lasting extremely dry periods and the very low amount of total monsoon rainfall (May–October) affect water scarcity and threaten the economy, livelihood, and food security in the country [4,31]. Water scarcity occurs in the central dry and deltaic regions during the summer (March–April) [38]. Water 2021, 13, 729 4 of 27 Water 2021, 13, x FOR PEER REVIEW 4 of 28

Figure 1. EElevationlevation map of the study area.

3. Data Data and and Methodology 3.1. Data 3.1. Data We used monthly gauge-based gridded rainfall datasets of the Global Rainfall Clima- We used monthly gauge-based gridded rainfall datasets of the Global Rainfall Cli- tology Centre (GPCC) and Climatic Research Unit version TS4.0 (CRU TS4.0) during the matology Centre (GPCC) and Climatic Research Unit version TS4.0 (CRU TS4.0) during period of 1950–2019 [39–42]. The gridded climate datasets were collected with a spatial the period of 1950–2019 [39–42]. The gridded climate datasets were collected with a spatial resolution of 0.5◦ × 0.5◦ over the study region. Several institutes have developed gridded resolution of 0.5° × 0.5° over the study region. Several institutes have developed gridded climate datasets during the past few decades, which are frequently used in hydro-climatic climate datasets during the past few decades, which are frequently used in hydro-climatic assessments [40,43,44]. Among others, the data products of CRUTS4.0 and GPCC are assessments [40,43,44]. Among others, the data products of CRUTS4.0 and GPCC are com- commonly used due to a longer temporal span [39,41]. The in situ observation data of 35 monly used due to a longer temporal span [39,41]. The in situ observation data of 35 me- meteorological stations were acquired from the Department of Meteorology and Hydrol- teorologicalogy (DMH), sta Myanmar.tions were The acquired long-term from in the situ Department observation of Meteorology data were further and Hydrology used after (DMH),validation Myanmar. with observed The long summer-term monsoonin situ observation rainfall (1950–2019), data were whichfurther is used highly after correlated valida- tionwith with GPCC observ griddeded summer data, to monsoon investigate rainfall the interdecadal (1950–2019), variability. which is highly correlated with GPCCThe gridded monthly data, data to ofinvestigate the Pacific the decadal interdecadal oscillation variability. (PDO) Index from May–October for theThe period monthly 1950 data to 2010of the were Pacific used decadal in this oscillation study [22 ,(PDO)45]. The Index AMO from index May is–October defined foras athe local period estimate 1950 overto 2010 the were entire used North in this Atlantic studylow-resolution [22,45]. The AMO annual index SST is anomaliesdefined as aafter local removing estimate over any directthe entire trend North [46]. Atlantic The North low Pacific-resolution (NP) annual index, SST North anomalies Atlantic after Os- removingcillation (NAO) any direct index, trend and [46] the. ArcticThe North Oscillation Pacific (AO)(NP) indexindex, data North were Atlantic also used Oscillation in this (NAO) index, and the Arctic Oscillation (AO) index data were also used in this study [47].

Water 2021, 13, 729 5 of 27

study [47]. Large-scale atmospheric circulation, such as air temperature and relative hu- midity, was collected from ERA5 the 5th generation reanalysis of the European Centre for Medium-Range Weather Forecasts (ECMWF) to determine the related breakouts and their moisture transport [48,49]. The extended version of the Reproduced Ocean Surface Temperature (ERSST, version3b), published in the National Maritime and Barometrical Administration/National Climatic Information Center, was used for seawater heating [50]. Analytical data from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) [51], such as speed, zonal and meridional wind, vertical speed (0.5◦ × 0.5◦), were used to explore long-term wind zones.

3.2. Methods The EOF is a statistical method broadly used to minimize the multidimensionality of complex climate information and recognize the most imperative physical modes with the least chance of data being misplaced [52,53]. Principal component analysis (PCA) is used for the assessment of atmospheric information based on the EOF method [54]. The technique describes the variance and covariance of the data, describing a few modes of variability. The modes that explain the largest percentage of the original variability are considered significant. The modes can be represented by orthogonal spatial patterns (eigenvectors) and corresponding time series (principal components). The EOF technique has been successfully used in Myanmar to investigate the interannual variability in summer monsoon rainfall [25]. With regard to the temporal behavior of the EOF mode, a 10-year running average was applied to obtain the long-term variability. The summer monsoon rainfall data were normalized to prevent areas and seasons of maximum variance from dominating the eigenvectors [55]. The standardized rainfall values were computed for all the years from the long-term mean, yearly mean, and the standard deviation using Equation (1): X − X Z = (1) Sd

where Z represents the standardized departure, X is the long-term mean value, and Sd is the standard deviation from the mean. The Z value provides immediate information about the significance of a specific deviation from the mean [25,56]. The orthogonal function of EOF is defined as:

N z(x, y, t) = ∑ PC(t) × EOF (x, y) (2) k−1

where z(x, y, t) denotes the function of space (x, y) and time (t), while EOF(x, y) represents the spatial structure in relation to the temporal variation in Z. A power spectrum was analyzed to evaluate the multiscale temporal variations result- ing from the wavelet transform employed in the long-term rainfall dataset [57–59]. The decadal spectra were used to estimate the changes that occur in the rainfall pattern on a decadal basis. Prior to performing the power spectrum analysis from the results of the EOF, PCA was used to verify the significance of the scales of interdecadal variation in this study. The red noise spectrum of the decadal spectrum was computed [54]. The confidence limits related to the read noise spectrum were calculated with 90% confidence levels. Composites include classifying and averaging one or more category of the variable fields according to their relation with the prevailing situations [60]. Composites results are used to produce hypotheses for patterns related to individual scenarios [53,61]. The goal of the composite analysis was to produce the wet and dry years separately. It is primarily used in China and Myanmar to detect the circulation anomalies connected with wet/dry events [25,45,62]. Horizontal moisture transfer rate/flux convergence (MFC), known as moisture con- vergence, was identified for wet and dry years using the vector character. The MFC results Water 2021, 13, 729 6 of 27

from the conservation of water vapor in pressure coordinates, which has been discussed in detail previous [63]. MFC can be calculated as:

MFC = − ∇.(qVh) = − Vh.∇q − q∇.Vh (3)     ˆ ∂ ˆ ∂ where ∇ = i ∂x + j ∂y , q is the specific humidity, and Vh = (u, v). The nonparametric Mann–Kendall test (MK) test was used to detect the trend in monsoon rainfall [64,65]. The MK test is widely used to evaluate trends in rainfall, agrome- teorological, and hydrological time series [66–72]. This trend test’s limitations are normally associated with its own null hypothesis, which assumes that the data are independently and identically distributed. However, the null hypothesis is commonly taken as evidence of the null hypothesis, which is commonly used as evidence of a trend in a given agro- meteorological and/or hydrological time series. The MK test investigates the sequential change point to emphasize an abrupt change significant at the 95% confidence level. This approach has been used successfully worldwide [70,73,74] to detect climate variables’ change points. For certain datasets consisting of x values with a sample size Sen slope (SS), the MK calculation begins by estimating the S statistic as:

SS−1 SS
S = ∑ ∑ sgn xj − xi ... f or j > I (4) i=1 j=i+1

As presented in Mann (1945) and Kendall (1975), when SS ≥ 8, the distribution of S approaches the Gaussian form with mean E(S) = o and variance V (S) is calculated using:

SS (SS − 1)(2SS + 5) − ∑SS ti (m − 1)(2m + 5)m V (S) = m=1 (5) 18 where ti is the number of ties of length m. The S statistic is standardized Z, as shown in Equation (6), and its significance can be estimated from the normal cumulative distribution function.  − √S 1 → S > 0  Vs Z = 0 → S = 0 (6)  √S+1 → S < 0 Vs

Singular value decomposition (SVD) analysis is used in two techniques: meteoro- logical and oceanographic data analysis. SVD is applied between two identical datasets of two jointly analyzed fields to identify pairs of the coupled spatiotemporal variations. Bretherton et al., (1992) [75] and Wallace et al., (1992) [76] provided a brief introduction to the SVD method as a fundamental matrix operation that can be considered an extension of rectangular matrices of the diagonalization of a square symmetric matrix. In the calcula- tion, each pair defines the covariance percentage between the two joint fields, allowing the extraction of the dominant modes of pair covariability between the two analyzed fields. Moreover, the SVD of the cross-covariance matrix identifies from two data fields pairs of spatial patterns that explain as much as possible of the mean-squared temporal variance between the two fields. This method is widely used [25] because it works on common and ambiguous data sets. In this work, SVD was used to find the maximum covariance or correlation between the summer monsoon rainfall over Myanmar and the SST over the Indian Ocean and Pacific Ocean. After this, we used ensemble empirical mode decomposition (EEMD), which is an adaptive time-frequency data analysis technique developed by Wu et al., (2007) [77] and Wu and Huang (2009) [78]. It is used to resolve problems of mixed-mode and false components in the empirical mode decomposition (EMD) decomposition process. This method is a major modification of the original EMD method, emphasizing the adaptiveness and temporal locality of the data decomposition [79]. The EMD method can decompose Water 2021, 13, 729 7 of 27

any complicated data series into a few amplitudes of frequency-modulated oscillatory components, called intrinsic mode functions (IMFs), of distinct timescales [80]. EMD is based on the local characteristic time scale of a signal, and decomposes the complicated signal into a number of IMFs. The main steps of the EEMD method employed in this study were similar to those outlined by Qian (2009) [81,82]. In this study, EEMD was used to decompose the summer monsoon rainfall and PDO index.

4. Results 4.1. General Characteristics of Monsoon Rainfall 4.1.1. Annual Cycle The climatological annual cycle of the monthly mean monsoon rainfall over Myanmar (1950–2019) was estimated based on: in situ observation, CRUTS4.0, and GPCC datasets; the results are shown in Figure2. The error bars shows the monthly standard deviation of climatological annual cycle with in each active and break spells. The horizontal solid lines show the monthly standard deviations of all datasets with blue, grey, and orange bars. Interestingly, the significant number of active spells (blue and orange bars) show positive rainfall pattern from June to August whereas the break spells show relatively weak rainfall pattern during September to October. Table1 reveals the amount of average monthly summer monsoon rainfall (May–October) from in situ observations, CRUTS4.0, and GPCC for the period of 1950–2019. The country experienced a relatively high unimodal rainfall pattern in June, July, and August, observed in all datasets. The average monthly rainfall from in situ observations was 480 mm in June, 520 mm in July, and 524 mm in August. From GPCC, these amounts were 478, 519, and 519 mm, respectively. The rainfall from CRUTS4.0 exhibited a weaker pattern in these three months (444, 466, and 476 mm, respectively). The higher rainfall agrees with the northern hemisphere summer and monsoon season. The rainfall season in Myanmar extends from May to October, and the rainfall season coincides with summer monsoon onset in the country, which generally occurs in the third week of May. Similar monsoon onset periods were observed by Sein et al., (2015) in a simulation study using Regional Climate Model (RegCM3). Notably, the GPCC dataset produced more accurate results compared to the in situ observations. CRUTS4.0 reported less rainfall compared to the in situ observations and GPCC. We found a strong correlation (0.94) between the in situ observations and GPCC datasets, but a relatively weak correlation (0.52) between the in situ observations and the CRUTS4.0 dataset (Table2). The root mean square error (RMSE) for GPCC (14.24) for summer rainfall was lower than that of the CRUTS4.0 dataset (45.29) (Table2). Therefore, GPCC was thus used in most analyses since the data are continuous and can be extrapolated to cover all parts of the country to identify longer time periods. Roy et al., (2011) studied the interannual variability in summer monsoon rainfall in relation to Indian Ocean Dipole (IOD) and ENSO using the GPCC dataset. Therefore, the use of GPCC data provided better results, due to extended longer periods, which were more complete as compared to the observed data. Water 2021, 13, x FOR PEER REVIEW 8 of 28

Table 1. Monthly average of the summer monsoon rainfall (mm) from in situ observations, and Global Rainfall Climatology Centre (GPCC) and Climatic Research Unit version TS4.0 (CRUTS4.0) datasets. The bold values represents the significance level (0.05).

