Temporal variations in the wind and wave climate at a location in the eastern Arabian Sea based on ERA-Interim reanalysis data

Temporal variations in wind speed and significant wave height (SWH) at a location in the eastern Arabian Sea are studied using ERA-Interim reanalysis data from 1979 to 2012. A shallow water location is selected for the study since measured buoy data are available close to the location for comparison with the reanalysis data. The annual mean wind speed shows a statistically significant decreasing trend of 1.5 cm s −1 year−1, whereas a statistically insignificant increasing trend of 3.6 cm s −1 year−1 is observed for annual maximum wind speed due to the local events that altered the trend in annual maximum wind speed. Weakening of SWH during one of the peak monsoon months (August) is identified from the monthly analysis of SWH, which shows a higher upward trend in SWH during the southwest monsoon period, with an exception during August. The annual mean SWH shows a slight upward trend (0.012 cm year −1), whereas a larger upward trend (1.4 cm year −1) is observed for annual maximum SWH. Both identified trends are statistically insignificant. The influence of tropical cyclone activity is also studied and it is found that the maximum SWH and wind speed during 1996 are directly related to the cyclonic event.


Introduction
Knowledge of the wind and wave climate is required for engineering design (e.g.ships and offshore platforms) and shipping industry (Krogstad and Barstow, 1999;Tucker and Pitt, 2001).Wind and wave climate is assessed based on the data obtained through (i) in-situ measurement, (ii) model dataset, (iii) satellite altimeter and (iv) voluntary observing ship (VOS) data.Understanding the long-term variations in the wind and wave parameter is a key element in contemporary physical oceanography and coastal science.Long-term wind and wave records based on in-situ measurements are still sparse in the Arabian Sea (AS) and only short-term records are available (Premkumar Introduction

Conclusions References
Tables Figures

Back Close
Full  et al., 2000;Kumar, 2006;Sajiv et al., 2012;Glejin et al., 2013).A comprehensive understanding of the properties of the approaching waves and their potential changes are the major knowledge necessary for sustainable management of both offshore activities and the coastal region (Soomere and Räämet, 2011).Generally significant wave height (SWH) is used to study the wave climate at a location.Globally there have been many studies on the change in wave climate especially in north Pacific and north Atlantic region (Carter and Draper, 1988;Allan and Komar, 2000;Caires and Swail, 2004;Gulev and Grigorieva, 2004;Soomere and Räämet, 2011;Vanem and Walker, 2013).Kumar and Sajiv (2010) carried out a study on variations in long-term wind speed estimates considering different decades in the AS and reported that extreme wind speed has an annual decreasing trend of 1.3 cm s −1 .As a part of the global study, Caires and Swail (2004) examined the long-term trend in SWH in the AS based on VOS data.But the long-term changes and decadal variation in wave climate in the eastern AS is not known at present.Since no regionalized study has been done on the temporal variations of SWH in the eastern AS, a study is carried out using ERA-Interim re-analysis dataset.The objective of the study is to quantify the monthly, inter-annual and decadal variability of the wind and wave climate at a location in the eastern AS.The analysis is based on the wind speed and SWH data obtained from ECMWF reanalysis ERA-Interim data for 34 yr .Shallow water location (14.25 • N, 74.25 • E) in the eastern AS is selected for the study since measured buoy data is available close to this location for comparison of the ERA-Interim SWH data (Fig. 1).

ERA-Interim reanalysis data
Present study is based on the ERA-Interim global atmospheric reanalysis data which are produced by European Center for Medium-Range Weather Forecasts (Berrisford Introduction

Conclusions References
Tables Figures

Back Close
Full  et al., 2009;Dee et al., 2011).ERA-Interim is the first re-analysis using adaptive and fully automated bias corrections of satellite radiance observations (Dee and Uppala, 2008) and contains improvements on ERA-40 such as the complete use of fourdimensional variational data assimilation.Wind speed and SWH are downloaded for the period from January 1979 to December 2012 at 6 hourly intervals with a resolution of 0.75 • × 0.75 • latitude/longitude and used in the study.Wind speeds are derived from the zonal and meridional component of winds at 10 m a.s.l.We studied monthly variation in parameters (wind speed and SWH) by creating monthly data from the 6 hourly data and taking the mean of those months over a period of 34 yr.

