Multifractal analysis of daily rainfall from a monsoon watershed in Eastern China

The multifractal behavior of daily rainfall was investigated for a watershed in Eastern China to better understand the temporal structure of rainfall under monsoonal climate. In this study, over periods of up to 46 years, daily rainfall recorded in 1962 to 2007 at 10 meteorological stations in the administrative area of Lin-Yi City in Shandong province were analyzed with focus on features of power spectra, standard statistical moments, and exceedance probability tails of these daily rainfall time series. Spectral analysis and study of the moments of the rainfall intensity showed that a scaling range from 1 day to 1 year is present. Empirical moment scaling functions of the rainfall intensity calculated for different moments of order suggested that the values of universal multifractal parameters α and C ₁ for all stations were approximated to 0.7 and 0.37, respectively. Comparing with the parameters estimated in other literatures, our results showed higher values for α but lower values for C ₁ in general, which suggested that the rainfall series in the study watershed influenced by the East-Asia monsoon climate have similarities to that in France, but are spikier and smoother than that in the semi-arid region in Portugal. The parameter H values were estimated as vary from ?0.18 to ?0.22, which is similar to the result obtained by Tessier et al. (J Geophys Res-Atmos 101:26427–26440, 1996).     

Multi-decadal scale variability in autumn-winter rainfall in south-western Australia since 1655 AD as reconstructed from tree rings of Callitris columellaris

We present the first tree-ring based reconstruction of rainfall for the Lake Tay region of southern Western Australia. We examined the response of Callitris columellaris to rainfall, the southern oscillation index (SOI), the southern annular mode (SAM) and surface sea temperature (SST) anomalies in the southern Indian Ocean. The 350-year chronology was most strongly correlated with rainfall averaged over the autumn-winter period (March–September; r = ?0.70, P < 0.05) and SOI values averaged over June–August (r = 0.25, P < 0.05). The chronology was not correlated with SAM or SSTs. We reconstructed autumn-winter rainfall back to 1655, where current and previous year tree-ring indices explained 54% of variation in rainfall over the 1902–2005 calibration period. Some variability in rainfall was lost during the reconstruction: variability of actual rainfall (expressed as normalized values) over the calibration period was 0.78, while variability of the normalized reconstructed values over the same period was 0.44. Nevertheless, the reconstruction, combined with spectral analysis, revealed that rainfall naturally varies from relatively dry periods lasting to 20–30 years to 15-year long periods of above average rainfall. This variability in rainfall may reflect low-frequency variation in the El Niño-Southern Oscillation rather than the effect of SAM or SSTs.     

Nonlinear dynamics of meteorological variables: multifractality and chaotic invariants in daily records from Pastaza, Ecuador

Weather represents the daily state of the atmosphere. It is usually considered as a chaotic nonlinear dynamical system. The objectives of the present study were (1) to investigate multifractal meteorological trends and rhythms at the Amazonian area of Ecuador and (2) to estimate some nonlinear invariants for describing the meteorological dynamics. Six meteorological variables were considered in the study. Datasets were collected on a daily basis from January 1st 2001 to January 1st 2005 (1,460 observations). Based on a new multifractal method, we found interesting fractal rhythms and trends of antipersistence patterns (Fractal Dimension >1.5). Nonlinear time series analyses rendered Lyapunov exponent spectra containing more than one positive Lyapunov exponent in some cases. This sort of hyperchaotic structures could explain, to some extent, larger fractal dimension values as the Kaplan–Yorke dimension was also in most cases larger than two. The maximum prediction time ranged from ξ?=?1.69 days (approximately 41 h) for E/P ratio to ξ?=?14.71 days for evaporation. Nonlinear dynamics analyses could be combined with multifractal studies for describing the time evolution of meteorological variables.     