Month In situ GPCC (mm) CRUTS4.0 (mm) Jan 4.66 5.28 5.60 Feb 9.94 10.54 12.12 Mar 20.93 20.19 20.72 Apr 55.25 53.95 62.75 May 248.03 237.92 224.41 Jun 480.32 478.80 444.51 Jul 520.03 519.32 466.11 Aug 524.53 519.38 475.89 Sep 340.86 344.49 328.81 Oct 175.52 179.23 190.30 Water 2021, 13, 729 Nov 53.25 57.80 64.77 8 of 27 Dec 9.10 9.36 9.54

600 OBS GPCC 500 CRU

400

300

Rainfall (mm) Rainfall 200

100

0 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Months

Figure 2. The annual cycle of mean rainfall from in situ observations (OBS), GPCC, and CRUTS4.0 over Myanmar (mm) (1950–2019).Figure The error 2. The bars annual show cycle the monthly of mean standardrainfall from deviation in situ ofobservation the interannuals (OBS) variability, GPCC, and within CRUTS4.0 the periods. The over Myanmar (mm) (1950–2019). The error bars show the monthly standard deviation of the in- horizontal solid lines shows the standard deviation of all datasets blue, grey, and orange bars. terannual variability within the periods. The horizontal solid lines shows the standard deviation of all datasets blue, grey, and orange bars. Table 1. Monthly average of the summer monsoon rainfall (mm) from in situ observations, and Table 2. SummaryGlobal of correlation Rainfall Climatology and root mean Centre square (GPCC) error and (RMSE Climatic) among Research GPCC Unitand CRUTS4.0 version TS4.0 (CRUTS4.0) datasets and in datasets.situ observation The bold stations values’ data represents over Myanmar. the significance level (0.05).

Dataset MonthCorrelation In SituCoefficient (R) GPCC (mm)RMSE CRUTS4.0 (mm) GPCC Jan 4.660.94 5.2814.24 5.60 CRUTS4.0 Feb 9.940.52 10.5445.29 12.12 Mar 20.93 20.19 20.72 Apr 55.25 53.95 62.75 4.1.2. Climatology of Summer Monsoon Rainfall May 248.03 237.92 224.41 Figure 3 shows the Junclimatology of observed480.32 May–October mean 478.80 monsoon rainfall over 444.51 Myanmar (Figure 3a) andJul GPCC rainfall (Figure520.03 3b) for the period of 519.32 1950 to 2019. Both da- 466.11 tasets showed relatively Augsimilar spatial rainfall 524.53patterns over the country 519.38 (Figure 3). Maximal 475.89 rainfall was received alongSep the western coast 340.86 (Rakhine State), and 344.49southern (Mon, Kayin 328.81 Oct 175.52 179.23 190.30 States and Tanintharyi NovRegion) and northern 53.25tips (Kachin State) of Myanmar. 57.80 This aligns 64.77 Dec 9.10 9.36 9.54

Table 2. Summary of correlation and root mean square error (RMSE) among GPCC and CRUTS4.0 datasets and in situ observation stations’ data over Myanmar.

Dataset Correlation Coefficient (R) RMSE GPCC 0.94 14.24 CRUTS4.0 0.52 45.29

4.1.2. Climatology of Summer Monsoon Rainfall Figure3 shows the climatology of observed May–October mean monsoon rainfall over Myanmar (Figure3a) and GPCC rainfall (Figure3b) for the period of 1950 to 2019. Both datasets showed relatively similar spatial rainfall patterns over the country (Figure3). Maximal rainfall was received along the western coast (Rakhine State), and southern (Mon, Kayin States and Tanintharyi Region) and northern tips (Kachin State) of Myanmar. This aligns with the recent extreme events (floods, strong winds, and landslides) that have occur in these areas due to the strong monsoon rainfall. Along the coastal area (west and south), Water 2021, 13, x FOR PEER REVIEW 9 of 28

Water 2021, 13, 729 9 of 27 with the recent extreme events (floods, strong winds, and landslides) that have occur in these areas due to the strong monsoon rainfall. Along the coastal area (west and south), regionsregions close clos toe to the the BoB, BoB, Andaman Andaman Sea, Sea, and and northern northern tip tip (Kachin (Kachin State) State) occupy occupy the the highest highest mountain,mountain called, called Hkakabo Hkakabo Razi.Razi. The The c centralentral (dry (dry zone) zone) and and eastern eastern area areass receive receive less less rain- rainfallfall than than other other place placess in in the the region. region. The The central central dry dry zone zone receives receives the the lowest lowest amount amount of ofrainfall rainfall compared compared to to rest of thethe studystudy areaarea (Figure(Figure3 ).3). The The observed observed low low rainfall rainfall in in the the centralcentral dry dry zone zone is attributedis attributed to theto the topography: topography the: areathe area is situated is situated between between two mountaintwo moun- ranges:tain range thes: Shan the Shan Plateau Plateau to the to east the east and and Rakhine Rakhine Yoma Yoma Mountain Mountain to theto the west. west Kumar. Kumar etet al., al. (2006) (2006) indicated indicated that that increased increased rains rains on on the the Myanmar Myanmar coast coast contribute contribute more more than than totaltotal rainfall rainfall in in the the West West Ghats, Ghats associated, associated to to the the central central and and northern northern regions. regions.

Figure 3. Climatology of mean summer monsoon (May–October) rainfall (mm) over Myanmar (1950–2019) (a) in situ observation rainfall; (b) GPCC rainfall. Figure 3. Climatology of mean summer monsoon (May–October) rainfall (mm) over Myanmar (1950–2019) (a) in situ ob- servation rainfall; (b) GPCC4.2. rainfall. Spatial and Temporal Interdecadal Variability in Summer Monsoon Rainfall. The change in decadal mean May–October rainfall from the long-term mean rainfall over4.2. MyanmarSpatial and was Temporal estimated Interdecadal using GPCC Variability data in (Figure Summer4). InMonsoon the six Rainfall decades,. the results show strongThe change negative in decadal anomalies mean duringMay–October the periods rainfall of from 1971–1980, the long 1981–1990,-term mean andrainfall 1991– over 2000Myanmar (Figure was4c–e). estimated From the using results, GPCC the data highest (Figure frequency 4). In the of six rainfall decades, occurred the results during show thestrong period negative 1961–1970, anoma followedlies during by the 2001–2010, periods of suggesting 1971–1980, frequent 1981–1990, extreme and 1991 rainfall–2000 in (Fig theure recent4 c–e). decade, From the whereas results, 1981–1990 the highest and frequency 1991–2001 of rainfall showed occurred the maximum during the displacement period 1961– toward1970, followed the right by side 2001 of the–2010 figure,, suggesting indicating frequent an increase extreme in rainfall the maximum in the recent intensity decade, of monsoonwhereas rainfall1981–1990 in theand region.1991–2001 Figure showed4a,b the indicates maximum less rainfalldisplacement in the toward western the and right centralside of parts the figure of Myanmar;, indicating however, an increase the oppositein the maximum was true: intensity the patterns of monsoon in Figure rainfall4d,e in exhibitthe region. strong Figure negative 4a,b changeindicates anomalies. less rainfall Sein in the et al.,western (2015) and studied central ENSO parts of associated Myanmar; withhowever, summer the monsoon opposite rainfall. was true: They the foundpatterns that in mostFigure of 4d the.e Elexhibit Niño strong years showednegative strong change negativeanomalies. decadal Sein rainfall et al. (2015) anomalies studied in Myanmar. ENSO associated In addition, with strong summer positive monsoon anomalies rainfall. wereThey recorded found that in 1961–1970 most of the and El Niño 2001–2010 years (Figureshowed4 b,f).strong The negative results decadal further supportrainfall anom- the findingsalies in ofMyanmar. Zaw et al.,In addition, (2020), who strong reported positive that anomalies the occurrence were recorded of extreme in 1961 events–1970 (i.e., and drought2001–2010 and ( flood)Figure is 4b,f). significantly The results associated further support with large-scale the findings atmospheric of Zaw et circulations al. (2020), who of the Pacific and Indian Oceans, particularly during the positive phase of the ENSO. The overall strong and positive decadal rainfall anomalies occurred in La Niña years over the

Water 2021, 13, x FOR PEER REVIEW 10 of 28

reported that the occurrence of extreme events (i.e., drought and flood) is significantly Water 2021, 13, 729 associated with large-scale atmospheric circulations of the Pacific and Indian Oceans,10 ofpar- 27 ticularly during the positive phase of the ENSO. The overall strong and positive decadal rainfall anomalies occurred in La Niña years over the target region. Table 3 reveals that the observed extreme rainfall between 1950 and 2019 occurred in the first three decades targetfor the region. west coast Table (Rakhine3 reveals State), that thefor observedtwo decades extreme along rainfallthe south between coast (Tanintharyi 1950 and 2019 Re- occurred in the first three decades for the west coast (Rakhine State), for two decades along gion), and for one decade in the Upper Sagaing Region. This result agrees with the spatial the south coast (Tanintharyi Region), and for one decade in the Upper Sagaing Region. decadal monsoon rainfall in Figure 4. This result agrees with the spatial decadal monsoon rainfall in Figure4.

Figure 4. Decadal change in mean summer monsoon rainfall (mm) over Myanmar: (a) 1951–1960; (b) 1961–1970; (c) 1971–1980;Figure 4. Decadal (d) 1981–1990; change (ine) mean 1991–2000; summer (f) 2001–2010.monsoon rainfall (mm) over Myanmar: (a) 1951–1960; (b) 1961–1970; (c) 1971– 1980; (d) 1981–1990; (e) 1991–2000; (f) 2001–2010. Table 3. Observed extreme decadal rainfall over Myanmar during the period of 1950–2019. Table 3. Observed extreme decadal rainfall over Myanmar during the period of 1950–2019. Sr. No Year Station Region/State Rainfall (mm) Sr. No Year Station Region/State Rainfall (mm) 1 4.9.1965 Kyaukpyu Rakhine 568 1 24.9.1965 24.8.1997 Kyaukpyu DaweiRakhine Tanintharyi 568 549 2 324.8.1997 29.6.1989 Dawei HkamtiTanintharyi Upper Sagaing 549 527 3 429.6.1989 4.7.2006 Hkamti DaweiUpper TanintharyiSagaing 527 447 4 54.7.2006 5.6.1980Dawei SittweTanintharyi Rakhine 447 422 6 18.7.1956 Thandwe Rakhine 353 5 5.6.1980 Sittwe Rakhine 422 6 18.7.1956 Thandwe Rakhine 353 From Figure4 and Table3, the decadal rainfall fluctuated during the period of 1950–2019From, Figure with a 4 positive and Table skewness 3, the decadal and increased rainfall fluctuated frequency during in recent the period decades. of 1950 This– suggests an increase in the extreme rainfall events with increased frequency in all study 2019, with a positive skewness and increased frequency in recent decades. This suggests regions, especially in the western and southern regions. This may be associated with an increase in the extreme rainfall events with increased frequency in all study regions, increased global-warming-induced changes in monsoon and circulation patterns in the especially in the western and southern regions. This may be associated with increased regions; similar rainfall patterns were previously reported [83]. The results for various regions in Myanmar revealed that the west, south, and north regions received the most rainfall, while the east and central regions received relatively less (Figure5). Figure5 shows that an overall increasing trend in rainfall occurred over all regions, which corresponds to previously reported findings [18,84,85]; however, the Water 2021, 13, x FOR PEER REVIEW 11 of 28