Buoy data
Measured wave data at 9 m water depth (14.304 • N, 74.391 • E) off Honnavar (Fig. 1) using a moored directional wave rider buoy during January 2011-December 2012 is used in the present study for comparison of the ERA-Interim SWH data.From the heave data recorded by the buoy for 30 min duration at 1.28 Hz interval, the wave spectrum is obtained through Fast Fourier Transform (FFT).FFT of 8 series, each consisting of 256 measured heave data of the buoy, are added to obtain the spectrum.The high frequency cut-off is set at 0.58 Hz and the resolution is 0.005 Hz.SWH is obtained from the wave spectrum.

Comparison of reanalysis data with measured data
Extensive inter comparison and evaluation of wind stress estimates from reanalysis dataset (ERA-Interim) against available in-situ observations are done for the tropical Indian ocean (Praveen et al., 2013).Evaluation of the data is based on the comparison with available observations from the global tropical moored buoy array (McPhaden et al., 1998)  Temporal variation of monthly maximum and mean wind speed is plotted for 34 yr (Figs. 3 and 4).Monthly maximum wind speed show a decreasing trend during February-April, June-August and in October (Table 1).Other months show upward trend in monthly maximum wind speed.The monthly mean wind speed show a decreasing trend in all months with higher values during the peak south west monsoon season (June and August).Even though, a higher decreasing trend (∼ 4 cm s −1 yr −1 ) of mean wind speed is observed during August, the decreasing trend is less (0.079 cm s −1 yr −1 ) for maximum wind speed (Table 1).The highest upward trend in maximum wind speed is observed during September (6.14 cm s −1 yr −1 ).
Upward trend (3.6 cm s −1 yr −1 ) is observed for annual maximum wind speed and a decreasing trend (1.5 cm s −1 yr −1 ) is observed for annual mean wind speed (Fig. 5).
The difference in trends for annual maximum and mean wind speed is due to the fact that, individual events would drive the trends of annual maximum and statistically that becomes meaningless.to carry out a relevant statistical analysis.The 90th and 99th percentile values show decreasing trend (1.9 and 1.6 cm s −1 yr −1 ) similar to the trend of annual mean wind speed (1.5 cm s −1 yr −1 ).
Statistical analysis of wind speed frequency is done by classifying wind speed into different range through decadal period during (i) 1980-1989, (ii) 1990-1999 and (iii) 2000-2009 (Table 2).The study shows a decreasing trend in frequency of occurrence of stronger wind.The weakening of strong winds can be also evident from the decreasing trend of monthly mean wind speed over 34 yr.
Study on long-term variation of wind speed by Kumar and Sajiv (2010) along the AS considering different decades reported that extreme wind speed has an annual decreasing trend of 1.3 cm s −1 .Some other studies also indicated declining trend of wind speed in the last decade around the globe (Jiang et al., 2010).The decrease in frequency of occurrence of stronger wind (90th and 99th percentile) in the present study also supports the fact that the cyclonic activity in the AS is decreasing.Based on 118 yr period data of cyclonic disturbances including cyclonic storms in AS, Shyamala and Iyer (1996) examined the decadal variability and found that maximum number of cyclonic storms in AS occurred during the decade 1901-1910 (15) followed by 1961-1970 (14), 1971-1980 (13), 1981-1990 (3) and 1991-2000 (7).The decade 1981-1990 had the lowest frequency (3) of cyclonic storm and severe cyclonic storm.These are statistically significant decadal variability with decreasing tendency in decadal frequency of cyclonic storms in AS since the last 3 decades from 1971-2000.Srivastava et al. (2000) studied trends in annual cyclonic disturbances for the period 1891-1997 over the BoB and the AS and found that there is a significant decreasing trend in the annual frequency of storms over both the basins and the slopes Introduction

