Drought diagnostics

The unusual twin failure of the south-west monsoon and the north-east monsoon following immediately in the southern States in 2016 is the chief reason for the unprecedented drought situation in the region.

A farmer showing dried paddy crop at Siruvapuri, 40 kilometres from Chennai. Tamil Nadu's 32 districts were declared drought affected following poor rainfall received during the north-east monsoon period of October to December 2016.

THE unusual failures of the south-west monsoon (SWM)—June to September—and the immediately following north-east monsoon (NEM)—October to December—in 2016, and the current heat wave conditions are the reasons for the unprecedented drought situation in the southern States. Figure 1 (a-c) shows the Standardised Precipitation Index (SPI), an index based on rainfall for drought monitoring, moderately/severely/extremely dry/wet situations for the SWM period, the NEM period and the whole of 2016, which is indicative of the grim situation.

Forecasts by the India Meteorological Department (IMD) for the 2016 SWM and NEM for the southern peninsular region had been that there would be normal rainfall, but in both cases they were off the mark. The prediction for the NEM, particularly, was hugely off the mark. The region comprises the following nine meteorological subdivisions (which cover five States and two Union Territories): coastal Andhra Pradesh (CAP), Telangana (TEL), Rayalaseema (RYS), Tamil Nadu plus Puducherry (TN), coastal Karnataka (CK), north interior Karnataka (NIK), south interior Karnataka (SIK), Kerala (KER) and Lakshadweep (LAK).

The NEM is important for agriculture and water resources in the meteorological subdivisions of TN, CAP, RYS, SIK and KER. It is particularly critical for TN, which is the biggest beneficiary of the NEM, as it receives nearly 48 per cent of its annual rainfall (of 914 millimetre) during October-December as against about 35 per cent during June-September (Table 1). The TN subdivision receives less rain during the SWM because, being on the leeward side of the Western Ghats, it falls in the rain-shadow region for the SWM. TN is the only subdivision that receives significantly more rainfall during the October-December period than during June-September (317 mm). Kerala, on the other hand, receives 70 per cent of its annual rainfall (2,928 mm) during June-September and only 16 per cent (478 mm) during October-December.

The NEM rainfall data for 1901-2010 show that there is a high degree of correlation between rainfall over the entire southern region and that over TN—a correlation coefficient of 0.84. For instance, during the 110-year period, while there have been 26 and 27 years of excess and deficient rainfall respectively over the southern region, the corresponding figures for TN were 26 and 25. Frequently, the NEM’s performance over TN is taken as an index for its performance over the entire southern peninsula. During the last two decades of the NEM up to 2010, there were more excess years than deficient years.

The performance of NEM 2016 was the worst in the past 140 years. It caused huge NEM rainfall deficits (>60 per cent) in all the five subdivisions of significance (Figure 2). TN, for instance, had a rainfall deficit of 62 per cent (against the long period average, or LPA, of 438.2 mm) and a deficit of this magnitude (63 per cent) was last experienced in 1876. Interestingly, this highly deficient NEM came after a year of huge excess rainfall (of +53 per cent of the LPA) in NEM 2015, which resulted in a deluge in and around Chennai accompanied by major devastation ( Frontline , December 25, 2015). For Chennai, this double blow was indeed a cruel irony.

Before we look at the reasons for the NEM’s failure, let us first examine the performance of the SWM and the reasons for its failure over the southern States. In absolute figures, the quantum of SWM rainfall for the country as a whole, though technically normal, was significantly lower than the IMD forecast. But, in terms of spatial distribution, the monsoon failed the southern States. Against the forecast of 106 per cent of the LPA, the actual rainfall for the country as a whole was only 97 per cent. But in terms of rainfall distribution over the southern peninsula, the actual rainfall was only 92 per cent of the LPA against a prediction of 113 per cent.

Table 2 shows the rainfall distribution in the subdivisions of the southern peninsula. Four subdivisions had deficient rainfall and TN was almost there with −19 per cent departure from the normal. (On a regional scale “normal” is when the rainfall departure is within +19 per cent and −19 per cent of the LPA.) Significantly, Kerala, which receives 70 per cent of its rainfall during the SWM, was in deficit by 34 per cent. District-wise, the SWM rainfall distribution (Figure 3) shows that, barring a few districts, the entire western coastal belt of the region (all districts in KER and many districts in CK) and most western districts of TN had deficient rainfall. Before June 2016, most of the climate forecasting models around the world were indicating a high probability for the development of La Nina (the cooling of waters in equatorial Pacific that is positively correlated to the SWM and the opposite of El Nino, which is negatively correlated to SWM) during the second half of the monsoon season, which would have favoured a normal to above normal monsoon.