global-warming-induced changes in monsoon and circulation patterns in the regions; sim- ilar rainfall patterns were previously reported [83]. The results for various regions in Myanmar revealed that the west, south, and north Water 2021, 13, 729 11 of 27 regions received the most rainfall, while the east and central regions received relatively less (Figure 5). Figure 5 shows that an overall increasing trend in rainfall occurred over all regions, which corresponds to previously reported findings [18,84,85]; however, the an- nual rainfall variabilityannual rainfall on the variability regional on scale the differ regionaled. In scale thediffered. northern In part, the northern the rainfall part, the rainfall variability appearedvariability to be appeared increasing to, bewith increasing, an amount with of 500 an amount mm. The of peak 500 mm. rainfall The years peak rainfall years (dry years) were(dry recorded years) wereduring recorded 1992, 1997, during and 1992,2010 (1982, 1997, and1991, 2010 2005, (1982, and 2009). 1991, In 2005, the and 2009). In eastern part, monsoonthe eastern rainfall part, slightly monsoon increased, rainfall as slightly depicted increased, by a rainfall as depicted amount by of a250 rainfall amount mm with dry andof 250 wet mm yearswith appearing dry and wetto be years the possible appearing reason to be for the balancing possible the reason overall for balancing the trend in the region.overall In trendthis region, in the the region. highest In (lowest) this region, monsoon the highest rainfall (lowest) was observed monsoon in rainfall was 1988 and 2013observed (1987 and in 2009). 1988 In and the 2013 central (1987 part and, the 2009). monsoon In the rainfall central exh part,ibited the an monsoon in- rainfall creasing trendexhibited at the rate an of increasing300 mm per trend year at. The the analysis rate of 300 show mmed per that year. 1960, The 1997 analysis, 2010, showed that and 2014 (1987,1960, 1991, 1997, 1999 2010, and and 2009) 2014 were (1987, the 1991,peak 1999wet and(dry) 2009) years were in the the northern peak wet and (dry) years in the central parts. Ournorthern results and concur central with parts. the Ourfindings results of Zaw concur et withal. (2021 the) findings, who found of Zaw that et 2– al., (2021), who 4 year high-frequencyfound that periodicities2–4 year high-frequency from spectral peaks periodicities, and predominant from spectral regions peaks, of high and predominant spatial correlationsregions indicated of high spatialthe summer correlations rainfall indicated in Myanmar the summer is associated rainfall with in Myanmar large- is associated scale atmosphericwith circulations, large-scale mainly atmospheric linked circulations, with the ENSO mainly events linked due to with SST thedeviation ENSOs events due to in the tropicalSST Pacific deviations Ocean. inThe the variations tropical Pacificin the west Ocean.ern Theregion variations’s dry condition in the western were region’s dry observed in thecondition late 1960s were and observed 1980s (Figure in the 5d) late and 1960s in the and late 1980s 2000s (Figure in southern5d) and region in the late 2000s in (Figure 5e). In southernthe western region part, (Figure the monsoon5e). In rainfall the western experienced part, the an monsoon increasing rainfall trend in- experienced an itially with an increasingamount of trend250 mm, initially but a withdecreasing an amount pattern of 250in the mm, last but few a decreasingdecades. In patternthe in the last southern part,few the decades.opposite In pattern the southern to the part,western the oppositeregion was pattern observed, to the western recording region the was observed, recording the highest rainfall (530 mm) over last decades, being the highest amongst all highest rainfall (530 mm) over last decades, being the highest amongst all regions. The regions. The maximum (minimum) rainfall was recorded in 1960, 1980, 1998, 2006, and maximum (minimum) rainfall was recorded in 1960, 1980, 1998, 2006, and 2011 (1987, 2011 (1987, 1991, 1999, 2002, 2004, 2005, and 2014) in the region. 1991, 1999, 2002, 2004, 2005, and 2014) in the region.

Figure 5. BlackFigure lines, 5. homogeneousBlack lines, homogeneous regions’ average regions summer’ average monsoon summer rainfall monsoon time series rainfall (mm); time red series line, (mm) the corresponding; 10 year runningred line, mean the variations corresponding (mm): 10 (a) year north, running (b) east, mean (c) central, variations (d) west,(mm) and: (a) (north,e) south (b) regions east, (c) in central, Myanmar. (d) west, and (e) south regions in Myanmar. The time series of interdecadal variability in rainfall over the whole region is shown in Figure6. The mean rainfall was observed to increase in the late 1950s, reaching its peak around 1965, and subsequently decreased during the mid-1980s before increasing again in the early 1990s. The second climax was observed in the early 2000s; however, the rainfall decreased in the recent decade. We classified wet and dry years based on the standardized

rainfall anomaly as ≥1 and ≤–1 for wet and dry years, respectively. The same approach was used effectively by other authors (Chen and Huang, (2017) in China, Lone et al., (2019) in India, and Sein et al., (2015) in Myanmar). The wet periods were 1962–1967, 1998, 2003, Water 2021, 13, x FORWater PEER 2021 REVIEW, 13, x FOR PEER REVIEW 12 of 28 12 of 28

The time series ofThe interdecadal time series variability of interdecadal in rainfall variability over the in wholerainfall region over theis shown whole region is shown in Figure 6. The meanin Figure rainfall 6. The was mean observed rainfall to increasewas observed in the tolate increase 1950s, reachingin the late its 1950s, peak reaching its peak around 1965, andaround subsequently 1965, and decreased subsequently during decreased the mid-1980s during before the mid increasing-1980s before again increasing again in the early 1990s.in The the secondearly 1990s. climax The was second observed climax in thewas early observed 2000s; in however, the early the2000s; rain- however, the rain- fall decreased in fallthe decreasedrecent decade. in the We recent classified decade. wet We and classified dry years wet based and ondry the years stand- based on the stand- Water 2021, 13, 729ardized rainfall anomalyardized rainfallas ≥1 and anomaly ≤–1 for aswet ≥1 and and dry ≤–1 years, for wet respectively. and dry years, The respectively.same ap- The same12 ofap- 27 proach was used proacheffectively was by used other effectively authors by(Chen other and authors Huang, (Chen (2017) and in Huang,China, Lone(2017) et in China, Lone et al. (2019) in India,al. and (2019) Sein in et India, al. (2015) and Seinin Myanmar et al. (2015)). The in wet Myanmar periods). Thewere wet 1962 periods–1967, were 1962–1967, 1998, 2003, and 20041998,; typical 2003, anddry years2004; weretypical observed dry years during were theobserved periods during of 1978 the–1986 periods and of 1978–1986 and and 2004; typical dry years were observed during the periods of 1978–1986 and 1988–1990. 1988–1990. The results1988– from1990. theThe s resultsequential from Mann the –sKendallequential test Mann statistic–Kendall were testcalculated statistic for were calculated for The results from the sequential Mann–Kendall test statistic were calculated for interdecadal interdecadal rainfallinterdecadal variability rainfall over the variability region to over detect the the region statistically to detect significant the statistically turn- significant turn- rainfall variability over the region to detect the statistically significant turning points in the ing points in the indecadalg points trend in the in decadalrainfall (trendFigure in 7). rainfall An abrupt (Figure interdecadal 7). An abrupt change interdecadal was change was decadal trend in rainfall (Figure7). An abrupt interdecadal change was noted during the noted during the noted1970s; duringhowever, the there 1970s was; however, a significant there reductionwas a significant (at the 95%reduction confidence (at the 95% confidence 1970s; however, there was a significant reduction (at the 95% confidence level) in rainfall level) in rainfall betweenlevel) in therainfall 1980s between and 2000s the (1980sFigure and 7), 2000swhich ( Figureagrees 7),with which spatial agrees map with spatial map between the 1980s and 2000s (Figure7), which agrees with spatial map shown in Figure4. shown in Figure 4.shown in Figure 4.

Figure 6.FigureInterdecadal 6. Interdecadal variabilityFigure variability estimated6. Interdecadal estimated with avariability 10with year a 10 running estimated year running mean with of mean a the 10 standardizedyear of the running standardized mean summer of sum- the monsoon standardized rainfall sum- (mm) anomalymer monsoon over Myanmar. rainfallmer (mm) monsoon anomaly rainfall over (mm) Myanmar. anomaly over Myanmar.

Figure 7. Abrupt change in summer monsoon rainfall derived from the sequential Mann–Kendall test statistic; U(t), for- FigureFigure 7. Abrupt change inin summersummer monsoon monsoon rainfall rainfall derived derived from from the the sequential sequential Mann–Kendall Mann–Kendall test test statistic; statistic; U(t) U, forward(t), for- ward sequential statistic; U’(t), backward sequential statistic; the upper and lower dashed lines represent significant at the wardsequential sequential statistic; statistic U’(t);, U backward’(t), backward sequential sequential statistic; statistic; the upper the upper and lowerand lower dashed dash linesed line represents represent significant significant at the at 95%the 95% confidence level.95% confidence level. confidence level. Figures 6 and 7 show that regional variability in monsoon rainfall over Myanmar was FiguresFigures 6 andand 7 show that that regional regional variability variability in in monsoon monsoon rainfall rainfall over over Myanmar Myanmar was obvious during the 1950–2019 period. The magnitude of the rainfall variability could pos- obviouswas obvious during during the 1950 the– 1950–20192019 period period.. The magnitude The magnitude of the of rainfall the rainfall variability variability could could pos- sibly be associated with the geographical location of the region and monsoon circulation siblypossibly be associated be associated with with the the geographical geographical location location of ofthe the region region and and monsoon monsoon circula circulationtion patterns. The temporal trend in the Asian and Indian monsoon system showed a strong variation on different time scales, including monthly, inter-annual, intra-seasonal, and annual time scale [18,84,85]. Such large-scale variation due to different factors could induce changes in monsoon rainfall over the diverse study regions [21,86,87]. Water 2021, 13, x FOR PEER REVIEW 13 of 28

Water 2021, 13, 729 13 of 27 patterns. The temporal trend in the Asian and Indian monsoon system showed a strong variation on different time scales, including monthly, inter-annual, intra-seasonal, and an- nual time scale [18,84,85]. Such large-scale variation due to different factors could induce changes in monsoon rainfall over the diverse study regions [21,86,87]. 4.3. Interdecadal Variations in Rainfall and Associated Circulation Influences 4.3.1.4.3. Interdecadal Interdecadal Variations Rainfall in Rainfall Variability and Associated Circulation Influences 4.3.1.The Interdecadal EOFs were Rainfall examined Variability with the first three eigenvectors of the summer monsoon rainfallThe in EOFs Myanmar were examined for the decadalwith the 10first year three running eigenvectors mean of (Figure the summer8). In monsoon Myanmar, EOF approachesrainfall in Myanmar have been for the used decadal to detect 10 year the running interannual mean (Figure variability 8). In Myanmar, in summer EOF monsoon rainfallapproaches [31 ,have83]. been These used methods to detect arethe interannual suitable for variability finding in the summer total variancemonsoon rain- in monsoon rainfallfall [31,83] in. theThese region. methods The are results suitable reveal for find thating the the first total EOF variance (EOF1), in monsoon explaining rainfall 47% of the totalin the variance, region. The showed results areveal dipole that spatial the first pattern EOF (EOF1), with positive explaining loading 47% of onthe the total western var- coast (Rakhineiance, showed region) a dipole and spatial negative pattern loading with positive in northwestern loading on region the western (Sagaing coast Region)(Rakhine over the targetregion) region and negative (Figure loading8a, top in and northwestern bottom), andregion explained (Sagaing Region) the corresponding over the target PC-1, re- which clearlygion (Figure exhibited 8a, top interdecadal and bottom), variability.and explained EOF2, the corresponding explaining 18%PC-1,of which the totalclearly variance, exhibited interdecadal variability. EOF2, explaining 18% of the total variance, exhibited a exhibited a dipole pattern, with the opposite pattern to EOF1 (Figure8b, top and bottom), dipole pattern, with the opposite pattern to EOF1 (Figure 8b, top and bottom), revealing revealing that the PC time series is related to EOF2 (PC2). The third EOF (EOF3), explaining that the PC time series is related to EOF2 (PC2). The third EOF (EOF3), explaining 11% of 11%the total of the variance total variance,, showed strong showed positive strong loading positive in loadingthe southern in the region, southern and the region, time and the timeseries series of the ofconsistent the consistent PC is connected PC is connected to EOF3 (PC3) to EOF3, as shown (PC3), in Figure as shown 8c. To in validate Figure 8c. To validatethe significance the significance of the scales of theof interdecadal scales of interdecadal variation, we variation, used the wepower used spectrum the power from spectrum fromthe results the results EOF: the EOF: PC for the 53 PC years for was 53 years analyzed was (Figure analyzed 9). PC1 (Figure–PC39 exhibited). PC1–PC3 a signif- exhibited a significanticant spectral spectral peak detected peak detected at nearlyat 20 nearly–30 year 20–30 time scales. year time scales.