Conclusions References
Tables Figures

Back Close
Full of decreasing trend are maximum during the last four decades.Singh (2001) examined the long-term trend in the frequencies of cyclonic disturbances (depression) and the cyclonic storms forming over the BoB and the AS in monsoon season and found that frequency of cyclonic disturbances has decreased at a rate of about 6-7 disturbances per hundred years whereas frequency of cyclonic storms has decreased at the rate of about 1-2 cyclonic storms per hundred years.

Variation in significant wave height during 1979-2012
Monthly mean and maximum value of SWH for 34 yr are presented in Figs. 6 and 7. Higher wave heights observed during the SW monsoon months are in line with the observation on monsoon wind speed and the earlier studies on waves in AS (Kumar et al., 2000(Kumar et al., , 2003(Kumar et al., , 2012;;Sajiv et al., 2012;Glejin et al., 2013).The monthly mean SWH shows decreasing trends in August and from November to April, whereas during other months upward trend is found.SW monsoon period showed higher upward trend in mean SWH with an exception during August and maximum upward trend is observed during September (0.9 cm yr −1 ).The significant weakening of strong winds during August also reflects the cause of weakening of mean SWH.Whereas, the monthly maximum SWH shows an upward trend for most of the months except during February-April.The decreasing trend during February-April is observed in the range 0.15-0.37cm yr −1 .During SW monsoon season, the monthly maximum SWH showed a higher upward trend, with maximum during September (1.9 cm yr −1 ) similar to the mean SWH.
The annual mean SWH show slight upward trend with an increase of 0.012 cm yr −1 , whereas an increasing trend of 1.4 cm yr −1 is observed for annual maximum SWH (Fig. 8).The trend in annual maximum SWH depend on individual events and hence the 90th and 99th percentile values of SWH are estimated.The 90th and 99th percentile values of SWH increased by 0.15 and 0.76 cm yr −1 respectively (Fig. 8).The increase in annual mean is much lower than the 90th and 99th percentile and the annual maximum values, suggesting larger positive trends for higher waves.Higher rate of increase for Introduction

Conclusions References
Tables Figures

Back Close
Full 99th and annual maximum SWH indicate that extremes are increasing at a faster rate than the mean.Since the annual mean, maximum and the 90th and 99th percentile SWH have the increasing trend, the study shows that the SWH has a increasing trend at the location studied.Similar results of increasing trend in SWH can be observed in the study of Hemer et al. (2010) for Indian Ocean sector of Southern Ocean.Gulev and Grigorieva (2004) studied over 100 yr of ship observations and reported slightly increasing trend in mean wave height of 0.14 m decade −1 for the North Atlantic and 0.08-0.1 m decade −1 for the North Pacific.Neu (1984) and Bouws et al. (1996) analyzed synoptic charts issued by the Meteorological and Oceanographic Center (Canada) for the North Atlantic and largely based on ship observations.They found a trend in average wave height of 0.23 m decade −1 (1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982) whereas Neu (1984) reported larger values between 0.6 and 1.4 m decade −1 .More recent studies of buoy data in the North Pacific has shown increasing trends in average SWH between 0.05 and 0.27 m decade −1 during 1979-1999 (Allan and Komar, 2000) and 0.15 m decade −1 for the period 1976-2007 (Ruggiero et al., 2010;Young et al., 2011).The present study shows that at the study location, the mean, 90th and 99th percentile wind speed have a decreasing trend whereas the corresponding SWH have a increasing trend.This contrasting result is due to the fact that waves in the eastern Arabian Sea are mainly the swells propagating from south Indian Ocean and southern Ocean (Glejin et al., 2013;Sajiv et al., 2012) and hence are not directly related to the local phenomenon.Hemer et al. (2013) reported that increased Southern Ocean wave activity influence a larger proportion of the global ocean as swell propagates northwards into the other ocean basins (Hemer et al., 2013).The increase in swells at the study location is also evident from the increase in wave period (Fig. 9).The increase in swells at the study location is due to the increase in wind speed over the southern ocean during the past 20 yr as observed by Young et al. (2011).Introduction

