Trends of winter precipitation extremes over Northwest Himalaya

ABSTRACT

The intensity and frequency of extreme precipitation events are of great concern for water resources over the Indian Himalayas, where climate change is rapidly progressive. The long-term trends of extreme precipitation events have been examined during the winter season (November to April; 1996–2016) using the data of 10 observatories maintained by the Snow and Avalanche Study Establishment (SASE) over the Northwest Himalaya (NWH) region. The extreme events are categorized into seven indices, which are 90^th and 95^th percentiles, rainy days, and maximum 1-d, 3-d, 5-d and total accumulated precipitation. The analysis shows increasing trends for all seven indices for the peak month of February for all Himalayan ranges and altitudes. The rainy days are significantly increased at a 95% confidence level in the Pir-Panjal Range. The teleconnection study reveals that the negative correlation between the North Atlantic Oscillation and winter precipitation extremes is critical for the inter-monthly precipitation behaviours over the NWH.

KEYWORDS:

• precipitation
• Northwest Himalaya
• climate extremes
• teleconnections
• western disturbances

Editor S. Archfield Associate Editor K. Ryberg

1 Introduction

The Intergovernmental Panel on Climate Change (IPCC) report Changes in Climate Extremes and their Impact on the Natural Physical Environment predicts that in the 21st century, extreme precipitation events will increase over many areas of the globe, mainly at high latitudes, in tropical regions, and in the northern mid-latitudes (Seneviratne et al. 2012). The Himalayan region is associated with sparse representation of meteorological data (Sharma et al. 2016), and better spatial and temporal coverage are needed to understand the effects of climate change in the region. Additionally, the presence of glaciers in this region is a matter of deep concern because glaciers have a significant impact on regional water and food security (Kohler and Maselli 2009, Romshoo et al. 2015, 2017). Due to the complex nature of the mountainous regions of the Indian Himalayas, very limited studies have been carried out to analyse its weather systems, like western disturbances (WDs) and monsoon circulations, in the past (Pisharoty and Desai 1956, Rao and Srinivasan 1969, Dhar et al. 1984, Dimri 2006, Puranik and Karekar 2009). WDs, while passing over the Western Himalayas (WH), form a short-lived mesoscale cloud structure underneath the cirrus canopy, which causes significant precipitation over the same region (Mohanty et al. 1999, Dimri 2004, Puranik and Karekar 2009).

The WH primarily comprises three prominent climatic zones; i.e. a lower Himalayan zone is composed of the Pir Panjal range, which receives most of its precipitation in the form of wet snow (Dimri 2004). The middle climatic zone consists of the Great Himalayan range; it records the lowest minimum temperature and receives a lesser amount of snow (Negi et al. 2016). The third and final upper climatic zone comprises the Zanskar, Ladhak and Karakoram ranges, which receive the least precipitation, with very low density; the temperature during winter remains well below freezing and hence the snow takes much longer to settle in this region (Negi et al. 2016).

Precipitation over the WH region occurs mainly during winter (December–March: DJFM) (Kumar et al. 2015) and WDs are the main source of the enormous amount of precipitation over northern India (Lang and Barros 2004, Dimri and Mohanty 2009, Shekhar et al. 2017a), which is important for the overall ecology and socio-economic balance. Recent studies reported a frequent increase of extreme temperature and precipitation events over the higher altitudes of the WH (Bhutiyani et al. 2007, 2010, Shekhar et al. 2010, 2017b, Malik et al. 2016). Shekhar et al. (2017a, 2017b) analysed different categories of extreme precipitation events over the WH during winter; they found that heavy precipitation events (>70 mm) significantly increased over the Pir-Panjal range and that the duration of dry spells has also increased over the WH region. Zhan et al. (2017) observed that the maximum rainy days (Rx1d, Rx3d, and Rx5d) significantly increased over the Hindu-Kush Himalayan region during 1961–2012. Subba et al. (2019) reported that the maximum precipitation (Rx1d and Rx5d) events decreased during 1997–2016 over the Eastern Himalayas (Central Himalaya). Kamil et al. (2019) showed an increase, significant at the 95% level, in the correlation between the storm frequency and precipitation over Western Himalaya–Karakoram–Hindukush (WHKH). A study by Zaz et al. (2019) revealed that winter temperature has increased significantly at all stations, and, conversely, that winter precipitation has decreased over Jammu and Kashmir region, India. Almazroui et al. (2020) reported that the winter precipitation over Northwestern Himalaya (NWH) is projected to increase in the future.