FigureFigure 8. The8. The first first three three empirical empirical orthogonalorthogonal functions functions (EOF (EOF__ (top (top) spatial) spatial modes modes and andEOF EOF(bottom (bottom) time) series time: series:(a) EOF (-a) EOF-1, (b) EOF-2,1, (b) EOF and-2, (andc) EOF-3. (c) EOF-3.

Water 2021, 13, 729 14 of 27 Water 2021, 13, x FOR PEER REVIEW 14 of 28

(a) (b) (c)

Figure 9. Power spectra:spectra: ( a) principal component (PC)1,(PC)1, (b) PC2, andand (c) PC3. The red line indicates the Markov redred noisenoise spectrum, the green dasheddashed lineline representsrepresents thethe 90%90% significance,significance, andand thethe blueblue dasheddashed lineline denotesdenotes thethe 10%10% significancesignificance boundary forfor thethe Markov analysis.analysis.

From FiguresFigures8 8and and9, 9, we we inferred inferred that that the the interdecadal interdecadal rainfall rainfall pattern pattern over over Myanmar Myan- marhas significantlyhas significantly changed changed over over the study the study period, period, especially especially during during EOF2 EOF2 and EOF3.and EOF3. PC1 PC1and and PC2 PC2 experienced experienced an increasingan increasing spectrum spectrum over over the the target target region region during during thethe periodperiod 1950–2019.1950–2019. The humid regions of the country appeared to experience decreasing rainfall, as can bebe seenseen inin FigureFigure8 c,8c, while while the the arid arid regions regions in thein the northwestern northwestern area area experienced experienced an anincrease increase in monsoonin monsoon rainfall. rainfall. The The humid humid central central region region appeared appeared to be to the be corethe core monsoon mon- soonregion region of Myanmar; of Myanmar; thus, thus, this couldthis could imply imply a shift a shift toward toward the northwesternthe northwestern parts parts of the of thecountry. country. Further Further studies studies with with more more data data and and modeling modeling approaches approaches are neededare needed to validate to val- idatethe current the current hypotheses. hypotheses.

4.3.2. Large Large-Scale-Scale Atmospheric Circulation Pattern Rainfall variabilityvariability is is influenced influenced by by atmospheric atmospheric circulation, circulation, geographical geographical location, location, and andlow low latitude latitude in a countryin a country like like Myanmar. Myanmar. In this In this section, section, the generalthe general circulation circulation anomalies anom- aliesover over low latitudeslow latitudes are examined are examined as a majoras a major issue. issue. Circulation Circulation patterns patterns on an on interdecadal an interde- cadalscale arescale analyzed are analyzed over the over region. the region. To examine To examine the interdecadal the interdecadal variability variability in the country’s in the countryrainfall’s related rainfall to related the typical to the circulation typical circulation patterns patterns on the interdecadal on the interdecadal time scale, time we scale em-, ployedwe employ a lowed levela low of level horizontal of horizontal and vertical and vertical wind wind anomalies anomalies of 850 of hPa 850 forhPa the for wet the andwet anddry periodsdry periods (Figure (Figure 10). The 10). results The results show ashow significant a significant negative negative westerly westerly or southwesterly or south- winds anomaly from the BoB and Andaman Sea over almost the whole region except the westerly winds anomaly from the BoB and Andaman Sea over almost the whole region eastern and southern parts during wet years, as shown in Figure 10a. During dry years except the eastern and southern parts during wet years, as shown in Figure 10a. During (Figure 10a), significant negative northeasterly and easterly wind anomalies prevailed in dry years (Figure 10a), significant negative northeasterly and easterly wind anomalies the region. The southwesterly and BoB appeared, which are assumed to be the prime prevailed in the region. The southwesterly and BoB appeared, which are assumed to be sources of oceanic water transport to continental land masses for rainfall during monsoon the prime sources of oceanic water transport to continental land masses for rainfall during seasons (Figure S2). A similar pattern was observed for wind anomalies, further explaining monsoon seasons (Figure S2). A similar pattern was observed for wind anomalies, further dry conditions and elevated air temperature due to less rainfall. The continental land mass explaining dry conditions and elevated air temperature due to less rainfall. The continen- of South East Asia (SEA) experiences a strong anticyclonic pattern, which (in the northern tal land mass of South East Asia (SEA) experiences a strong anticyclonic pattern, which hemisphere) is associated with decreased wind and rainfall. The anticyclonic pattern (in the northern hemisphere) is associated with decreased wind and rainfall. The anticy- appeared to be pushing the cyclonic activity toward the tropical region and northern parts clonicof the BoBpattern by pushingappeared the to southbe pushing easterlies the cyclonic to the peninsular activity toward tip of Indiathe tropical (Figure region S2). and northern parts of the BoB by pushing the south easterlies to the peninsular tip of India (Figure S2).

WaterWater 20212021, ,1313, ,x 729 FOR PEER REVIEW 1515 of of 28 27

FigureFigure 10. 10. TheThe mean mean summer monsoonmonsoon seasonseason (May–October) (May–October) wind wind anomalies anomalies (m/s) (m/s) at at 850 850 hPa hPa for for (a) wet(a) wet years years (1962–1967, (1962– 1967,1998, 1998, and 2003–2004) and 2003–2004) and ( band) dry (b years) dry (1978–1986years (1978 and–1986 1988–1990). and 1988– The1990). red The (blue) red color(blue) indicates color indicates the significant the significant positive positive(negative) (negative) anomalies anomal overies the over target the region. target region.

DuringDuring wet wet years years (Figure (Figure 10b), 10b), the the subtro subtropicalpical and and Somali Somali jet jet streams streams appeared appeared to to be be pushingpushing the the oceanic oceanic moisture moisture-rich-rich winds winds toward toward the the continental continental land land mass. mass. The The southern southern regionregion received received northeasterly northeasterly and and easterly easterly winds winds from from the the South South China China Sea, Sea, which which is is the the corecore region region experiencing experiencing monsoon monsoon rain rainfall.fall. However, However, the the anomalous anomalous cloud cloud liquid liquid water water contentscontents were were obvious obvious over over the the Western Western SEA, SEA, which which includes includes Myanmar, Myanmar, with with enhanced enhanced windswinds speed speed and and reduced reduced air air temperature, temperature, implying implying increased increased rainfall. rainfall. The The anticyclonic anticyclonic patternpattern appeared appeared to movemove toto thethe BoB BoB region region strengthening strengthening the the intensity intensity of sub-tropicalof sub-tropical jets, jets,pushing pushing them them further further into into continental continental land land masses. masses. ToTo assess assess the the rainfall rainfall variability variability and and its its relationship relationship with with regional regional-scale-scale atmospheric atmospheric circulation,circulation, oceanic oceanic temperature temperature,, and and monsoon monsoon indices, indices, a a compos compositeite analysis analysis was was per- per- formed.formed. To To do do so, so, the the regional regional-scale-scale monthly monthly and and annual annual rainfall rainfall standardize standardizedd index index was was developeddeveloped to to the the inter inter-annual-annual scale. scale. ToTo f furtherurther clarify clarify the the atmospheric atmospheric circulation circulation and and its possible its possible association association with withmonsoon mon- rainfall,soon rainfall, we assessed we assessed the pressure the pressure vertical velocity vertical (ω) velocity for the (ω wet) for and the dry wet periods and dry along periods lati- ◦ ◦ ◦ tudealong 15°, latitude 20°, and 15 25°, 20N, as, and shown 25 inN, Figure as shown 11. The in Figureresults reveal11. The that results upward reveal (rising) that motion upwards ◦ ◦ over(rising) the region motions were over significant the region positive were anomal significanties at 15° positive and 20° anomalies N (Figure at 11a,c 15 ,e).and Sinking 20 N motion(Figure occurred 11a,c,e). Sinkingover the motion region occurredduring dry over years, the regioncharacterized during dryby the years, significant characterized posi- ◦ ◦ tiveby theanomaly significant at 15° positive and 20° anomaly N (Figure at 15 11b,d,f).and 20 TheN (Figuredifference 11b,d,f). in wet The minus difference dry period in wet moistureminus dry transport period moistureat 850 hPa transport revealed at that 850 anomalous hPa revealed moisture that anomalous convergence moisture (positive con- anomalies)vergence (positive emerge anomalies)in the central emerge and northern in the central region ands where northern moisture regions comes where from moisture BoB andcomes the fromGulf of BoB Thailand and the (Figure Gulf of S1). Thailand We also (Figure identified S1). a We relationship also identified between a relationship the distri- butionbetween of air the space distribution at lower of and air spacehigher at levels lower and and the higher flow levelsof moi andsture the transport. flow of moisture An un- familiartransport. airfield An unfamiliar at low levels airfield aligned at low with levels the location aligned withof a large the location junction of of a largethe complete junction flowof the of completemoisture in flow a stormy of moisture area. Figure in a stormy S2 shows area. the Figure distribution S2 shows of field the streams distribution for the of diversityfield streams of summer for the varieties diversity for of summerwet drying varieties seasons for at wet 850 drying hPa. The seasons cyclonic at 850 broadcast hPa. The cyclonic broadcast in Southern Myanmar during the period 1950–2019 occurred in the in Southern Myanmar during the period◦ 1950–2019 occurred in the southwest, with its centersouthwest, around with 10° its N center(Taninthary around region). 10 N (TanintharyTherefore, more region). water Therefore, vapor enter moreed water the region, vapor entered the region, especially since the convergence center of the corresponding humidity especially since the convergence center of the corresponding humidity was associated was associated with a clear climate zone that cuts the region, increasing rainfall in that with a clear climate zone that cuts the region, increasing rainfall in that region. In these region. In these cases, water vapor from the BoB and the Andaman Sea was transported cases, water vapor from the BoB and the Andaman Sea was transported into the land. into the land.

Water 2021, 13, 729 16 of 27 Water 2021, 13, x FOR PEER REVIEW 16 of 28

FigureFigure 11. 11.Pressure Pressure vertical vertical velocity velocity (ω (ω))(× (×1010−−33 Pa s−−1)1 )over over Myanmar Myanmar during during wet wet years years (1962 (1962–1967,–1967, 1998, 1998, and and 2003 2003–2004)–2004) andand dry dry years years (1978–1986 (1978–1986 and and 1988 1988 –1990): –1990): (a) wet(a) wet years years at 15 at◦ 15°N, (N,b) ( dryb) dry years years at 15 at◦ 15°N, (N,c) wet(c) wet years years at 20 at◦ 20°N, (N,d) ( dryd) dry years years at 20° N, (e) wet years at 25° N, and (f) dry years at 25° N. The red (blue) color indicates a positive (negative) signif- at 20◦ N, (e) wet years at 25◦ N, and (f) dry years at 25◦ N. The red (blue) color indicates a positive (negative) significant icant anomaly. The dashed (solid) lines show ascending (descending) motion. Negative (positive) values indicate an up- anomaly.ward (downward) The dashed motion. (solid) lines show ascending (descending) motion. Negative (positive) values indicate an upward (downward) motion. 4.3.3. Role of Oceans 4.3.3. Role of Oceans From the composite analysis, we elucidated that large-scale atmospheric circulation From the composite analysis, we elucidated that large-scale atmospheric circulation had had less (not-significant) influence on the summer monsoon rainfall variability over My- lessanmar (not-significant). The observed influence relationship on the summerwas rather monsoon weaker rainfall and could variability not be over associated Myanmar. with The observedrainfall variability relationship over was the rather study weaker region. and However, could not the be composite associated analysis with rainfall revealed variability that overthe theanomalous study region. wind pattern However, led the the rainfall composite variability analysis due revealed to the regional that the rainfall anomalous depend- wind patternence on led moisture/water the rainfall variability vapor. The due trend to the in regionalrainfall was rainfall obvious dependence but the driver on moisture/water of the trend vapor.appeared The trendto be complex in rainfall; a was simple obvious mean but trend the may driver not of provide the trend enough appeared explanation to be complex; since a simple mean trend may not provide enough explanation since sub-seasonal variability may balance each other, which may smooth or even remove the trend.