Conclusions References
Tables Figures

Back Close
Full Wave climate of the AS and that along the west coast of India mainly depends on the wind conditions prevailing during SW monsoon (June-September), north east monsoon (October-January) and pre monsoon (February-May) (Glejin et al., 2013).
Winds during the SW monsoon produce high wave activity (SWH > 1.5 m) and relatively calm condition (SWH < 1.5 m) prevails during the rest of the year.Stronger storm events like cyclone produce higher waves in the region.Gray (1985) found that north Indian Ocean accounts for 7 % of global tropical cyclones.Occurrences of tropical cyclones are more in the BoB than in the AS, and the ratio of their frequencies is about 4 : 1 (Dube et al., 1997).During 1979-2008, there has been an average of 4.7 cyclonic storm days per year over the AS with 1981AS with , 1990AS with , 1991AS with , 2000AS with , 2005AS with , 2008 having 0 storms; and 1998 and 2004 having more than 15 cyclonic storm days (Evan and Camargo, 2010 ).Hence, to examine the influence of these cyclonic events on the waves, the extreme SWH during the study period are analysed.During 1979-2012, there are 3 peak events with SWH more than 4.5 m; peak1 in 1989, peak2 in 1996 and peak3 in 2000.During 1989During , 1996During and 2000, the occurrence of cyclone is observed along the west coast of India.During 1989, tropical cyclone 02A developed off the west coast of India from 7-13 June with wind speed upto 15.3 m s −1 (30 mph) but the peak1 is on 23 July and is not due to the cyclone period (7-13 June).In 1996, a tropical depression formed in the AS from 9-12 June and developed into a tropical