Devi et al. (2020) attempted to find the relationship between the North Atlantic Oscillation (NAO) and winter precipitation, and noted an inverse correlation over the WH. Various global parameters, such as sea surface temperature (SST), El Niño, outgoing longwave radiation (OLR) and NAO, have been used by researchers to study teleconnections with the winter precipitation mechanisms over WH (Shreshtha et al. 2000, Asnani 2005, Yadav et al. 2007, 2012, Wang et al. 2011). Hurrell et al. (2003) conducted a study outside of the WH and reported that stronger westerlies are mainly associated with the positive phase of the NAO over Western Europe.

Such significant changes in the spatial and temporal distributions of extreme precipitation events over the WH make the region a focus area to understand atmospheric processes from the perspective of planning, policy construction, and global warming. This study is the first of its kind to analyse index trends of precipitation with respect to range and altitude in the NWH region. The main objective of this study is to determine the teleconnection or link between global features, i.e. the NAO index, and the precipitation indices in the area selected for the study. The article is organized as follows: the data and methodology are presented in Section 2. Results and discussion are given in Section 3, and the conclusion is presented in Section 4.

2 Data and methodology

2.1 Data

The present study focuses on the 10 well-maintained observatories of the Snow and Avalanche Study Establishment (SASE), India, representing the NWH, as shown in Fig. 1. Daily raingauge precipitation data from 1995–96 to 2015–16 (21 years) recorded by SASE are used to represent the whole NWH. The study of seven extreme precipitation indices was carried out for the winter period (November to April) and for the seasonal precipitation (NDJFMA) for 21 years of data. The definition of the winter season (November–April) follows Shekhar et al. (2010). Seasonal precipitation depicts the cumulative amount of the monthly precipitation values. As the Himalayas are essentially affected by winter precipitation, the present study was carried out for a similar period.

Figure 1. Geographical location of 10 stations (numbered S1 to S10) over different ranges and altitudes in meters used in this study (note: the background does not depict the political boundary)

Before proceeding with the analysis of the precipitation indices, a quality check of the dataset was carried out utilizing the RClimDex toolkit. Many researchers use RCimDex to analyse indices (e.g. Almazroui 2020a, 2020b). The quality check assists in identifying the outliers and negative values in the precipitation data. The missing values in the daily precipitation data are filled using a quantile mapping approach (QM) (Devi et al. 2019). Empirical QM is useful for bias correction because it maps quantile with quantile and it is distribution-free. The study area has been categorized into three ranges (i.e. Pir-Panjal: PP, Shamshawari: SW, and Great Himalaya: GH) and three altitudes (lower altitude: 2400–2800 m, middle altitude: 2800–3200 m, and higher altitude: 3200–3800 m) (Shekhar et al. 2017b). Along with the trend for the whole NWH, the analysis of indices was also carried out over the three ranges and three altitudes.

The required North Atlantic Oscillation Index (NAOI) data used for the present study were downloaded from the site http://www.cpc.ncep.noaa.gov and further details are given by Hurrell and Deser (2009). NAOI is the difference in normalized sea level pressure between Lisbon (Portugal) and Stykkisholmur/Reykjavik (Iceland). NAOI controls the strength and direction of westerly winds across the North Atlantic. Winter precipitation over NWH is strictly linked with WDs that originate from the Mediterranean and the Caspian Sea or the Atlantic (Krishnan et al. 2019). Two previous studies, David and Fowler (2004) and Bhutiyani et al. (2010), found a connection between winter precipitation of NWH and seasonal (September–November: SON) NAOI. Hence, the current study used only one parameter – NAOI (SON) – and 21 years of data to find the relationship with the studied indices.