Figure 4 shows the temperature anomalies over the Central Pacific (Nino 3.4), which is known to have considerable influence over the Indian monsoon. The oscillatory mode of the sea-surface temperature (SST), known as El Nino/Southern Oscillation (ENSO), has maintained a near-neutral-to-weak La Nina condition right into 2017. Moreover, according to the IMD, the atmospheric response to the neutral-to-mildly-La Nina conditions, as indicated by the corresponding Southern Oscillation Index (SOI), which is a measure of the observed sea-level pressure difference between Tahiti and Darwin, Australia, on the Pacific, lagged behind and could be observed only after the SWM season was over. (Negative SOI values correspond to El Nino and positive values to La Nina.) Thus the impact of ENSO on the SWM was neutral and the non-development of La Nina against expectations was one of the chief reasons for the SWM rainfall forecast going awry.

Monthly distribution of rainfall If one looks at the monthly distribution of rainfall over the southern peninsula, August 2016 was a particularly bad month (Table 3), the region receiving only 60 per cent of the LPA. According to the IMD, in August 2016 there was a strong ant-cyclonic circulation anomaly in the lower troposphere in the southern parts of the Arabian Sea, the southern peninsular region and south-western Bay of Bengal, which resulted in atmospheric subsidence over the southern peninsula, resulting in deficient monsoon in the entire region during the month. An important meteorological influence on the monsoon is that of the Indian Ocean Dipole (IOD), which is related to SSTs in the western and eastern Indian Ocean. A positive IOD, when the SST on the western side is warmer (and the eastern side is cooler), boosts the SWM winds. On occasions, positive IOD has helped counter an unfavourable ENSO condition. Unfortunately, in 2016, the IOD, too, conspired against the monsoon and remained negative throughout (Figure 5) with significantly above normal SSTs.

Another phenomenon that is known to have a role in the performance of the monsoon is the Madden Julian Oscillation (MJO). It is a tropical oscillatory disturbance that propagates eastward around the globe with a cycle of about 30-60 days. Although MJO does not cause ENSO, it is known to influence the impact of ENSO. It can speed up and intensify the development of El Nino and La Nina phases. Also, depending on the phase of the MJO oscillation, it can either enhance or suppress convection over the region where its activity is located. According to the IMD, after about August 10, the MJO was active over the Pacific causing above normal convective activity over there. As a consequence, the northward propagation of the Inter-Tropical Convergence Zone (ITCZ), or the “Equatorial Trough”, into the Indian landmass, which aids convection and rainfall, was absent. The End of the Monsoon Report 2016 of the IMD also points out that there was anomalous subsidence over the peninsular region causing below normal convective activity and below normal rainfall over most of the peninsular region. Increased convective activity and associated rising motion over the Indian Ocean (owing to the negative IOD) on the one side and the passage of the remnants of a number of low pressure systems from the Pacific along the monsoon trough over central India on the other resulted in this anomalous subsidence between these two regions, says the report.

It is instructive to point out the special features of the NEM in appreciating the difficulty in forecasting as well as understanding the possible reasons for its failure in 2016. The NEM system is essentially the retreating SWM when there is a reversal of the south-westerly surface and lower tropospheric winds associated with the SWM to become north-easterlies that begin to flow over the Indian region in October. The monsoon trough (in the 850 millibar level winds at a height of about 1.5 kilometres where the atmosphere essentially begins) is normally located near 20 N latitude in September, which gradually moves southwards to 16-17 N in October and 12-13 N in November, which is a clear indication of the reversal of the low-level wind regime.

Correspondingly, the monsoon withdraws from the northern latitudes down up to 15 N in October, and south of 15 N there is a marked increase in rainfall, indicating the onset of the NEM. The reversal of winds is well defined over the belt of coastal Tamil Nadu (CTN), which marks the onset of the NEM, and the normal onset date on the basis of the 110-year database is taken as October 20. Before the onset, the SWM continues over the five subdivisions into the first half of October as well. The normal date of withdrawal is taken to be December 30. In 2016, the SWM withdrew completely from the Indian region on October 28, 2016, and the NEM onset occurred on October 31, 2016.