Influence of ENSO and Indian Ocean dipole on wind and waves
During the three peak events, the SWH observed are 4.8, 5.5 and 5.4 m and the wind speed observed are 14.0, 14.1 and 16.5 m s −1 respectively.The most logical cause of the peak identified in the wave height might be due to the increase in wind speeds over time.When the wave height is compared with the cyclonic events in the previous section, it is clear that they are not directly related for all three events.Maximum SWH during the peak2 event is only during the cyclonic event (Fig. 10) and the high SWH observed during peak1 and peak3 are not due to the cyclonic storm.This finding is not surprising since the wave climate in AS region is also influenced by the distant storms (Glejin et al., 2013).ENSO and Indian Ocean dipole (IOD) are two dominant modes of climate variations in the tropical Pacific and Indian Oceans.Both modes are shown to influence the climate conditions of several parts of the world.Some studies noticed the influence of Tropical cyclone (TC) activity on IOD mode (Saji et al., 1999;Webster et al., 1999), which in turn influences the wind pattern and can be clearly seen as high energy waves since the wind and waves are correlated.Also it is seen in literature that North Indian Ocean TC activity is notably influenced by the IOD mode.When the Indian Ocean is in a positive (negative) phase of the IOD, North Indian Ocean SST (sea surface temperature) anomalies are warm in the west (east) and cold in the east (west), which can weaken (strengthen) convection over the BoB and eastern AS, and cause anticyclone (cyclonic) atmospheric circulation anomalies at low levels.This result in less (more) TC genesis and reduced (increased) opportunities for TC occurrence in the North Indian Ocean (Yuan and Cao, 2012).Webster et al. (1999) indicated that IOD events are tied to strongly coupled ocean-atmosphere-land interactions, and are important in Indian-Pacific regional climate variations.On the other hand, ENSO, is a coupled ocean-atmosphere phenomenon of the tropical Pacific, also show significant effects on tropical cyclone as well as in climatic signals in Indian Ocean region (Jury, 1993;Reason et al., 2000).To extend the argument for a relationship between the IOD, ENSO with observed SWH and wind speed in the AS, the annual average maximum SWH and wind speed is correlated with the IOD and ENSO indices.An IOD event can be detected using Dipole Mode Index (DMI) which is defined as the difference in SST anomalies between western and eastern equatorial Indian Ocean.The oceanic Nino index (ONI) is used for identifying El-Niño (warm) and La-Niña (cool) events in the tropical Pacific.It is the running 3 month mean SST anomaly for the Niño 3.4 region (5 • N-5 • S, 120-170 • W).Events are defined as 5 consecutive months at or above the +0.5 • anomaly for warm (El-Niño) events and at or below the −0.5 • anomaly for cold (La-Niña) events.The threshold is further broken down into weak (with a 0.5-0.9 • SST anomaly), moderate (1.0-1.4 • ) and strong (≥ 1.5 • ) events.For determining the El-Niño/La-Niña and positive/negative IOD years, we have downloaded Nino 3.4 index and IOD mode index from the NOAA and JAMSTEC websites and both of ONI and dipole mode indices were overlaid to one graph (Fig. 10c).We have analysed DMI and Nino 3.4 index with SWH and wind speed (Fig. 10a and b) and found that during the peak events (peak1 and peak2), the Nino 3.4 index and DMI shows high negative trend.During a positive IOD year, the eastern Indian Ocean is colder than normal while the western Indian Ocean is warmer than normal.Since the DMI is defined as a difference between the western box and the eastern box, the index is positive during a positive IOD year and is negative during negative IOD year.It has been noted that ENSO remote forcing leads to an Indian Ocean basin scale warming in the following spring after the ENSO matures (Xie et al., 2002).Nearly 40 % of IOD occurs simultaneously with ENSO; the positive IOD is associated with El-Niño and the negative IOD generally with La-Niña (Saji and Yamagata, 2003;Ashok et al., 2004).After carefully observing the pattern of these two index with yearly maximum SWH and wind speed, it is found that, during certain years when both the index shows positive value, the SWH is found to be less.The co-occurrence of both events, positive IOD events (Yamagata et al., 2002;Vinayachandran et al., 2007) and El-Niño can be observed in many years (1982, 1991, 1994 and 1997) noted as phase 1.The co-occurrence of negative IOD events Introduction

Conclusions References
Tables Figures

Back Close
Full  (Meyers et al., 2007) and La-Niña can also be observed during 1989 and 1998 noted as phase 2. During phase 1, 1982, 1994 and 1997 are strongest positive IOD years, an anticyclonic atmospheric circulation at low level can be observed (Yuan Junpeng and Cao Jie, 2012) which in fact reduces the cyclonic occurrence over AS.Significant reduction in the maximum wind speed and maximum SWH can be observed during this period.Also in reverse, extreme SWH and wind speed are observed during phase 2 (1989 and 1998).

Statistical trend analysis for the time series data
A statistical analysis of the ERA-Interim wave and wind time series has been performed to test the significance of trends during the 34 yr period.A Student's t test and z test have been used to assess the probability distribution with N-1 • of freedom with level of significance (α) fixed at 5 %.The result of the t test and z test are presented in Table 3.The null hypothesis of zero-slope is rejected for all the parameter at a 5 %level of significance.For all parameter as the computed p value for t test and z test is lower than the significance level α = 0.05, one should reject the null hypothesis H0, and accept the alternative hypothesis Ha (Where H0: the difference between the means is equal to 0, Ha: the difference between the means is different from 0).There are several statistical tests available for testing the time series data (Hirsch et al., 1993).These tests start from a null hypothesis that the observations are samples from a stationary process.The likelihood of this hypothesis is evaluated based on the value of a test statistic, a property of the data set.A p value of 5 % is a common critical value for accepting statistical significance, but there is no reason why other values cannot be used.Rejection of the null hypothesis at the 5 % significance level means that we are 95 % confident of non-stationary.The Mann-Kendall test is a non-parametric significance test for a monotonic trend in a time series based on the Kendall's tau (τ) (Mann, 1945;Kendall, 1975).We have carried out a two tailed Mann-Kendall test with the linear fitted trend for the long-term trend of annual wave and wind parameter.The result is presented in Table 4.As the computed p value is lower than the significance Introduction