Validating a rapid-update satellite precipitation analysis across telescoping space and time scales

In order to properly utilize remotely sensed precipitation estimates in hydrometeorological applications, knowledge of the accuracy of the estimates are needed. However, relatively few ground validation networks operate with the necessary spatial density and time-resolution required for validation of high-resolution precipitation products (HRPP) generated at fine space and time scales (e.g., hourly accumulations produced on a 0.25° spatial scale). In this article, we examine over-land validation statistics for an operationally designed, meteorological satellite-based global rainfall analysis that blends intermittent passive microwave-derived rainfall estimates aboard a variety of low Earth-orbiting satellite platforms with sub-hourly time sampling capabilities of visible and infrared imagers aboard operational geostationary platforms. The validation dataset is comprised of raingauge data collected from the dense, nearly homogeneous, 1-min reporting Automated Weather Station (network of the Korean Meteorological Administration during the June to August 2000 summer monsoon season. The space-time RMS error, mean bias, and correlation matrices were computed using various time windows for the gauge averaging, centered about the satellite observation time. For ±10 min time window, a correlation of 0.6 was achieved at 0.1° spatial scale by averaging more than 3 days; coarsening the spatial scale to 1.8° produced the same correlation by averaging over 1 h. Finer than approximately 24-h and 1° time and space scales, respectively, a rapid decay of the error statistics was obtained by trading-off either spatial or time resolution. Beyond a daily time scale, the blended estimates were nearly unbiased and with an RMS error of no worse than 1 mm day^?1.     

Impact of different regression frameworks on the estimation of the scaling properties of radar rainfall

Rainfall is characterized by high variability both in space and time. Despite continuous technological progress, the available instruments that are used to measure rainfall across several spatio-temporal scales remain inaccurate. To remedy this situation, scaling relationships of spatial rainfall offer the potential to link the observed or predicted precipitation quantities at one scale to those of interest at other scales. This paper focuses on the estimation of the spatial rainfall scaling functions. Standard scaling analysis constructed by means of the ordinary least squares method often violates such basic assumptions implicit in its use and interpretation as homoschedasticity, independence, and normality of the errors. Consequently, the authors consider alternative regression frameworks i.e. bootstrapping regression, semi parametric linear model, and multilevel normal linear model to show how these different approaches exert a significant impact on the multifractal analysis of radar rainfall. In addition, the uncertainties associated with the construction of the scaling function due solely to the regression procedure are quantified. The radar data come from the polarimetric C-band weather radar located in Rome, Italy, and the scaling properties are computed for a square domain centred on the radar site with a side length of 128 km and a finest resolution of 1 km².     

Integral Turbulence Statistics Over a Central European City Centre

Atmospheric measurements over 5 years (2005–2010) at two sites in ?ód?, central Poland have been analyzed to develop a better understanding of turbulence in urban areas. Fast response wind velocity, temperature, humidity and CO₂ concentration were measured using sonic anemometers and gas analyzers, placed on narrow masts at 37 and 42 m above the ground. The measurements were used to calculate standard deviations of each parameter, and were then normalized according to local Monin–Obukhov similarity theory and plotted as a function of stability parameter ζ = z′/L. Results for the wind components show typical scaling with a power law with exponent ±1/3 in the free convection limit, and that approaches a constant value close to neutral stratification. For stable conditions, the constant value in the neutral limit remains the same for stability parameters lower than 0.1–0.2, then increases. The normalized standard deviation of temperature fits the ?1/3 law in the free convection limit, approaching a constant value within a stable limit. However, it exhibits hyperbolic characteristics for close to neutral stratification. The normalized standard deviations for humidity and CO₂ concentration exhibit scaling similar to the wind components in the unstable regime and remain constant in the stable domain. The results for the wind components and for temperature are in the range of various functions found in other studies. The absolute values for humidity and CO₂ concentration seem to be slightly higher, but only single examples of such investigations can be found in the literature.     

Influence of a Two-scale Surface Roughness on a Neutral Turbulent Boundary Layer

Flow in the urban boundary layer is strongly influenced by the surface roughness, which is composed principally of isolated buildings or groups of buildings. Previous research has shown that the flow regime depends on the characteristic height of these obstacles (H), and the spacing between them (W). In reality, the urban boundary layer contains roughness elements with a wide range of length scales; in many practical situations these can be classified into large-scale roughness—buildings, or groups of buildings—and small-scale roughness, such as street furniture and elements on the façades and roofs. It is important to understand how the small-scale roughness might modify mass and momentum transfer in the urban boundary layer, but relatively little information is available concerning the potential interaction between large- and small-scale roughness elements in the different flow regimes. This problem has been studied using wind-tunnel experiments, by measuring vertical velocity profiles over a two-dimensional obstacle array, adding small-scale roughness elements to the top of larger parallel square bars. The experiments were performed for different cavity aspect ratios: the results show that the small-scale roughness increases the turbulence intensities and the momentum transfer when the large-scale obstacles are closely packed (H/W > 1) but it has very little effect for more widely-spaced obstacles (H/W < 1).     