2.2 Methodology

The value of a parameter in a specific range/altitude over NWH is calculated by taking a simple average of the parameters of the stations falling under the same range/altitude. Several precipitation indices have been defined by experts, such as the Standard Precipitation Index (SPI) and Actual Precipitation Index (API) (Şen and Almazroui 2021). The extreme precipitation indices considered in the present study were defined and recommended by the Expert Team on Climate Change Detection, Monitoring and Indices (ETCCDMI), in association with the World Meteorological Organization (WMO) Commission for Climatology and the Research Programme on Climate Variability and Predictability (CLIVAR) (Karl et al. 1999, Peterson 2005). Seasonal indices are calculated using the average of each index. The extreme precipitation indices (R95p, R90p, RD, RX1d, RX3d, RX5d, and PRCPTOT) considered here are mainly the indicators of disasters that caused avalanches and landslides. Definitions of the indices are given in Table 1.

Table 1. Extreme precipitation indices (EPIs) and their definitions. Each index was computed for November–April and Seasonal. The base period used is 1995–1996 to 2015–2016. “Wet days” are days with daily rainfall ≥1 mm

Index (abbreviation)	Definition	Unit
R95p	95th percentile of wet days	mm/d
R90p	90th percentile of wet days	mm/d
RD	Rainy days	d
RX1d	Maximum 1-day precipitation	mm/d
RX3d	Maximum 3-day precipitation	mm/d
RX5d	Maximum 5-day precipitation	mm/d
PRCPTOT	Seasonal wet days total precipitation	mm

2.2.1 Mann-Kendall test

Mann-Kendall is the most widely used and robust non-parametric method for trend analysis. A Microsoft Excel template called MAKESENS, developed by the Finnish Meteorological Institute (Salmi et al. 2002), is used in the present study to estimate the significance of trends for time series indices. This test is useful because it does not require any particular distributional assumptions. The relationship between NAOI and the indices was found by calculating Pearson’s correlation coefficient and the year-to-year correspondence. For the year-to-year correspondence, the indices are normalized using Equation (1).

(1) $NormalizedIndices=\frac{X_i-\bar{X}}{s}$(1)

where $\bar{X}$ and s represent the mean and standard deviation of the respective indices.

3 Results and discussion

To understand the changes in precipitation over complex mountain regions, studying extreme events is required. The amount and intensity of precipitation can be used to reconstruct the overall behaviour of the climate system. For the trend analysis of precipitation indices, both parametric (i.e. linear) and non-parametric (i.e. Mann-Kendall) methods were used, at 0.1, 0.05, 0.01, and 0.001 levels of significance (Kothyari and Singh 1996).

3.1 Inter-monthly and seasonal winter precipitation indices

Figure 2(a–h) presents a comparative study of the precipitation indices (R90p, R95p, RD, RX1d, RX3d, and RX5d and PRCPTOT) for NWH during November–April and seasonally. The pattern of all seven indices for the winter months is the same, i.e. increasing from November to February and slowly decreasing thereafter, in March and April. We also noted that February is the peak month in the winter precipitation for NWH. The monthly average R90p recorded during the study period from November to April and seasonally was found to be 4.3, 9.0, 19.8, 28.7, 22.9, 17.2, and 28.8 mm, respectively, as shown in Fig. 2(a). The maximum amount of precipitation was observed in February and the minimum in November. Apart from this, a large variance was observed in January. The 95^th percentile pattern followed that of the 90^th percentile, but January’s data distribution is very small (Fig. 2b). The average R95p recorded for November to April and seasonally was found to be 10.7, 17.6, 31.8, 43.6, 34.5, 25.8, and 46.4 mm, respectively.

Figure 2. Inter-monthly and seasonal variation of seven precipitation indices for Northwest Himalaya

The number of rainy days is high in February, March, and April and low in November (Fig. 2c). Interestingly, after peak winter the rainy days increase, which indicates March and April precipitation is critical for snow cover and glacier dynamics over high mountains. The consecutive 1-d, 3-d, and 5-d maximum variance of precipitation was observed in February, whereas the minimum was observed in December (Fig. 2d, e and f). The total precipitation (PRCPTOT) of winter is high in February and low in November (Fig. 2g). In brief, the February and March precipitation is high in all events in terms of amount and intensity. We also noted that April precipitation was slightly increased in all indices and included the same number of rainy days as in February and March, which are the peak winter months.

