Homogeneous Indian Monsoon rainfall: Variability and prediction

Abstract

The Indian summer monsoon rainfall is known to have considerable spatial variability, which imposes some limitations on the all-India mean widely used at present. To prepare a spatially coherent monsoon rainfall series for the largest possible area, fourteen subdivisions covering the northwestern and central parts of India (about 55~o of the total area of the country), having similar rainfall characteristics and associations with regional/global circulation parameters are merged and their area-weighted means computed, to form monthly and seasonal Homogeneous Indian Monsoon (HIM) rainfall series for the period 1871-1990. This paper includes a listing of monthly and seasonal rainfall of HIM region. HIM rainfall series has been statistically analysed to understand its characteristics, variability and teleconnections for long-range prediction.HIM rainfall series is found to be homogeneous, Gaussian distributed and free from persistence. The mean (R) rainfall is 757mm (87~ of annual) and standard deviation (S) 119 mm, with a Coefficient of Variation (CV) of 16~o. There were 21 dry (R i ~</~-S) and 19 wet (R i/>/~ + S) years during 1871-1990. There were clusters of frequent negative departures during 1899-1920 and 1965-1987 and positive departures during 1942-1961. The recent three decades show very high rainfall variability with 10 dry and 6 wet years. The decadal averages were alternatively positive and negative for three consecutive decades, viz., 1871-1900 (positive); 1901-1930 (negative); 1931-1960 (positive) and 1961-1990 (negative) respectively. Significant QBO and autocorrelation at 14th lag have been found in HIM rainfall series.To delineate the changes in the climatic regime of the Indian summer monsoon, sliding correlation coefficients (CCs) between HIM rainfall series and (i) Bombay msl pressure, (ii) Darwin mst pressure and (iii) Northern Hemisphere surface air temperature over the period 1871-1990 have been examined. The M-year sliding CCs showed the systematic turning points of positive and negative CCs around the years, 1900 and 1940. In the light of other corroborative evidences, these turning points seem to delineate 'meridional' monsoon regime during 1871-1900 and 1940-1990 and 'zonal' monsoon regime during 1901-1940. The monsoon signal is particularly dominant in many regional and global circulation parameters, during 1951-1990. Using the teleconnections of HIM series with 12 regional/global circulation parameters during the recent 36-year period 1951-86 regression models have been developed for long-range prediction. In the regression equations 3 to 4 parameters were entered, explaining upto 80~ of the variance, depending upon the data period. The parameters that prominently enter the multiple regression equations are (i) Bombay msl pressure, (ii) April 500 mb Ridge at 75~ (iii) NH temperature, (iv) Nouvelle minus Agalega msl pressure and (v) South American msl pressure. Eleven circulation parameters for the period 1951-80 were subjected to Principal Component Analysis (PCA) and the PC's were u...

Sub‐Annual to Interannual Arabian Sea Upwelling, Sea Surface Temperature, and Indian Monsoon Rainfall Reconstructed Using Congruent Micrometer‐Scale Climate Proxies

“…(a) Annual‐scale comparison of Indus River discharge from Kotri Barrage (downstream; blue line), Indus River sediment discharge from Kotri Barrage (downstream, black line) (Inam et al., 2007), and the annual mean PC1 score (green). (b) Annual‐scale comparison of homogenous India June‐September (JJAS) total rainfall percent‐departures (Parthasarathy et al., 1993) from a temporally‐centered 30‐year sliding mean (black); Karachi, Pakistan total annual rainfall percent departures from the 1948–1993 mean (orange);and the linearly detrended mean annual PC1 score (green). (c) As in B, but comparing the 3‐year moving means of the annual values for each series.…”

“…The core contains a distinct laminated interval from approximately 0-13.5 cm core depth that was deposited in the twentieth century (Schulz & von Rad, 2014; Figure 2; hereafter we omit the use of 'core depth'). The homogenous India summer monsoon seasonal rainfall (June through September) area of Parthasarathy et al (1993) is shaded gray. The blue circle is Extended Reconstructed Sea Surface Temperature (ERSST), version 5 grid point 24°N, 66°E.…”

“…After linear detrending, annual mean PC1 captures many of the homogeneous India June-September and Karachi annual rainfall departures prior to ∼1975 (Figure 5b). On a multi-annual scale, the 3-year moving mean of detrended annual PC1 is positively correlated with the 3-year moving mean of homogeneous India monsoonal (1900( -1993black line), normalized SST reconstructed using the reciprocal CCaT proxy (red), normalized SST reconstructed using the 𝐴𝐴 U K ′ 37 proxy (blue), and detrended PC1 (green) on the varve age-depth model. CCaT has been offset by −8 power units and PC1 has been offset by −10 power units for clarity.…”

“…Figure 4. (a) Thomson multi-taper power spectrum of normalized Extended Reconstructed Sea Surface Temperature (ERSST) at 24°N 66°E(1900( -1993 black line), normalized SST reconstructed using the reciprocal CCaT proxy (red), normalized SST reconstructed using the 𝐴𝐴 U K ′ 37 proxy (blue), and detrended PC1 (green) on the varve age-depth model. CCaT has been offset by −8 power units and PC1 has been offset by −10 power units for clarity.…”

On the All‐India Rainfall Index and Sub‐India Rainfall Heterogeneity

“…In this study, we address three long-running controversies regarding AIRI's interpretation in light of this spatiotemporal variability, attempting to synthesize and reconcile past arguments made with differing and often short and/or coarse datasets by using state-of-the-art, 120-year datasets of Indian rainfall at high resolution and of sea surface temperature (SST).First is the relationship between AIRI and the overall spatiotemporal extent of wetting or drying across India. Spatially, in most summers there are parts of India that experience drought and others excess rainfall, and for rain-fed agriculture for example, quite plausibly the spatial extent of drought relative to local rainfall normals is more relevant than the average rainfall anomaly in mm day −1 across India (Parthasarathy et al, 1993). This has motivated attempts to define bulk indices in terms of the fraction of the Indian surface area experiencing rainfall anomalies exceeding specified thresholds (Mooley et al, 1981;Parthasarathy et al, 1987).…”