Since the country receives three-fourths of its annual rainfall during the SWM period, the extent of research in forecasting the NEM has been far less compared with the SWM. Being a large-scale system, the SWM’s synoptic features are well defined and have been extensively studied and researched upon. On the other hand, the meteorological factors influencing the smaller-scale NEM (being confined to the southern peninsula) seem to be not as clearly discernible. But on a broad scale, it is known that just as the south-westerly winds of the SWM are driven by what is known as the Mascarene High (the high pressure region in the southern Indian Ocean near the Mascarene Islands), the reversal of the south-westerly winds to north-easterlies is driven by the cold outflow from the Siberian High (a strong surface high pressure region over Siberia) in the Northern Hemisphere. Recent studies have also demonstrated that the ENSO phenomenon, which we know has a strong influence over the summer monsoon, affects the NEM, too. Similarly, the IOD mode has been identified to have an influence on the NEM with the positive IOD phase being associated with a good NEM and a negative IOD phase with a poor or suppressed NEM.

The seasonal NEM rainfall exhibits high variability, but the year-to-year variability is such that there are seasons of highly deficient rainfall and seasons of large-scale floods. During a given season, the NEM often witnesses long dry spells. As is well known, occurrence of intense cyclonic storms over the Bay of Bengal is another characteristic feature of the NEM. Because of this somewhat chaotic behaviour of the NEM as against the well-organised SWM system, the skill in forecasting the NEM has been far less than in the case of the SWM. In particular, reasons for failure in the NEM’s performance have been especially difficult to identify, points out Y.E.A. Raj, former Deputy Director General of Meteorology at the Chennai Regional Centre of the IMD, who has done extensive research on the NEM.

The IMD’s forecast for NEM 2015 was fairly accurate. It had predicted above normal rainfall with higher than 111 per cent of the LPA of seasonal rainfall over the southern peninsula. The rainfall caused unprecedented floods in Chennai. For TN, it had forecast over 112 per cent of the LPA. The actual rainfall for the southern peninsula was 132 per cent of the LPA and the rainfall over TN was 152 per cent of its LPA. On the basis of model predictions that the weak La Nina condition would turn to cool neutral ENSO condition and that the weak IOD condition would continue to remain so during the NEM period, the IMD had in September 2016 forecast a normal NEM 2016 for the five met subdivisions of the southern peninsular region and normal rainfall for TN.

Interestingly, unanimously agreeing on the above perspective on the ENSO and the IOD, the South Asian Climate Outlook Forum (SASCOF), of which India is a member, during its ninth session in Myanmar in September 2016, made the following observation: “There is unanimity among the experts that the prevailing cool neutral La Nina conditions in the equatorial Pacific are likely to continue or reach to borderline La Nina conditions during the OND [October to December] season. However, it is recognised that there is some uncertainty on the potential impacts of weak La Nina on the climate of the region due to strong day-to-day atmospheric variability observed in the region.” On the basis of this and other considerations, it issued the following forecast for NEM 2016 in its consensus statement, with an accompanying map for the consensus outlook (Fig. 6): “The outlook suggests that during the 2016 NEM, normal rainfall is likely over most parts of South Asia. However, below normal activity is likely over some areas of the Southern part of the region consisting of south-east peninsular India, Sri Lanka and Maldives…. Below normal rainfall is also likely over some areas of north and eastern parts of the region.” Figure 6 shows a 40 per cent deficient rainfall over the coastal regions and slightly lower deficiency over the other regions. In hindsight, it would seem that this was a better forecast.

“Positive SST in Nino 3.4 (El Nino) and negative SOI during August-September and their persistence in October-November favour a good NEM,” Raj said in an email to Frontline . “But the correlation coefficients are of the order of 0.4-0.5 only and the relation is not one-to-one. There are several years when ENSO did not indicate excess/deficient NEM rainfall, but overall there is some relation,” he added. Raj also pointed out that for NEM 2016, the SST anomalies over Nino 3.4 varied between −0.41 and −0.73, indicating weak La Nina only. The corresponding SOI values for August to December months were 5.3, 13.5, −4.3, −0.7, and 2.6 respectively providing no proper signal, he said.