Figure 2(h) shows the seasonal distribution of the seven indices for the whole NWH during the winter. The minimum variation was observed on rainy days, as compared to the other indices. Negi et al. (2018) found the variance of precipitation at three altitudes – Lower Himalayan (LH), Greater Himalayan (GH), and Karakoram Himalayan (KH) – over NWH to have a range of 431–804 mm.

The monthly and seasonal linear time series analysis and Mann-Kendall statistics are shown in Fig. 3(a–g) and Table 2, respectively, for the seven precipitation indices over NWH. A statistically decreasing trend was observed in November for all indices (Fig. 3a), but it is not significant. Similar trends were observed for December and January, except for RX1d in December which shows positive trends. The peak winter month (i.e. February) shows statistically increasing trends for all indices, and it is also significant for PRCPTOT at the 90% confidence level by the Mann-Kendall trend analysis, as shown in Fig. 3(d) and Table 2. Slightly decreasing and increasing trends were observed in March and April, respectively, but these were not significant (as shown in Fig. 3(e) and (f)). All the indices show increasing seasonal trends (except for RD which shows a decreasing but not significant trend) over the whole NWH. The increasing trends might be due to an influence of the peak winter month (February) on the seasonal trend (Fig. 3g). More details of the trend analysis of all months and seasons can be found in Table 2.

Table 2. Mann-Kendall trends statistics for winter months (November to April) and seasonally over the Northwest Himalaya

Mann-Kendall statistics
Index	First year	Last year	Test Z
			November	December	January	February	March	April	Seasonal
R90p	1996	2016	−1.15	−0.33	−0.33	1.00	−0.27	0.88	0.18
R95p	1996	2016	−1.06	−0.09	−0.75	0.94	−0.57	0.51	1.21
RD	1996	2016	−0.03	−0.49	−0.15	0.92	−0.46	0.76	−0.18
RX1d	1996	2016	−0.57	0.51	−0.94	1.06	−0.75	0.33	0.94
RX3d	1996	2016	−1.18	−0.15	−0.75	1.06	−0.75	0.63	1.29
RX5d	1996	2016	−1.00	−0.21	−0.82	1.18	−0.75	0.45	0.82
PRCPTOT	1996	2016	−1.12	−0.21	−0.63	1.84^+	−0.45	1.12	0.81

⁺@ 0.1 level of significance.

Figure 3. Time series of inter-monthly and seasonal variations of seven precipitation indices over Northwest Himalaya

3.2 Study of indices via ranges and altitude wise over NWH

The trend analysis of the seven precipitation indices for the three ranges of NWH (i.e. PP, SH, and GH) during the winter months and seasonally is shown in Fig. 4. November and January show decreasing trends for all seven precipitation indices in the PP range, but these are not significant. Mixed trends of indices are found for December. February, the peak winter month, shows increasing trends for all seven indices in the PP range, and it is significant for the 95^th, RX3d, RX5d, and PRCPTOT indices at a 90% confidence level. No significant trends were observed in March. In April, significantly increasing trends were observed for RD at a 95% confidence level. Increasing trends are shown for all indices seasonally in PP, which is significant for RX1d and RX3d at a 95% confidence level. Similar results were observed for extreme precipitation events over the PP range by Shekhar et al. (2017b). There are, however, mixed and non-significant trends in the SW range. GH shows decreasing trends for November, which are significant for the 95^th, RX1d, and RX3d indices at a 90% confidence level. Mixed and decreasing trends were observed for December and January, respectively. Increasing trends were observed in February for all indices. As in the PP range, RD also significantly increases during April in the GH range. However, no significant trends were observed for seasonal precipitation in the GH range, a result supported by Shekhar et al. (2017b).

Figure 4. Range-wise trend analysis of seven precipitation indices

Study of the seven indices using three altitudes [lower (2400–2800 m), middle (2800–3200 m), and higher (3200–3800 m)] of NWH was carried out for the winter months and seasonally and is shown in Fig. 5. Decreasing trends were observed at all three altitudes for all indices during November, and were significant for the higher altitude at a 90% confidence level. For December, lower and higher altitudes showed no significantly increased trends, but the middle altitude displayed increasing trends. The lower altitude showed increasing trends for January, whereas decreasing trends were observed for middle and higher altitudes. Similar to the range results, all three altitudes showed increasing trends for all indices, and these were significant at the 0.05 and 0.1 levels for February. Lower and higher altitudes showed mixed trends for all indices, whereas the middle altitude has significant decreasing trends for RX1d and RX3d, in March. A positive trend was observed for all three altitudes in April, and in the middle altitudes, decreasing trends were found from November to March. The increasing trend in April for the middle altitude might be the reason for increasing summer duration and temperature (Kohler and Maselli 2009, Immerzeel et al. 2010, Romshoo et al. 2015, 2017). Mixed seasonal trends were observed for lower and middle altitudes; however, increasing seasonal trends were observed for higher altitudes.