Water 2021, 13, x FOR PEER REVIEW 17 of 28

Water 2021, 13, 729 17 of 27 sub-seasonal variability may balance each other, which may smooth or even remove the trend. FigureFigure 12 12 reveals reveals a correlationa correlation map map between between the the mean mean May–October May–October SST SST (over (over the Indian the In- Oceandian Ocean and the and Pacific the Pacific Ocean) Ocean) and mean and May–Octobermean May–October rainfall rainfall over Myanmar. over Myanmar. A significant A sig- correlationnificant correlation (hatched regions)(hatched is regions) displayed is atdisplayed the 95% confidenceat the 95% levelconfidenc in Figuree level 12 .in The Figure results 12. showThe results a strong show negative a strong correlation negative between correlation rainfall between over rainfall Myanmar over and Myanmar Indian Ocean and Indian SST, whichOcean means SST, thatwhich more means (less) that rainfall more was (less) recorded rainfall over was Myanmar recorded and over with Myanmar cooling (warming) and with overcooling the Indian(warming) Ocean. over The the results Indian further Ocean. indicate The results that the further air temperature indicate that in the air northwest temper- partature of thein the SEA northwest was obviously part of higher the SEA than was in the obviously southeast higher part of than the region, in the whichsoutheast may part affect of thethe indicator region, which of regional may affect extreme the temperature. indicator of regional Interestingly, extreme we detectedtemperature. increasing Interestingly, trends inwe warm detected temperature increasing events trends in the in northernwarm temperatur part of thee region,events in in agreement the northern with part previous of the studiesregion, in in the agreement target regions with [previous81,88]. Kreft studies and Ecksteinin the target (2013b) regions examined [81,88] the. Kreft monsoon and Eckstein rainfall over(2013b) Myanmar examined and foundthe monsoon a negative rainfall correlation over Myanmar with the positive and found phase a ofnegative the Indian correlat Oceanion dipolewith the (IOD). positive phase of the Indian Ocean dipole (IOD).

FigureFigure 12. 12Correlation. Correlation map map between between the the mean mean decadal decadal May–October May–October SST (overSST (over the Indian the Indian Ocean Ocean and Pacific and Pacific Ocean) Ocean) and mean and decadalmean decadal May–October May–October rainfall over rainfall Myanmar. over Myanmar. Hatched regionsHatched show regions a significant show a significant correlation correlation at the 95% confidenceat the 95% level.confidence level. This instability in the increase/decrease in SST and maximum/minimum rainfall may be associatedThis instability with the in increasing the increase/decrease (decreasing) in trend SST inand temperature maximum/minimum over the target rainfall region, may whichbe associated might affect with thethe extremeincreasing temperature (decreasing) occurrences trend in temperature in the study over area. the The target increase region, in frequencywhich might and magnitudeaffect the extreme of daily temperature extreme warm occurrences temperature in andthe decreasestudy area. in extreme The increase cold temperaturein frequency may and occurmagnitude on the of global daily scope,extreme resulting warm temperature in the increase and in decrease length, frequency,in extreme and/orcold temperature intensity of may warm occur periods on the or heatglobal waves scope, for resulting most land in areasthe increase [4,33]. Strongin length, winds fre- canquency, transport and/or water intensity vapor of from warm the periods ocean to or nearby heat waves coastal for areas, most which land canareas affect [4,33] regional. Strong temperatureswinds can transport and ultimately water vapor affect from the cooling the ocean trends, to ne andarby vice coastal versa areas, [4,5]. which can affect regionalTheinterdecadal temperatures variability and ultimately in the 10affect year the running cooling mean trends, in Myanmar and vice rainfallversa [4,5] anomaly. was computedThe interdecadal using standardized variability in deviation,the 10 year PDO, running and mean AMO in indices, Myanmar with rainfall the results anom- providedaly was computed in Figure 14 using. The standardized effect of positive deviation PDO was, PDO low, and rainfall AMO (dry/drought), indices, with the and results that ofprovided negative in PDO Figure was 13. more The rainfalleffect of (wet/flood) positive PDO in was the regionlow rainfall (Figure (dry/drought), 14). The PDO and index that wasof negative statistically PDO significant was more with rainfall a strong (wet/flood) negative in correlation the region with (Figure rainfall 13). (–0.79),The PDO and index the AMOwas statistically index correlation significant coefficient with a strong explained negative the positive correlation correlation with rainfall (0.53), (– as0.79), shown and inthe TableAMO4. index The correlation correlation coefficients coefficient betweenexplained the the Myanmar positive correlation rainfall and (0.53) various, as shown climate in indicesTable 4. for The the correlation interannual coefficients and interdecadal between components the Myanmar are rainfall presented and invarious Table4 climate. The correlationindices for coefficientsthe interannual between and PDOinterdecadal and AMO components indices and are Myanmar presented rainfall in Table for 4. both The thecorrelation interannual coefficients and interdecadal between componentsPDO and AMO were indices significant and at Myanmar the 95% confidence rainfall for level. both The Myanmar rainfall showed a highly negative correlation with the PDO index and a positive correlation with the AMO index, particularly for the interdecadal component. The

Water 2021, 13, 729 18 of 27

cumulative role of these large-scale indices, especially during monsoon season, is perceived as affecting the SEA monsoon circulation pattern and thus resulting in dry/wet conditions during El Niño/La Niña years. The findings of the current study showed no obvious or significant association between ENSO, PDO, and monsoon onset timing with rainfall variability over Myanmar. The cold phase of the AMO is generally associated with negative Myanmar rainfall anomalies, which is lower than normal rainfall. The warming or positive phase PDO may decrease rainfall (drought) conditions, and cooling or negative-phase PDO may increase rainfall (flood) conditions over Myanmar. Moreover, there were nine extreme years (1952, 1959, 1961, 1965, 1970, 1973, 1974, 1999, and 2001), as shown in Figure 13. The results of extreme rainfall measures during the negative phases of the PDO show that no extreme events occurred during the positive phase of the PDO (Figure 13). Various studies of linearly coupled dominating patterns between the global SST and rainfall variations in Myanmar related to the interdecadal variability have been conducted using the SVD method [4,31,37,81]. This method has successfully been applied in the region to identify the interannual variability in summer monsoon rainfall [89]. Figure 15 indicates the heterogeneous correlation of the first three SVD modes amongst global SST and Myanmar rainfall. The SVD of the SST modes closely represents the long-term global warming pattern (Figure 15a,e), with a correlation coefficient of 0.94 between the time series of its principal components and the low-pass-filtered global SST anomaly. The southern and eastern regions of Myanmar and regions of the mid-west extending up to Indian Ocean are strongly correlated with SST in the tropical and northern Pacific (Figure 15a,b). We found subtle differences in the global spatial structure of the tropical Pacific SST correlation pattern in this period (Figure 15c). Although it is unclear if the differences in the global SST spatial patterns for the three modes regional rainfall are statistically significant, they do provide some insights. The findings reported using heterogeneous correlation patterns [11] as the first SVD of SST mode 1 were mainly associated with the cold phase of PDO in the Pacific Ocean, north of 20◦ N, and the warm phase of AMO in the North Atlantic Ocean (Figure 15a). The rainfall mode 1 explains much of the rainfall over the regions except for the central core and eastern regions (Figure 15b). The squared covariance fraction of this mode is about 53%, and the correlation coefficient between the two fields is 0.94. SVD2 of SST mode 2 demonstrated a warm SST in the eastern equatorial Pacific SST, like the El Niño SST pattern in the Pacific Ocean, and a cold SST in the North Atlantic Ocean (Figure 15c). The result show that relatively low rainfall over the region contributed to rainfall mode 2 (Figure 15d). Thus, the El Niño SST pattern in the Pacific Ocean and cold phase of the AMO were associated with dry/drought conditions over the region. The time series of the coefficients for the three SVD modes of global ocean SST and Myanmar rainfall are presented in Figure 16. The squared covariance fraction of SVD mode 2 was detected at a rate of 33%, and the correlation coefficient between the two fields’ time coefficients is around 0.94. SST mode 3 explains cold SSTs in the central and eastern equatorial Pacific SSTs as a La Niña SST pattern (Figure 15e). Rainfall mode 3 exhibits enhanced rainfall over the region (Figure 15f). SVD3 of the squared covariance fraction of this mode was detected with a 6.7% rate, and the correlation coefficient between the two fields is 0.93 (Table5). Therefore, we inferred that the role of the oceanic circulation is important for decadal Myanmar rainfall.

Table 4. Correlation coefficients between various indices and Myanmar rainfall for interannual and interdecadal components. * Significant values at the 95% confidence level. NP, North Pacific index; NAO, North Atlantic oscillation index; AO, Arctic oscillation.

Indices Interannual Component Interdecadal Component NP –0.24 –0.17 NAO –0.38 * –0.15 AO –0.02 –0.21 PDO –0.39 * –0.79 * AMO 0.53 * 0.54 * Water 2021, 13, 729 19 of 27

Table 5. Statistics of the leading singular value decomposition (SVD) sea surface temperature (SST) and rainfall modes over the study area.

Squared Temporal Rainfall Mode SST Variance Covariance Correlation Variance Mode 1 53% 0.94 47% 20% Mode 2 33% 0.94 21% 28% Water 2021, 13, x FOR PEER REVIEW 19 of 28 Mode 3 6.7% 0.93 9% 13%

Water 2021, 13, x FOR PEER REVIEW 18 of 28

the interannual and interdecadal components were significant at the 95% confidence level. The Myanmar rainfall showed a highly negative correlation with the PDO index and a positive correlation with the AMO index, particularly for the interdecadal component. The cumulative role of these large-scale indices, especially during monsoon season, is per- ceived as affecting the SEA monsoon circulation pattern and thus resulting in dry/wet conditions during El Niño/La Niña years. The findings of the current study showed no obvious or significant association between ENSO, PDO, and monsoon onset timing with rainfall variability over Myanmar. The cold phase of the AMO is generally associated with negative Myanmar rainfall anomalies, which is lower than normal rainfall. The warming or positive phase PDO may decrease rainfall (drought) conditions, and cooling or nega- tive-phase PDO may increase rainfall (flood) conditions over Myanmar. Moreover, there were nine extreme years (1952, 1959, 1961, 1965, 1970, 1973, 1974, 1999, and 2001), as shown in Figure 14. The results of extreme rainfall measures during the negative phases of the PDO show that no extreme events occurred during the positive phase of the PDO Figure 14. Time series of the normalized summer monsoon rainfall in Myanmar (1950–2010). The blue bars indicate the Figure 13. Time series of( theFigure normalized 14). summer monsoon rainfall in Myanmar (1950–2010). The blue bars indicate the multidecadalmultidecadal variability variability (MDV) (MDV) in the in PDOthe PDO index index obtained obtained using using the ensemble the ensemble empirical empirical mode decomposition mode decomposition (EEMD) method. (EEMD) method.

Table 4. Correlation coefficients between various indices and Myanmar rainfall for interannual and interdecadal components. * Significant values at the 95% confidence level. NP, North Pacific index; NAO, North Atlantic oscillation index; AO, Arctic oscillation.