The onset for NEM 2016 was delayed because of the cyclonic storm Kyant that formed over the Bay of Bengal on October 21, 2016. Also, according to the IMD, the onset phase of NEM 2016 was somewhat unusual. Normally, the onset occurs over southern CTN and the rain belt then spreads to the interior and northern parts of Tamil Nadu and Kerala. In 2016, the onset occurred over SCAP and extreme northern parts of CTN and then spread southwards. But the onset phase was short-lived as the MJO continued to play spoilsport. Over the peninsular region, it continued to be in the unfavourable phase because of which the rainfall activity did not sustain.

A second low pressure system that developed into a depression occurred on November 2, lasting up to November 6. This depression first moved towards NCAP but curved back and moved towards Bangladesh, carrying with it precious moisture that was available along and off TN coast. This made way for dry and cold wind from the north to penetrate, which caused the weakening of the monsoon over the southern peninsular region. While the depression of November 2 brought heavy to very heavy rainfall at isolated places, the transient low-intensity weather systems that formed subsequently over the southern parts of the Bay of Bengal during the rest of November, which had the potential to develop into rainfall-bearing systems over land, could not be carried by the easterlies up to the south-eastern peninsula because of dry and cold winds from the north, and the easterlies became north-easterlies and carried these rain-bearing systems to Sri Lanka, skipping the southern peninsular region altogether.

Only when the cyclonic storm Nada (November 29-December 2, 2016) and the very severe cyclonic storm Vardah (December 6-13, 2016) over the south-east Bay of Bengal moved westwards and crossed over Tamil Nadu—the former near Nagapattinam and the latter close to Chennai—the associated landfalls and westward movement over land triggered enhanced rainfall activity over the southern peninsular region. Vardah, of course, brought with it some amount of devastation along the coast as well as more than one-metre-high storm surges and sustained surface winds of 100-110 kilometres per hour speed. But it also brought heavy to very heavy rainfall days in various parts of TN (where even extremely heavy rainfall days occurred), CAP and RYS.

Besides the above unfavourable meteorological conditions, there were other meteorological factors that have identified influence over the performance of the NEM, which too were unfavourable. As noted earlier, the MJO activity was unfavourable during the entire season. There were also strong westerly wind anomalies, which prevented proper north-easterly winds to blow over the southern peninsular region. The equatorial trough had migrated too far to southern latitudes closer to the equator, rather than being over the Indian southern region, which could otherwise have assisted enhanced precipitation. The locations of the Siberian High (compared with 2015) were unfavourable for strong movement of north-easterlies that would have resulted in a good monsoon. Also, except for the solitary instance that got deflected towards Sri Lanka, there was general lack of easterly wave activity in the latitude belt (8-12 N) of relevance to the NEM.

“Despite two low pressure systems crossing the TN coast and easterly wave troughs traversing over the the Bay of Bengal, the NEM rainfall of 2016 ended up deficient over all the five subdivisions benefited by the NEM,” notes the IMD’s end of season report on NEM 2016 (Figure 2). In sum, month-wise, October and November rainfall over all the subdivisions was largely deficient. TN became largely deficient in October because of the late onset and so were the other four subdivisions. In November, as pointed out earlier, the depression that had formed over the south-east Bay of Bengal but moved towards the Bangladesh coast, the penetration of dry and cold air from the north into the south-west Bay of Bengal and the confinement of convective activity because of troughs in the easterlies to the south of the Indian latitudes resulted in poor NEM activity over the southern peninsular region. Finally, in December, rainfall activity picked up over TN, CAP, RYS and SIK, thanks to the passage of Nada and Vardah. KER, being too far to the west, did not benefit from even this minor spurt in the otherwise largely suppressed overall NEM activity (Table 4).

At the end of the season, the SPI, which is a measure of rainfall-based wetness/dryness of a region, indicated moderately dry to extremely dry over most of the NEM region during October-December 2016 (Figure 1b). The prevailing heat wave over the southern States have only accentuated this condition. The satisfactory reservoir position, thanks to the excess rainfall during NEM 2015, unfortunately got undone because of the poor SWM 2016 over the southern region. Otherwise, the extreme drought situation prevailing now could have been reasonably managed.