Figure 5. Altitude-wise trend analysis of seven precipitation indices

3.3 Teleconnection between NAOI and precipitation indices

The present study examined the teleconnections between winter and seasonal precipitation indices (R95p, R90p, RD, RX1d, RX3d, RX5d and PRCPTOT) with an average of NAOI (SON) using the correlation coefficient (CC), as shown in Table 3. During November, the precipitation indices show an inverse relationship with NAOI at the 95% and 99% confidence levels. We also detected a significantly decreasing effect of NAOI during pre-winter. Mixed CC was observed between precipitation indices and NAOI in December, although this was not significant. An inverse relationship was observed for January, February, April, and seasonal precipitation between the precipitation indices and NAOI.

Table 3. Correlation Coefficient (CC) between the North Atlantic Oscillation Index anomalies and the precipitation indices anomalies over Northwest Himalaya

Pearson’s correlation coefficient
	90th	95th	RD	RX1d	RX3d	RX5d	PRCPTOT
November	−0.33	−0.467*	−0.19	−0.568**	−0.527*	−0.492*	−0.441*
December	−0.03	0.06	−0.10	0.10	0.06	0.04	0.05
January	−0.19	−0.30	−0.13	−0.31	−0.29	−0.25	−0.17
February	−0.07	−0.29	0.10	−0.05	−0.16	−0.12	−0.01
March	0.15	0.21	0.12	0.18	0.18	0.18	0.20
April	−0.13	−0.24	−0.33	−0.39	−0.29	−0.26	−0.28
Seasonal	−0.08	−0.23	−0.18	−0.21	−0.21	−0.16	−0.05

A reduction in Arctic sea ice increases the negative phase of NAO (Nakamura et al. 2015). As the Mediterranean region is highly influenced by the negative phase of NAO, this weakens the WDs which are the major source of winter storms over the Himalayas (Syed et al. 2006). This supports the results of the present study, i.e. that NAO during SON has a negative relationship with precipitation. The influence of the mid-latitude circumglobal wave train (CGT) on winter precipitation over the Arabian Peninsula is distinct from that of the NAO (Saeed and Almazroui 2019). However, indices over NWH for March show a positive relationship with NAOI, which might be the reason for the shifting of the winter over the NWH. Apart from the CC, the year-to-year correspondence of the NAOI with seasonal precipitation index anomalies (R95p, R90p, RD, RX1d, RX3d, RX5d and PRCPTOT) over NWH is shown in Fig. 6(a–g). As the figure shows, NAOI and the indices for seasonal precipitation have an inverse relationship.

Figure 6. Year-to-year correspondence of the North Atlantic Oscillation Index with seasonal precipitation index anomalies over Northwest Himalaya

4 Conclusion

In this study, extreme precipitation events were analysed using seven precipitation indices (R90p, R95p, RD, RX1d, RX3d, RX5d, and PRCPTOT) over the NWH. The relationship between these precipitation indices and NAOI was examined. The results are as follows:

1. The pattern of all seven indices during the winter months shows a similar trend, i.e. increasing from November to February and decreasing in March and April.

2. In the Himalayan ranges, the lower range (i.e. Pir-Panjal) shows significant increasing trends in all indices for February, April, and seasonal precipitation.

3. The lower and higher altitudes show increasing trends for all indices in February.

4. A significant inverse relationship was observed between the seven precipitation indices and NAO for November, and a positive relationship was observed for March.

Acknowledgements

The authors are thankful to the Director of SASE (DRDO) for providing the necessary support and guidance to complete this work. Thanks to the technical staff of SASE for the collection of data in extreme conditions over the rugged high mountainous region.

Disclosure statement

No potential conflict of interest was reported by the authors.