Indices Interannual Component Interdecadal Component NP –0.24 –0.17 NAO –0.38 * –0.15 AO –0.02 –0.21 PDO –0.39 * –0.79 * AMO 0.53 * 0.54 *

Various studies of linearly coupled dominating patterns between the global SST and rainfall variations in Myanmar related to the interdecadal variability have been conducted using the SVD method [4,31,37,81]. This method has successfully been applied in the re- gion to identify the interannual variability in summer monsoon rainfall [89]. Figure 15 indicates the heterogeneous correlation of the first three SVD modes amongst global SST Figure 14. Interdecadal variability in average 10 year running mean summer monsoon rainfall anomaly based on standard Figure 13. Interdecadal variabilityand inMyanmar average 10 rainfall. year running The SVD mean of summer the SST monsoon modes closelyrainfall anomalyrepresents based the on long standard-term global deviation, Pacific decadal oscillation (PDO), and Atlantic multidecadal oscillation (AMO) indices. deviation, Pacific decadal oscillationwarming (PDO pattern), and Atlantic (Figure multidecadal 15a,e), with oscillation a correlation (AMO coefficient) indices. of 0.94 between the time series of its principal components and the low-pass-filtered global SST anomaly. The southern and eastern regions of Myanmar and regions of the mid-west extending up to Indian Ocean are strongly correlated with SST in the tropical and northern Pacific (Figure 15a,b). We found subtle differences in the global spatial structure of the tropical Pacific SST correlation pattern in this period (Figure 15c). Although it is unclear if the differences in the global SST spatial patterns for the three modes regional rainfall are statistically sig- nificant, they do provide some insights. The findings reported using heterogeneous cor- relation patterns [11] as the first SVD of SST mode 1 were mainly associated with the cold phase of PDO in the Pacific Ocean, north of 20° N, and the warm phase of AMO in the North Atlantic Ocean (Figure 15a). The rainfall mode 1 explains much of the rainfall over the regions except for the central core and eastern regions (Figure 15b). The squared co- variance fraction of this mode is about 53%, and the correlation coefficient between the two fields is 0.94. SVD2 of SST mode 2 demonstrated a warm SST in the eastern equatorial Pacific SST, like the El Niño SST pattern in the Pacific Ocean, and a cold SST in the North Atlantic Ocean (Figure 15c). The result show that relatively low rainfall over the region

Water 2021, 13, 729 20 of 27 Water 2021, 13, x FOR PEER REVIEW 21 of 28

FigureFigure 15. 15. HeterogeneousHeterogeneous correlations correlations of of the the first first three three SVD SVD modes modes among among global global SST SST and and Myanmar Myanmar rainfall rainfall:: ( (aa)) SST SST mode mode 1,1, ( (bb)) rainfall rainfall mode mode 1, 1, ( (cc)) SST SST mode mode 2, 2, ( (dd)) rainfall rainfall mode mode 2, 2, ( (ee)) SST SST mode mode 3, 3, and and ( (ff)) rainfall rainfall mode mode 3. 3.

WaterWater2021 2021, 13, 13, 729, x FOR PEER REVIEW 2122 ofof 2728

FigureFigure 16. 16.Normalized Normalized time time series series of of the the first first three three SVD SVD modes: modes: ( a(a)) mode mode 1, 1, ( b(b)) mode mode 2, 2, and and ( c()c) mode mode 3 3 SST SST (red (red bars) bars) and and rainfallrainfall (black (black bars). bars).

5.5. DiscussionDiscussion InIn this this study, study we, we evaluated evaluated the the interdecadal interdecadal variability variability in thein the summer summer monsoon monsoon rainfall rain- (May–October)fall (May–October) in Myanmar in Myanmar (1950–2019) (1950– using2019) EOF, using SVD, EOF, EEMD, SVD, and EEMD, correlation and correlation analyses basedanalyses on 10based year on running 10 year means. running This mea studyns. This is the study first is of the its first kind of to its use kind a large to use number a large ofnumber statistical of statistical approaches approaches to investigate to investigate the interdecadal the interdecadal variability variability in monsoon in monsoon rainfall, whichrainfall is, beneficialwhich is beneficial for the scientific for the community scientific community and climate and change climate adaptation change adaptation in Myanmar. in TheMyanmar. results revealThe results that thereveal dominant that the summer dominant monsoon summer regions monsoon (south, regions central, (south, and central, east) observedand east) dryobserved conditions dry condition during 1998s during and 2002.1998 and Sein 2002. et al., Sein (2015) et al. described (2015) described the lowest the rainfalllowest over rainfall Myanmar over Myanmar since 1997, since and the 1997 most, and intensive the most heat intensive wave (60%) heat occurred wave (60%) in 1998 oc- duringcurred anin ENSO1998 during year. Websteran ENSO and year. Yang Webster (1992) and [90] Yang showed (1992) that [90 1998] showed was the that driest 1998 year was duringthe driest the year period during of 1950–2019, the period with of 1950 extreme–2019 temperature, with extreme recorded temperature in May recorded 1998 (43.6 in ◦MayC). Our1998 results (43.6 °C). agree Our with results the findings agree with of previousthe findings studies of previous [27], reporting studies similar [27], report resultsing over sim- theilar target results regions. over the Wetarget used regions. EOF analysis We used with EOF theanalysis 10 year with running the 10 year means running of summer means monsoonof summer rainfall, monsoon and rainfall, the results and show the results that the show first that leading the first EOF leading pattern EOF explained pattern 47% ex- ofplained the variance. 47% of Thethe maximumvariance. The loading maximum occurs loading along the occurs western along coast the andwestern the minimum coast and

Water 2021, 13, 729 22 of 27

in the northwest region, while the corresponding PC1 exhibits a dipole pattern. Our results support the findings of Zhang et al., (2014), who reported similar results over South China using a drought index. The circulation patterns on an interdecadal scale 2343 analyzed using the composites (dry and wet years) in this study. Interdecadal variability in circulation mode consists of upward (rising) motion over the region, significantly positive anomaly over central Myanmar. The southwest moisture flux comes from the north Indian Ocean across the entire west coast of Myanmar. Reckien and Petkova (2019) [91] studied the first component of EOF of the vertically integrated moisture transport vector in summer, and strong northward moisture transport from the tropical Indian and West Pacific Oceans dominated the whole East Asian region. In this study, PDO index values were found to be significantly related to Myanmar summer monsoon rainfall (–0.79) during the 1950–2019 period. The result show a positive correlation with phase changes in the PDO, but negative PDO change was found with lower summer monsoon rainfall over Myanmar, which agrees with Roy et al., (2011) and Kumar et al., (2006). Lone et al., (2019) investigated the ENSO, PDO, and the summer monsoon rainfall in Myanmar over 52 years (1951–2002). They found that the overall negative relationship between ENSO and rainfall patterns had a strong positive association across Myanmar. The role of ENSO in the resulting rainfall patterns during El Niño years was stronger during the cold phase of the PDO compared with during La Niña years. The difference between rainfall during the different phases of ENSO was superimposed with the warm and cold periods of the PDO, showing overall higher average annual rainfall during cold PDO episodes [92]. The results reveal that the Indian monsoon is more vulnerable to drought when El Niño events occur during warm phases of the Pacific interdecadal variability. Conversely, wet monsoons are more likely to prevail when La Niña events coincide during cold phases of the Pacific interdecadal variability. Based on the EEMD analysis, highly extreme summer monsoon rainfall (i.e., 1952, 1959, 1961, 1965, 1970, 1973, 1974, 1999, and 2001) occurs during the negative phases of the PDO; no extreme events occurred during the positive phase of the PDO. Our results support the findings of Chen and Huang (2017), who described that between 1951 and 2012, there were five extreme events that had a strong correlation with the negative phases of the PDO. The AMO index was significantly associated with Myanmar summer monsoon rainfall recorded, at 0.54. Here, to determine the interdecadal summer monsoon rainfall modes and SST, a further SVD analysis was performed. The results of SVD mode 1 explain that the positive phase of the AMO receives higher summer monsoon rainfall, except in central and east regions in Myanmar. The positive phase of the AMO significantly correlated with higher summer monsoon rainfall and negatively correlated with lower summer monsoon rainfall over Myanmar, which is in agreement with Sein et al., (2015). Our findings do not negate the potential impact of large-scale oceanic indices but rather postulate the hypothesis that these controls need to be reviewed with clear links to summer monsoon rainfall variability over Myanmar. Another possible reason could be the quality of station data, since they are recorded for synoptic-scale use and climatological processes usually occur on longer time scale; thus, the data may lack such large-scale signals. These are some of the points that need to be thoroughly investigated in future studies, including the frequency and intensity of extreme rainfall events and their role as drivers of regional rainfall variability.

6. Conclusions In this study, we attempted to explore the spatiotemporal trend and variability in monsoon rainfall during the period of 1950–2019 over Myanmar. Using statistical and composite analyses, we found significant variation in rainfall on seasonal, annual, and interdecadal scales. The main conclusions based on the findings are as follows: Water 2021, 13, 729 23 of 27

(1) The interdecadal variation over the target region increased in the late 1950s, reaching its peak around 1965, and subsequently decreased in the mid-1980s before increasing again in the early 1990s. An abrupt rainfall shift was recorded during the 1970s and a significant reduction (at the 95% confidence level) in rainfall was evident between the 1980s and the 2000s. (2) In terms of rainfall trend, the regional variability in monsoon rainfall over Myanmar was obvious during the 1950–2019 period. The magnitude of the rainfall variability could possibly be associated with the geographical location of the region and monsoon circulation patterns. The temporal trend over the Asian and Indian monsoon systems showed strong variation on different time scales, including monthly, inter-annual, intra-seasonal, and annual time scales. Such large-scale variation caused by different factors could induce changes in monsoon rainfall over the diverse study regions. (3) The widely used EOF method showed that the interdecadal rainfall pattern over Myanmar significantly changed over the study period, especially during EOF2 and EOF3. PC1 and PC2 experienced an increasing trend over the target region during the period 1950–2019. (4) The results further demonstrate the significant negative westerly and southwesterly wind anomalies from the BoB and Andaman Sea over almost the whole region except for the eastern and southern regions during wet years. The moisture transport at 850 hPa revealed that the anomalous moisture convergence (positive anomalies) emerged in the central and northern region, transporting moisture from the Western BoB and Eastern Gulf of Thailand. This cyclonic circulation may cause more water vapor from the BoB and Andaman Sea to be transported into the country. In addition, warm or positive PDO phase was associated with low rainfall in Myanmar, and cooling or negative PDO phase can increase rainfall over Myanmar. The AMO index had a mean average value of 0.53 for the same period. In addition, the effects of excessive rainfall occurred in the negative stages of the PDO, and no serious adverse events occurred in the positive phase of the PDO. The first SVD of SST mode 1 was mainly linked to the PDO cold phase in the Pacific Ocean above the North Pacific Ocean (20◦ N) and the warm AMO phase in the North Atlantic Ocean. (5) The SST patterns showed that the warm SST of the eastern equatorial Pacific SST is a pattern of the El Niño SST in the Pacific Ocean and the cold SST in the North Atlantic Ocean. The fractional covariance fraction of SVD2 was recorded as 33%, whereas the coefficient of correlation between the coefficients for the two-phase period was 0.94. Therefore, we concluded that the findings of the present study provide valuable results regarding the interdecadal variability in Myanmar summer monsoon rainfall, which is shown as a negative correlation with the PDO and a positive association with the AMO. In addition, Myanmar’s decadal summer monsoon rainfall and relationships with the PDO and AMO indexes explain the flood and dry conditions (drought) in the region.