Conclusions References
Tables Figures

Back Close
Full level alpha=0.05,one should reject the null hypothesis H0, and accept the alternative hypothesis Ha, (where Ha: there is a trend in the series, H0: there is no trend in the series).Statistically significant trend is found for wind speed and wave period mean trends, the null hypothesis of zero-slope is rejected for this cases at a 5 %level of significance.The Theil-Sen slope (Sen, 1968;Theil, 1950), also known as "Kendall's slope" or the "Nonparametric linear regression slope", is an alternative to the standard linear regression slope.Theil Sen's slope estimate is also is method for robust linear regression that chooses the median slope among all lines through pairs of two-dimensional sample points.Positive and negative value in Sen's slope indicates positive and negative trend respectively.

Conclusions
Long-term trend in wind and wave along the eastern AS is analysed using ERA-interim dataset.The study shows that during 1979-2012, the annual mean wind speed shows a decreasing trend (1.5 cm s −1 yr −1 ) whereas the annual maximum wind speed showed an upward trend of 3.6 cm s −1 yr −1 .The annual maximum and average wind speed during 1979-2005 has a decreasing trend of 1.4 and 0.8 cm s −1 yr −1 respectively.The higher wind speed associated with the local depression events occurred in AS during 2006 and 2007 shifted the trend of annual maximum wind speed during 1979 to 2012 from negative to positive.During the SW monsoon period higher upward trend in significant wave height (SWH) is observed with an exception during August.Annual mean SWH show small increasing trend (0.012 cm yr −1 ), whereas higher increasing trend of 1.4 cm yr −1 is observed for annual maximum SWH.Detailed analyses have been carried to examine the influence of the cyclonic events on the intensity of wave activity and wind speed and found that the peak2 event during 1996 is due to a cyclonic event.Further analysis with ENSO and IOD events shows that, during a positive IOD and La-Niña year, the observed annual average SWH and wind speed is comparatively lower whereas high SWH and wind speed are observed during a negative IOD and El- Full  Full  Full  Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 3.3 Intensity of wind speed and wave height during the peak events in AS

storm on 17
June off the west coast off India.This severe cyclonic storm during 17-20 June had peak intensity of 30.6 m s −1 and generated the peak2 event.During 2000, a depression formed at BoB, on 26 November and intensified into a storm and crossed the Indian sub-continent.But the peak3 is on 7 June and not during the cyclone period (26-30 November).Simon et al. (2001) found an increase in wind speed over the AS off Karnataka coast on 6-7 June based on the multi frequency microwave radiometer study.The increase in the local wind speed during 6-7 June is the reason for unusual increase in SWH during the peak3 event.Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Niño year and resulted in peak1.Whereas the peak3 was due to the locally formed low pressure eventDiscussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

3 Results and discussions 3.1 Variation in wind speed during 1979-2012
and OceanSITES (http://www.OceanSITES.org,2009) data with observations and reported that ERA-Interim data captures well temporal variability with better performance (∼ 0.86 correlation).For the present study, the nearest Introduction

Table 2 .
Frequency of occurrence of wind speed in different decades.

Table 3 .
Student t test and Z test for wave and wind.

Table 4 .
Mann-Kendall significance test for trend analysis.