Ground-based GPS measurements: time behavior from half-hour to years

Ground-based GPS and weather stations time series for the period 2010–2012 of precipitable water vapor (PWV), relative humidity (RH), and surface temperature (T) of half-hourly resolution are analyzed to demonstrate their value for dynamical analyses and weather forecasting. Three sample stations in the USA from the SoumiNet network are considered, which have rather continuous data for the last 3 years and a few missing values. Results for the three stations reveal the following features: (1) PWV time behavior is dominated by the annual cycle superimposed on high-frequency fluctuations with missing daily cycle, indicating a prevailing large-scale transport source of precipitable water at these sites; (2) RH is characterized by the daily cycle and high-frequency variability, while the annual cycle is missing; (3) T mainly varies following the annual and diurnal cycles; and (4) all variables show similar scaling properties of their variance spectra, S(f)?～?f ^–β , with a high-frequency regime of red noise type scaling (β?～?2) up to a day and long-term persistence beyond a week (β?～?0.5), with a week-long frequency interval of transition. Detrended fluctuation analysis of relative humidity indicates a clear long-term persistence scaling covering more than three decades. Implications of these findings on weather forecasting and climate modeling are discussed.     

RELATIONSHIP BETWEEN SUMMER LOW-FREQUENCY RAINFALL OVER SOUTHERN CHINA AND PROPAGATION OF TROPICAL INTRASEASONAL OSCILLATION

This study investigates the relationship between summer low-frequency rainfall over southern China and tropical intraseasonal oscillation (ISO) in the atmosphere by examining systematically the propagation features of the tropical ISO in terms of focusing on five large-scale low-frequency rainfall regimes in summer over southern China. It is demonstrated that there is a close linkage between the five rainfall regimes over southern China and the northward propagation of the tropical ISO. The moist ISO signals, which influence the low-frequency rainfall events in different regions of southern China, mainly propagate northwestward from the tropical ocean to the southeast of China. The southeast China rainfall regime is intimately associated with the moist ISO signals propagating northwestward from the equatorial mid-western Pacific Ocean. For both the Yangtze River regime and South of Yangtze River regime, the moist ISO signals over the northern South China Sea show an evident northward propagation towards the Yangtze River region, and then propagate westward. It is further found that the interaction between the northward propagation of low-latitude ISO signals and the southward propagation of high-latitude ISO signals can also make a clear influence on the low-frequency rainfall in southern China. For the Southern China regime, the moist ISO signals show a significant northward propagation from the Philippines. Moreover, for the rainless regime, southern China is under dry ISO signals’ control, and the latter shows no clear propagation to southern China. This study may provide insights for the extended-range forecasting of summer rainfall in southern China.     

Estimation of atmospheric surface layer parameters and numerical simulation using MM5 at Mangalore, West Coast of India

Atmospheric surface layer meteorological observations obtained from 20-m-high meteorological tower at Mangalore, situated along the west coast of India are used to estimate the surface layer scaling parameters of roughness length (z _o) and drag coefficient (C _D), surface layer fluxes of sensible heat and momentum. These parameters are computed using the simple flux–profile relationships under the framework of Monin–Obukhov (M–O) similarity theory. The estimated values of z _o are higher (1.35–1.54 m) than the values reported in the literature (>0.4–0.9 m) probably due to the undulating topography surrounding the location. The magnitude of C _D is high for low wind speed (<1.5 m s^?1) and found to be in the range 0.005–0.03. The variations of sensible heat fluxes (SHF) and momentum fluxes are also discussed. Relatively high fluxes of heat and momentum are observed during typical days on 26–27 February 2004 and 10–11 April 2004 due to the daytime unstable atmospheric conditions. Stable or near neutral conditions prevail after 1700 h IST with negative SHF. A mesoscale model PSU/NCAR MM5 is run using a high-resolution (1 km) grid over the study region to examine the influence of complex topography on the surface layer parameters and the simulated fluxes are compared with estimated values. Spatial variations of the frictional velocity (u _*), C _D, surface fluxes, planetary boundary layer (PBL) height and surface winds are noticed according to the topographic variations in the simulation.     