Supplementary Materials: The following are available online at https://www.mdpi.com/2073-4 441/13/5/729/s1, Figure S1: Difference fields of wet period (1962–1967, 1998, 2003–2004) minus dry period (1978–1986, 1988–1990) of moisture transport at 850 hPa (g kg−1 ms−1) over Myanmar computed from ERA-interim dataset. Vector shows moisture transport, while the shaded regions specify convergence (positive) and divergence (negative) moisture fluxes convergence., Figure S2: Difference stream fields of wet period (1962–1967, 1998, 2003–2004) minus dry period (1978–1986, 1988–1990) at 850 hPa in summer. Author Contributions: Conceptualization, Z.M.M.S., I.U., and X.Z.; Methodology, Software, and Validation, Z.M.M.S. and X.Z.; Formal Analysis and Investigation, I.U. and Z.M.M.S.; Writing— Review and Editing, I.U., F.S. and S.S.; Data curation, K.A., S.S. and Z.M.M.S. All authors have read and agreed to the published version of the manuscript. Funding: This study supported by the National (Key) Basic R&D Program of China (grant no. 2012CB955204). Water 2021, 13, 729 24 of 27

Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not Applicable. Data Availability Statement: Not Applicable. Acknowledgments: We thank the two anonymous reviewers for their time and support in re- shaping the manuscript. Their comments helped us to improve the final version of the manuscript. This research was encouraged by Binjiang College, Nanjing University of Information Science and Technology, Wuxi City, Jiangsu Province, China. Special appreciation to the Department of Meteorology and Hydrology, Myanmar, for the provision of the datasets used in the study. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Goswami, B.N.; Madhusoodanan, M.S.; Neema, C.P.; Sengupta, D. A physical mechanism for North Atlantic SST influence on the Indian summer monsoon. Geophys. Res. Lett. 2006, 33.[CrossRef] 2. Saleem, F.; Zeng, X.; Hina, S.; Omer, A. Regional changes in extreme temperature records over Pakistan and their relation to Pacific variability. Atmos. Res. 2021, 250, 105407. [CrossRef] 3. Hina, S.; Saleem, F. Historical analysis (1981–2017) of drought severity and magnitude over a predominantly arid region of Pakistan. Clim. Res. 2019, 78, 189–204. [CrossRef] 4. Suman, M.; Maity, R. Southward shift of precipitation extremes over south Asia: Evidences from CORDEX data. Sci. Rep. 2020, 10, 1–11. [CrossRef] 5. Ren, Y.-Y.; Ren, G.-Y.; Sun, X.-B.; Shrestha, A.B.; You, Q.-L.; Zhan, Y.-J.; Rajbhandari, R.; Zhang, P.-F.; Wen, K.-M. Observed changes in surface air temperature and precipitation in the Hindu Kush Himalayan region over the last 100-plus years. Adv. Clim. Chang. Res. 2017, 8, 148–156. [CrossRef] 6. Lone, S.A.; Jeelani, G.; Deshpande, R.D.; Mukherjee, A. Stable isotope (δ18O and δD) dynamics of precipitation in a high altitude Himalayan cold desert and its surroundings in Indus river basin, Ladakh. Atmos. Res. 2019, 221, 46–57. [CrossRef] 7. Zhi, X.; Tian, X.; Liu, P.; Hu, Y. Interdecadal variations in winter extratropical anticyclones in East Asia and their impacts on the decadal mode of East Asian surface air temperature. Theor. Appl. Clim. 2019, 131, 1763–1775. [CrossRef] 8. Hou, M.; Duan, W.; Zhi, X. Season-dependent predictability barrier for two types of El Niño revealed by an approach to data analysis for predictability. Clim. Dyn. 2019, 53, 5561–5581. [CrossRef] 9. Tangang, F.; Chung, J.X.; Juneng, L.; Supari; Salimun, E.; Ngai, S.T.; Jamaluddin, A.F.; Mohd, M.S.F.; Cruz, F.; Narisma, G.; et al. Projected future changes in rainfall in Southeast Asia based on CORDEX–SEA multi-model simulations. Clim. Dyn. 2020, 55, 1247–1267. [CrossRef] 10. Almazroui, M.; Saeed, S.; Saeed, F.; Islam, M.N.; Ismail, M. Projections of Precipitation and Temperature over the South Asian Countries in CMIP6. Earth Syst. Environ. 2020, 4, 297–320. [CrossRef] 11. Ge, F.; Zhu, S.; Luo, H.; Zhi, X.; Wang, H. Future changes in precipitation extremes over Southeast Asia: Insights from CMIP6 multi-model ensemble. Environ. Res. Lett. 2021, 16, 024013. [CrossRef] 12. Song, B.; Zhi, X.; Pan, M.; Hou, M.; He, C.; Fraedrich, K. Turbulent Heat Flux Reconstruction in the North Pacific from 1921 to 2014. J. Meteorol. Soc. Jpn. 2019, 97, 893–911. [CrossRef] 13. Francis, R.C.; Hare, S.R. Decadal-scale regime shifts in the large marine ecosystems of the North-east Pacific: A case for historical science. Fish. Oceanogr. 1994, 3, 279–291. [CrossRef] 14. Latif, M.; Barnett, T.P. Causes of Decadal Climate Variability over the North Pacific and North America. Science 1994, 266, 634–637. [CrossRef] 15. Yang, Y.; Gan, T.Y.; Tan, X. Spatiotemporal Changes in Precipitation Extremes over Canada and Their Teleconnections to Large-Scale Climate Patterns. J. Hydrometeorol. 2019, 20, 275–296. [CrossRef] 16. Ge, F.; Zhu, S.; Peng, T.; Zhao, Y.; Sielmann, F.; Fraedrich, K.; Zhi, X.; Liu, X.; Tang, W.; Ji, L. Risks of precipitation extremes over Southeast Asia: Does 1.5 ◦C or 2 ◦C global warming make a difference? Environ. Res. Lett. 2019, 14, 044015. [CrossRef] 17. Ge, F.; Zhi, X.; Babar, Z.A.; Tang, W.; Chen, P. Interannual variability of summer monsoon precipitation over the Indochina Peninsula in association with ENSO. Theor. Appl. Clim. 2016, 128, 523–531. [CrossRef] 18. Liu, B.; Wu, G.; Ren, R. Influences of ENSO on the vertical coupling of atmospheric circulation during the onset of South Asian summer monsoon. Clim. Dyn. 2014, 45, 1859–1875. [CrossRef] 19. Xu, J.; Koldunov, N.V.; Remedio, A.R.C.; Sein, D.V.; Rechid, D.; Zhi, X.; Jiang, X.; Xu, M.; Zhu, X.; Fraedrich, K.; et al. Downstream effect of Hengduan Mountains on East China in the REMO regional climate model. Theor. Appl. Clim. 2018, 135, 1641–1658. [CrossRef] 20. Zhang, L.; Zhu, X.; Fraedrich, K.; Sielmann, F.; Zhi, X. Interdecadal variability of winter precipitation in Southeast China. Clim. Dyn. 2014, 43, 2239–2248. [CrossRef] 21. Zheng, Y.; Zhang, Q.; Luo, M.; Sun, P.; Singh, V.P. Wintertime precipitation in eastern China and relation to the Madden-Julian oscillation: Spatiotemporal properties, impacts and causes. J. Hydrol. 2020, 582, 124477. [CrossRef] Water 2021, 13, 729 25 of 27

22. Mantua, N.J.; Hare, S.R.; Zhang, Y.; Wallace, J.M.; Francis, R.C. A Pacific Interdecadal Climate Oscillation with Impacts on Salmon Production. Bull. Am. Meteorol. Soc. 1997, 1069–1079. [CrossRef] 23. Zhang, L.; Sielmann, F.; Fraedrich, K.; Zhu, X.; Zhi, X. Variability of winter extreme precipitation in Southeast China: Contributions of SST anomalies. Clim. Dyn. 2015, 45, 2557–2570. [CrossRef] 24. Chhin, R.; Shwe, M.M.; Yoden, S. Time-lagged correlations associated with interannual variations of pre-monsoon and post- monsoon precipitation in Myanmar and the Indochina Peninsula. Int. J. Clim. 2020, 40, 3792–3812. [CrossRef] 25. Sein, Z.M.M.; Islam, A.R.M.T.; Maw, K.W.; Moya, T.B. Characterization of southwest monsoon onset over Myanmar. Theor. Appl. Clim. 2015, 127, 587–603. [CrossRef] 26. Burki, T. Floods in Myanmar damage hundreds of health facilities. Lancet 2015, 386, 843. [CrossRef] 27. Chen, F.-H.; Huang, W. Multi-scale climate variations in the arid Central Asia. Adv. Clim. Chang. Res. 2017, 8, 1–2. [CrossRef] 28. Zaw, Z.; Fan, Z.-X.; Bräuning, A.; Liu, W.; Gaire, N.P.; Than, K.Z.; Panthi, S. Monsoon precipitation variations in Myanmar since AD 1770: Linkage to tropical ocean-atmospheric circulations. Clim. Dyn. 2021, 1–16. [CrossRef] 29. Ge, F.; Peng, T.; Fraedrich, K.; Sielmann, F.; Zhu, X.; Zhi, X.; Liu, X.; Tang, W.; Zhao, P. Assessment of trends and variability in surface air temperature on multiple high-resolution datasets over the Indochina Peninsula. Theor. Appl. Clim. 2019, 135, 1609–1627. [CrossRef] 30. Xu, J.; Koldunov, N.; Remedio, A.R.C.; Sein, D.V.; Zhi, X.; Jiang, X.; Xu, M.; Zhu, X.; Fraedrich, K.; Jacob, D. On the role of horizontal resolution over the Tibetan Plateau in the REMO regional climate model. Clim. Dyn. 2018, 51, 4525–4542. [CrossRef] 31. Kreft, S.; Eckstein, D. Global Climate Risk Index 2014: Who suffers most from extreme weather events? Weather-related loss events in 2012 and 1993 to 2012. Ger. Brief. Pap. 2013, 28. 32. Zhang, L.; Sielmann, F.; Fraedrich, K.; Zhi, X. Atmospheric response to Indian Ocean Dipole forcing: Changes of Southeast China winter precipitation under global warming. Clim. Dyn. 2016, 48, 1467–1482. [CrossRef] 33. Sen Roy, N.; Kaur, S. Climatology of monsoon rains of Myanmar (Burma). Int. J. Climatol. 2000, 913–928. [CrossRef] 34. Zaw, Z.; Fan, Z.; Bräuning, A.; Xu, C.; Liu, W.; Gaire, N.P.; Panthi, S.; Than, K.Z. Drought Reconstruction Over the Past Two Centuries in Southern Myanmar Using Teak Tree-Rings: Linkages to the Pacific and Indian Oceans. Geophys. Res. Lett. 2020, 47. [CrossRef] 35. Zhang, L.; Zhi, X.F. Multimodel consensus forecasting of low temperature and icy weather over central and Southern China in early 2008. J. Trop. Meteorol. 2015, 67–75. [CrossRef] 36. Sein, M.M.Z.; Ogwang, B.A.; Ongoma, V.; Ogou, F.K.; Batebana, K. Inter-annual variability of Summer Monsoon Rainfall over Myanmar in relation to IOD and ENSO. J. Environ. Agric. Sci. 2015, 4, 28–36. 37. Oo, S.S.; Hmwe, K.M.; Aung, N.N.; Su, A.A.; Soe, K.K.; Mon, T.L.; Lwin, K.M.; Thu, M.M.; Soe, T.T.; Htwe, M.L. Diversity of Insect Pest and Predator Species in Monsoon and Summer Rice Fields of Taungoo Environs, Myanmar. Adv. Èntomol. 2020, 8, 117–129. [CrossRef] 38. Omer, A.; Elagib, N.A.; Zhuguo, M.; Saleem, F.; Mohammed, A. Water scarcity in the Yellow River Basin under future climate change and human activities. Sci. Total. Environ. 2020, 749, 141446. [CrossRef] 39. Schneider, U.; Becker, A.; Finger, P.; Meyer-Christoffer, A.; Ziese, M.; Rudolf, B. GPCC’s new land surface precipitation climatology based on quality-controlled in situ data and its role in quantifying the global water cycle. Theor. Appl. Clim. 2014, 115, 15–40. [CrossRef] 40. Aadhar, S.; Mishra, V. A substantial rise in the area and population affected by dryness in South Asia under 1.5 ◦C, 2.0 ◦C and 2.5 ◦C warmer worlds. Environ. Res. Lett. 2019, 14, 114021. [CrossRef] 41. Ahmed, K.; Shahid, S.; Sachindra, D.; Nawaz, N.; Chung, E.-S. Fidelity assessment of general circulation model simulated precipitation and temperature over Pakistan using a feature selection method. J. Hydrol. 2019, 573, 281–298. [CrossRef] 42. Onyutha, C. Analyses of rainfall extremes in East Africa based on observations from rain gauges and climate change simulations by CORDEX RCMs. Clim. Dyn. 2020, 54, 4841–4864. [CrossRef] 43. Liu, Y.B.; Song, P.; Peng, J.; Fu, Q.N.; Dou, C.C. Recent increased frequency of drought events in Poyang Lake Basin, China: Climate change or anthropogenic effects? Hydro-Climatol. Var. Chang. 2011, 344, 99–104. 44. Khan, N.; Sachindra, D.; Shahid, S.; Ahmed, K.; Shiru, M.S.; Nawaz, N. Prediction of droughts over Pakistan using machine learning algorithms. Adv. Water Resour. 2020, 139, 103562. [CrossRef] 45. Zhang, T.; Lin, X. Assessing future drought impacts on yields based on historical irrigation reaction to drought for four major crops in Kansas. Sci. Total. Environ. 2016, 550, 851–860. [CrossRef] 46. Knight, J.H.; Minasny, B.; McBratney, A.B.; Koen, T.B.; Murphy, B.W. Soil temperature increase in eastern Australia for the past 50 years. Geoderma 2018, 313, 241–249. [CrossRef] 47. Rayner, N.A.; Parker, D.E.; Horton, E.B.; Folland, C.K.; Alexander, L.V.; Rowell, D.P.; Kent, E.C.; Kaplan, A.L. Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res. Space Phys. 2003, 108.[CrossRef] 48. Hersbach, H.; Bell, B.; Berrisford, P.; Horányi, A.; Sabater, J.M.; Nicolas, J.; Radu, R.; Schepers, D.; Simmons, A.; Soci, C.; et al. Global reanalysis: Goodbye ERA-Interim, hello ERA5. ECMWF Newsl. 2019, 17–24. [CrossRef] 49. Hersbach, H.; Bell, B.; Berrisford, P.; Hirahara, S.; Horányi, A.; Muñoz-Sabater, J.; Nicolas, J.; Peubey, C.; Radu, R.; Schepers, D.; et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. 2020, 146, 1999–2049. [CrossRef] Water 2021, 13, 729 26 of 27