Multifractal detrended fluctuation analysis of the δ 18 O record of NGRIP ice core

The multifractal properties and scaling behaviors of the long-term and recent 2000-year δ ¹⁸ O records of NGRIP ice core are investigated by the multifractal detrended fluctuation analysis method. The generalized Hurst exponents, multifractal scaling exponents, and singularity spectrums of two δ ¹⁸ O records are derived to verify the multifractiality of two records. And the multifractal behaviors of two records are obviously different, which may reflect the climate change of the recent 2000-year time is quite different from one of the long-term time. In addition, the probability distribution analysis of two δ ¹⁸ O records is presented to manifest the different multifractality between two δ ¹⁸ O records of NGRIP ice core. Our results will be helpful to research the climate change.     

The Critical Richardson Number and Limits of Applicability of Local Similarity Theory in the Stable Boundary Layer

Measurements of atmospheric turbulence made over the Arctic pack ice during the Surface Heat Budget of the Arctic Ocean experiment (SHEBA) are used to determine the limits of applicability of Monin–Obukhov similarity theory (in the local scaling formulation) in the stable atmospheric boundary layer. Based on the spectral analysis of wind velocity and air temperature fluctuations, it is shown that, when both the gradient Richardson number, Ri, and the flux Richardson number, Rf, exceed a ‘critical value’ of about 0.20–0.25, the inertial subrange associated with the Richardson–Kolmogorov cascade dies out and vertical turbulent fluxes become small. Some small-scale turbulence survives even in this supercritical regime, but this is non-Kolmogorov turbulence, and it decays rapidly with further increasing stability. Similarity theory is based on the turbulent fluxes in the high-frequency part of the spectra that are associated with energy-containing/flux-carrying eddies. Spectral densities in this high-frequency band diminish as the Richardson–Kolmogorov energy cascade weakens; therefore, the applicability of local Monin–Obukhov similarity theory in stable conditions is limited by the inequalities Ri < Ri _cr and Rf < Rf _cr. However, it is found that Rf _cr  =  0.20–0.25 is a primary threshold for applicability. Applying this prerequisite shows that the data follow classical Monin–Obukhov local z-less predictions after the irrelevant cases (turbulence without the Richardson–Kolmogorov cascade) have been filtered out.     

Regional intensity–duration–frequency analysis in the Eastern Black Sea Basin,Turkey, by using L-moments and regression analysis

The analysis of rainfall frequency is an important step in hydrology and water resources engineering. However, a lack of measuring stations, short duration of statistical periods, and unreliable outliers are among the most important problems when designing hydrology projects. In this study, regional rainfall analysis based on L-moments was used to overcome these problems in the Eastern Black Sea Basin (EBSB) of Turkey. The L-moments technique was applied at all stages of the regional analysis, including determining homogeneous regions, in addition to fitting and estimating parameters from appropriate distribution functions in each homogeneous region. We studied annual maximum rainfall height values of various durations (5 min to 24 h) from seven rain gauge stations located in the EBSB in Turkey, which have gauging periods of 39 to 70 years. Homogeneity of the region was evaluated by using L-moments. The goodness-of-fit criterion for each distribution was defined as the Z^DIST statistics, depending on various distributions, including generalized logistic (GLO), generalized extreme value (GEV), generalized normal (GNO), Pearson type 3 (PE3), and generalized Pareto (GPA). GLO and GEV determined the best distributions for short (5 to 30 min) and long (1 to 24 h) period data, respectively. Based on the distribution functions, the governing equations were extracted for calculation of intensities of 2, 5, 25, 50, 100, 250, and 500 years return periods (T). Subsequently, the T values for different rainfall intensities were estimated using data quantifying maximum amount of rainfall at different times. Using these T values, duration, altitude, latitude, and longitude values were used as independent variables in a regression model of the data. The determination coefficient (R ²) value indicated that the model yields suitable results for the regional relationship of intensity–duration–frequency (IDF), which is necessary for the design of hydraulic structures in small and medium sized catchments.     

Turbulence characteristics in the atmospheric surface layer during summer monsoon of 1997 over a semi-arid location in India

A land surface processes experiment (LASPEX) was conducted in the semi-arid region of Northwest India during January 1997–February 1998. Analysis of turbulent components of wind and air temperature collected in the surface layer (SL) at Anand (22°35′N, 72°55′E) during the Indian summer monsoon season from June to September 1997 is presented. Turbulent fluctuation of wind components and air temperature observed at Anand varied as a function of terrain features and stability of the surface layer. Under neutral conditions, the standard deviation of vertical velocity (σ _w ) and temperature (σ _T ) were normalized using respective surface layer scaling parameter u _* and T _* which fitted the expressions σ _w /u _* = 1.25 and σ _T /T _* ≈ 4. Micrometeorological spectrum of wind and temperature at 5 m above ground level (AGL) at Anand showed peaks at time scale of 1–3 min at the low-frequency end. The inertial sub-range characteristics (?2/3 slope) of the spectrum are exhibited mostly. However, in some occasions, slope of ?1 denoting brown noise was depicted by the wind and temperature spectrum, which indicated anisotropy in turbulence.     

Estimation of sub-hourly DDF curves using scaling properties of hourly and sub-hourly data at partially gauged site

In urban drainage systems, knowledge of short-duration rainfall events can be considered as one of the most critical elements when their hydrological behaviour wants to be investigated. The temporal resolution of rainfall data usually available for practical applications is often lower than the data requested for the design procedures or mathematical models application, greatly affecting their reliability. Moreover, when high resolution rain gauges are available in the catchment, the registration period cannot be sufficiently long for obtaining practically usable statistical analyses. The present study proposes a method for estimating the distribution of sub-hourly extreme rainfalls at sites where data for time interval of interest do not exist, but rainfall data for longer duration are available. The proposed method is based on the “scale-invariance” (or “scaling”) theory whose concepts imply that statistical properties of the extreme rainfall processes for different temporal scales are self-related by a scale-changing operator involving only the scale ratio. The methodology is applied to extreme rainfall data from a rain gauge network within the metropolitan area of Palermo (Italy). Following the application, it is shown that the statistical properties of the rainfall series have a simple scaling property over the range of duration 10 min–24 h. A simple parsimonious analytical formulation for the DDF curves, which embodies the scaling behaviour, is presented.     

Universal inverse power-law distribution for temperature and rainfall in the UK region

Meteorological parameters, such as temperature, rainfall, pressure, etc., exhibit selfsimilar space–time fractal fluctuations generic to dynamical systems in nature such as fluid flows, spread of forest fires, earthquakes, etc. The power spectra of fractal fluctuations display inverse power-law form signifying long-range correlations. A general systems theory model predicts universal inverse power-law form incorporating the golden mean for the fractal fluctuations. The model predicted distribution was compared with observed distribution of fractal fluctuations of all size scales (small, large and extreme values) in the historic month-wise temperature (maximum and minimum) and total rainfall for the four stations Oxford, Armagh, Durham and Stornoway in the UK region, for data periods ranging from 92 years to 160 years. For each parameter, the two cumulative probability distributions, namely c_max and c_min starting from respectively maximum and minimum data value were used. The results of the study show that (i) temperature distributions (maximum and minimum) follow model predicted distribution except for Stornowy, minimum temperature c_min. (ii) Rainfall distribution for c_min follow model predicted distribution for all the four stations. (iii) Rainfall distribution for c_max follows model predicted distribution for the two stations Armagh and Stornoway. The present study suggests that fractal fluctuations result from the superimposition of eddy continuum fluctuations.     

Retrieving air humidity,global solar radiation,and reference evapotranspiration from daily temperatures: development and validation of new methods for Mexico. Part I: humidity

The occurrence of flood and drought frequency is highly correlated with the temporal fluctuations of streamflow series; understanding of these fluctuations is essential for the improved modeling and statistical prediction of extreme changes in river basins. In this study, the complexity of daily streamflow fluctuations was investigated by using multifractal detrended fluctuation analysis (MF-DFA) in a large heterogeneous lake basin, the Poyang Lake basin in China, and the potential impacts of human activities were also explored. Major results indicate that the multifractality of streamflow fluctuations shows significant regional characteristics. In the study catchment, all the daily streamflow series present a strong long-range correlation with Hurst exponents bigger than 0.8. The q-order Hurst exponent h(q) of all the hydrostations can be characterized well by only two parameters: a (0.354 ≤ a ≤ 0.384) and b (0.627 ≤ b ≤ 0.677), with no pronounced differences. Singularity spectrum analysis pointed out that small fluctuations play a dominant role in all daily streamflow series. Our research also revealed that both the correlation properties and the broad probability density function (PDF) of hydrological series can be responsible for the multifractality of streamflow series that depends on watershed areas. In addition, we emphasized the relationship between watershed area and the estimated multifractal parameters, such as the Hurst exponent and fitted parameters a and b from the q-order Hurst exponent h(q). However, the relationship between the width of the singularity spectrum (Δα) and watershed area is not clear. Further investigation revealed that increasing forest coverage and reservoir storage can effectively enhance the persistence of daily streamflow, decrease the hydrological complexity of large fluctuations, and increase the small fluctuations.     