50. Smith, T.M.; Reynolds, R.W.; Peterson, T.C.; Lawrimore, J. Improvements to NOAA’s Historical Merged Land–Ocean Surface Temperature Analysis (1880–2006). J. Clim. 2008, 21, 2283–2296. [CrossRef] 51. Kalnay, E.; Kanamitsu, M.; Kistler, R.; Collins, W.; Deaven, D.; Gandin, L.; Iredell, M.; Saha, S.; White, G.; Woollen, J.; et al. The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteorol. Soc. 1996, 37–472. [CrossRef] 52. Mardia, K.V. Multi-dimensional multivariate Gaussian Markov random fields with application to image processing. J. Multivar. Anal. 1988, 24, 265–284. [CrossRef] 53. Hannachi, A.; Jolliffe, I.T.; Stephenson, D.B. Empirical orthogonal functions and related techniques in atmospheric science: A review. Int. J. Clim. 2007, 27, 1119–1152. [CrossRef] 54. Levine, R.A.; Wilks, D.S. Statistical Methods in the Atmospheric Sciences. J. Am. Stat. Assoc. 2000, 95, 344. [CrossRef] 55. Walsh, J.E.; Mostek, A. A Quantitative Analysis of Meteorological Anomaly Patterns Over the United States, 1900–1977. Mon. Weather. Rev. 1980, 108, 615–630. [CrossRef] 56. Smakhtin, V.U.; Hughes, D.A. Automated estimation and analyses of meteorological drought characteristics from monthly rainfall data. Environ. Model. Softw. 2007, 22, 880–890. [CrossRef] 57. Gilman, D.L.; Fuglister, F.J.; Mitchell, J.M. On the Power Spectrum of “Red Noise”. J. Atmos. Sci. 1963, 182–184. [CrossRef] 58. Mitchell, J.M. Further Remarks on the Power Spectrum of “Red Noise”. J. Atmos. Sci. 1964, 461. [CrossRef] 59. Joshi, M.K.; Pandey, A.C. Trend and spectral analysis of rainfall over India during 1901–2000. J. Geophys. Res. Space Phys. 2011, 116.[CrossRef] 60. Zhu, S.; Ge, F.; Fan, Y.; Zhang, L.; Sielmann, F.; Fraedrich, K.; Zhi, X. Conspicuous temperature extremes over Southeast Asia: Seasonal variations under 1.5 ◦C and 2 ◦C global warming. Clim. Chang. 2020, 160, 343–360. [CrossRef] 61. Folland, C.K.; Knight, J.; Linderholm, H.W.; Fereday, D.; Ineson, S.; Hurrell, J.W. The Summer North Atlantic Oscillation: Past, Present, and Future. J. Clim. 2009, 22, 1082–1103. [CrossRef] 62. Wen, Z.; Niu, F.; Yu, Q.; Wang, D.; Feng, W.; Zheng, J. The role of rainfall in the thermal-moisture dynamics of the active layer at Beiluhe of Qinghai-Tibetan plateau. Environ. Earth Sci. 2013, 71, 1195–1204. [CrossRef] 63. Banacos, P.C.; Schultz, D.M. The Use of Moisture Flux Convergence in Forecasting Convective Initiation: Historical and Operational Perspectives. Weather Forecast. 2005, 20, 351–366. [CrossRef] 64. Mann, H.B. Nonparametric Tests against Trend. Econometrica 1945, 13, 245. [CrossRef] 65. Kendall, M.G. Rank Correlation Methods; Griffin: London, UK, 1975; ISBN 0852641990. 66. Ghosh, K.G. Spatial and temporal appraisal of drought jeopardy over the Gangetic West Bengal, eastern India. Geoenvironmental Disasters 2019, 6, 1. [CrossRef] 67. Waseem, M.; Ahmad, I.; Mujtaba, A.; Tayyab, M.; Si, C.; Lü, H.; Dong, X. Spatiotemporal Dynamics of Precipitation in Southwest Arid-Agriculture Zones of Pakistan. Sustainability 2020, 12, 2305. [CrossRef] 68. Payab, A.H.; Türker, U. Comparison of standardized meteorological indices for drought monitoring at northern part of Cyprus. Environ. Earth Sci. 2019, 78, 309. [CrossRef] 69. Naz, F.; Dars, G.H.; Ansari, K.; Jamro, S.; Krakauer, N.Y. Drought Trends in Balochistan. Water 2020, 12, 470. [CrossRef] 70. Musonda, B.; Jing, Y.; Iyakaremye, V.; Ojara, M. Analysis of Long-Term Variations of Drought Characteristics Using Standardized Precipitation Index over Zambia. Atmosphere 2020, 11, 1268. [CrossRef] 71. Wang, F.; Yang, H.; Wang, Z.; Zhang, Z.; Li, Z. Drought Evaluation with CMORPH Satellite Precipitation Data in the Yellow River Basin by Using Gridded Standardized Precipitation Evapotranspiration Index. Remote. Sens. 2019, 11, 485. [CrossRef] 72. Almazroui, M.; ¸Sen,Z. Trend Analyses Methodologies in Hydro-meteorological Records. Earth Syst. Environ. 2020, 4, 713–738. [CrossRef] 73. Ahmed, K.; Shahid, S.; Wang, X.; Nawaz, N.; Khan, N. Spatiotemporal changes in aridity of Pakistan during 1901–2016. Hydrol. Earth Syst. Sci. 2019, 23, 3081–3096. [CrossRef] 74. Rahman, G.; Dawood, M. Spatial and temporal variation of rainfall and drought in Khyber Pakhtunkhwa Province of Pakistan during 1971–2015. Arab. J. Geosci. 2018, 11, 46. [CrossRef] 75. Bretherton, C.S.; Smith, C.; Wallace, J.M. An Intercomparison of Methods for Finding Coupled Patterns in Climate Data. J. Clim. 1992, 541–560. [CrossRef] 76. Wallace, J.M.; Smith, C.; Bretherton, C.S. Singular Value Decomposition of Wintertime Sea Surface Temperature and 500-mb Height Anomalies. J. Clim. 1992, 561–576. [CrossRef] 77. Wu, Z.; Huang, N.E.; Long, S.R.; Peng, C.-K. On the trend, detrending, and variability of nonlinear and nonstationary time series. Proc. Natl. Acad. Sci. USA 2007, 104, 14889–14894. [CrossRef][PubMed] 78. Wu, Z.; Huang, N.E. Ensemble Empirical Mode Decomposition: A Noise-Assisted Data Analysis Method. Adv. Adapt. Data Anal. 2009, 1, 1–41. [CrossRef] 79. Zhu, S.; Remedio, A.R.C.; Sein, D.V.; Sielmann, F.; Ge, F.; Xu, J.; Peng, T.; Jacob, D.; Fraedrich, K.; Zhi, X. Added value of the regionally coupled model ROM in the East Asian summer monsoon modeling. Theor. Appl. Clim. 2020, 140, 375–387. [CrossRef] 80. Wu, J.; Zhou, L.; Mo, X.; Zhou, H.; Zhang, J.; Jia, R. Drought monitoring and analysis in China based on the Integrated Surface Drought Index (ISDI). Int. J. Appl. Earth Obs. Geoinf. 2015, 41, 23–33. [CrossRef] 81. Roy, S.S.; Roy, N.S. Influence of Pacific decadal oscillation and El Niño Southern oscillation on the summer monsoon precipitation in Myanmar. Int. J. Clim. 2010, 31, 14–21. [CrossRef] Water 2021, 13, 729 27 of 27

82. Kumar, K.K.; Rajagopalan, B.; Hoerling, M.; Bates, G.; Cane, M. Unraveling the Mystery of Indian Monsoon Failure during El Niño. Science 2006, 314, 115–119. [CrossRef][PubMed] 83. Kreft, S.; Eckstein, D. Global Climate Risk Index 2014. Who Suffers Most from Extreme Weather Events? Germanwatch: Bonn, Germany, 2016; ISBN 9783943704143. 84. Delworth, T.L.; Mann, M.E. Observed and simulated multidecadal variability in the Northern Hemisphere. Clim. Dyn. 2000, 16, 661–676. [CrossRef] 85. Kerr, R.A. A North Atlantic Climate Pacemaker for the Centuries. Science 2000, 288, 1984–1985. [CrossRef][PubMed] 86. Bowling, L.C.; Lettenmaier, D.P.; Nijssen, B.; Graham, L.P.; Clark, D.B.; El Maayar, M.; Essery, R.; Goers, S.; Gusev, Y.M.; Habets, F.; et al. Simulation of high-latitude hydrological processes in the Torne–Kalix basin: PILPS Phase 2(e). Glob. Planet. Chang. 2003, 38, 1–30. [CrossRef] 87. Chang, C.-P.; Zhang, Y.; Li, T. Interannual and Interdecadal Variations of the East Asian Summer Monsoon and Tropical Pacific SSTs. Part I: Roles of the Subtropical Ridge. J. Clim. 2000, 13, 4310–4325. [CrossRef] 88. Krishnan, R.; Sugi, M. Pacific decadal oscillation and variability of the Indian summer monsoon rainfall. Clim. Dyn. 2003, 21, 233–242. [CrossRef] 89. Zhu, S.; Ge, F.; Sielmann, F.; Pan, M.; Fraedrich, K.; Remedio, A.R.C.; Sein, D.V.; Jacob, D.; Wang, H.; Zhi, X. Seasonal temperature response over the Indochina Peninsula to a worst-case high-emission forcing: A study with the regionally coupled model ROM. Theor. Appl. Clim. 2020, 142, 613–622. [CrossRef] 90. Webster, P.J.; Yang, S. Monsoon and Enso: Selectively Interactive Systems. Q. J. R. Meteorol. Soc. 1992, 118, 877–926. [CrossRef] 91. Reckien, D.; Petkova, E.P. Who is responsible for climate change adaptation? Environ. Res. Lett. 2018, 14, 014010. [CrossRef] 92. Supari; Tangang, F.; Juneng, L.; Cruz, F.; Chung, J.X.; Ngai, S.T.; Salimun, E.; Mohd, M.S.F.; Santisirisomboon, J.; Singhruck, P.; et al. Multi-model projections of precipitation extremes in Southeast Asia based on CORDEX-Southeast Asia simulations. Environ. Res. 2020, 184, 109350. [CrossRef]