相空间中划分大尺度异常雨型的初步研究

文中针对以往雨型研究中的不足 ,从动力学角度将相空间引入到大尺度异常雨型的划分 ,通过估计EOF相平面上的概率密度函数、并确定局部极大值区域来划分主要雨型。分析表明 ,新雨型不仅再现了传统的 3类雨型特征 ,还得到了其他分型 ,并且能够反映出异常降水分布的季节内低频变化 ;分别采用 10和 2 0d低通滤波数据划分的雨型形态比较接近 ,但其种类、持续性和季节内分布等特征存在一定差异 ;雨型的气候分布与季风雨带有一定对应关系 ,其形成与夏季风异常关系密切。通过对简单直观的二维相平面问题的探讨 ,不仅证实了在相空间中划分大尺度雨型的可行性 ,也为进一步在高维空间中讨论雨型问题提供了依据。     

Scaling fluctuation analysis and statistical hypothesis testing of anthropogenic warming

Although current global warming may have a large anthropogenic component, its quantification relies primarily on complex General Circulation Models (GCM’s) assumptions and codes; it is desirable to complement this with empirically based methodologies. Previous attempts to use the recent climate record have concentrated on “fingerprinting” or otherwise comparing the record with GCM outputs. By using CO₂ radiative forcings as a linear surrogate for all anthropogenic effects we estimate the total anthropogenic warming and (effective) climate sensitivity finding: ΔT _anth  = 0.87 ± 0.11 K, $\uplambda_{{2{\text{x}}{\text{CO}}_{2} ,{\text{eff}}}} = 3.08 \pm 0.58\,{\text{K}}$ . These are close the IPPC AR5 values ΔT _anth  = 0.85 ± 0.20 K and $\uplambda_{{2{\text{x}}{\text{CO}}_{2} }} = 1.5\!-\!4.5\,{\text{K}}$ (equilibrium) climate sensitivity and are independent of GCM models, radiative transfer calculations and emission histories. We statistically formulate the hypothesis of warming through natural variability by using centennial scale probabilities of natural fluctuations estimated using scaling, fluctuation analysis on multiproxy data. We take into account two nonclassical statistical features—long range statistical dependencies and “fat tailed” probability distributions (both of which greatly amplify the probability of extremes). Even in the most unfavourable cases, we may reject the natural variability hypothesis at confidence levels >99 %.     

Mid-Holocene to present-day evolution of the Indian monsoon in transient global simulations

The low-frequency evolution of Indian rainfall mean-state and associated interannual-to-decadal variability is discussed for the last 6000 years from a multi-configuration ensemble of fully coupled global transient simulations. This period is marked by a shift of Indian Summer Monsoon Rainfall (ISMR) distribution towards drier conditions, including extremes, and a contraction of the rainy season. The drying is larger in simulations with higher horizontal resolution of the atmosphere and revised land surface hydrology. Vegetation–climate interactions and the way runoff is routed to ocean modulate the timing of the monsoon onset but have negligible effects on the evolution of seasonal rainfall amounts in our modeling framework in which carbon cycling is always active. This drying trend is accompanied by changes in ISMR interannual-to-decadal variability decreasing over north and south India but increasing over central India (20°–25° N). The ISMR interannual-to-decadal variability is decomposed into six physically consistent regimes using a clustering technique to further characterize its changes and associated teleconnections. From 6 to 3.8 kyr bp, the century-to-century modulations in the frequency of occurrence associated to the regimes are asynchronous between the simulations. Orbitally-driven trends can only be detected for two regimes over the whole 6–0 kyr bp period. These two regimes reflect increased influence of ENSO on both ISMR and Indian Ocean Dipole as the inter-hemispheric energy gradient weakens. Severe long-term droughts are also shown to be a combination of long-term drying and internally generated low-frequency modulations of the interannual-to-decadal variability.