The Effect of Rainfall Anomalies Due to Enso on tea Productivity in West Java Indonesia (Case Study: Bandung Regency)

This study aims to determine the effect of rainfall anomalies due to ENSO on tea productivity in Bandung Regency West Java Indonesia. This study used data on monthly rainfall, tea productivity, and the Nino 3.4 index from 2003-2020. The methods used are Fast Fourier Transform (FFT), Butterworth Bandpass Filter, and Inverse Fast Fourier Transform (IFFT) for the processing of these three data. The results of this study indicate that the rainfall pattern in the Bandung Regency area in 4 districts (Rancabali, Pasirjambu, Pangalengan, and Kertasari) is monsoon. ENSO correlation to rainfall anomalies and rainfall anomalies correlation to tea productivity in each district is low. The highest correlation between ENSO to rainfall anomalies is found in Pangalengan District (-0.309) and the highest correlation between rainfall anomalies to tea productivity is found in Kert asari District for lag 0 and 1 which are -0.304 and -0.285 respectively. The low correlation (-0.148 to - 0.309) showed that the effect of rainfall anomalies due to ENSO on tea productivity of each district in Bandung Regency has a small effect but with a significance value of less than 5% (0.000 to 0.030) which means it has a significant relationship.


Introduction
The territory of Indonesia is an area traversed by the equator and is located between two continents and oc eans, as well as the place of convergence of two major circulations in the world, namely Walker and Hadley circulation which greatly affects the level of rainfall variability in Indonesia.This condition causes Indonesia as a tropical region to have high rainfall variability both in space and time scales.Moreover, both the spatial and temporal rainfall variability is also amplified: higher intensity and more frequent rainfall have occurred during the wet season and longer drought during the dry season (Held & Soden 2006;Zhang & Fueglistaler 2019) An extensive study using a global rainfall data series had suggested an increase in global annual maximum daily precipitation (Westra, Alexander & Zwiers 2013).Positive trends were also indicated through regional studies , such as in northern Australia (Roderick, Wasko & Sharma 2019), India (Goswami et al 2006), and many other regions.(As-syakur et al 2013) observed an increasing rainfall over Indonesia, especially in Kalimantan, J ava, Sumatra, and Papua, from 1998 to 2010, based on the TRMM multisatellite dataset.
According to Aldrian and Susanto (2003), the rainfall spectrumin Indonesia shows the dominance of annual and semi-annual climate signals.Rainfall in Indonesia is influenced by various weather or climate factors, both globally and locally (Yulihastin et al., 2010).Globally, rainfall conditions in Indonesia are influenced by the Asian-Australian monsoon, ENSO, and Indian Ocean Dipole (IOD) phenomena (Zubaidah, 2012).ENSO and Indian Ocean Dipole (IOD) phenomena are related to inter-annual rainfall variations in Indonesia.The main factor affecting climate variability in Indonesia is ENSO.ENSO is the existence of Walker circulation disorder in the Pacific Ocean (Aldrian and Susanto, 2003).The El Nino phenomenon is caused by the warming of the sea surface temperature in the equatorial Pacific Ocean which causes a decrease in rainfall so that the dry seas on is longer than the rainy season Ocean which causes a decrease in rainfall so that the dry season is longer than the rainy season Ocean which causes a decrease in rainfall so that the dry season is longer than the rainy season while La Nina is caused by the cooling of the sea surface temperature in the equatorial Pacific Oc ean w hic h 1245 (2023) 012033 IOP Publishing doi:10.1088/1755-1315/1245/1/012033 2 causes an increase in rainfall so that the rainy season is longer than the dry season (As-syakur, 2010).In addition, the high variability of rainfall is also influenced by local conditions.The amount of rainfall variability can have an impact on the agricultural sector which indirectly affects food, welfare, life, and s ocial economy in the community (Baidu et al., 2017).
West Java province is a major tea producer that produces 79% of tea in Indonesia.West Java province has a plantation area of 484.234 ha.A total of 84.70 thousand ha are tea plantations.PT.Perkebunan Nusantara VIII exports 40% of the tea that has been processed abroad.According to the West Java provincial plantation offic e in 2020, West Java produced 101.441 tons out of 128.016 tons, equivalent to 79% of national tea produc tion.Bandung Regency produced 35.633 tons, equivalent to 35% of West Java province's tea pr oduc tion in 2020.This indicates that Bandung Regency is one of the largest tea producers in Indonesia.However, tea production in West Java over the past 10 years (2011-2020), especially in Bandung Regency varies so that there is a decrease and increase that causes instability of tea productivity shown in Figure 1  The 1987 drought reported by Sukasman (1987) showed that almost all tea plants in West Java experienced severe drought resulting in decreased production and high mortality in young plants, especially in the lowlands.Then, when the drought occurred in 1997 the tea plant experienced a drought that resulted in a decrease in production to reach 53% (Kartawijaya, 1992).During the period 2005-2014 El-Nino phenomenon occurred with the strongest intensity in late 2009 and early 2010 (Dalimoenthe et al., 2016).According to Dalimoenthe (2016), there was a decrease in the amount of rainfall in the highlands (Bandung) in 2010-2014 (after El-Nino) compared to 2005-2009 (before El Nino).Also, in 2015 there were El-Nino symptoms with very strong intensity, with the Nino index reaching its peak at 2.483 (BMKG, 2015).Rainfall intensity in West Java and sea surface temperatures on the north and south coasts of West Java decreased during the 2015 El Nino.
Tea productivity in Bandung Regency is influenced by climate variability, one of which is ENSO.In 2012 there was an event of La Nina effect on the length of the rainy season and high rainfall that caused the produc tivity of tea fell in Bandung Regency.Meanwhile, in 2015 there was an El Nino effect on the length of the dry season and low rainfall which caused annual rainfall in Bandung Regency of 2311 mm and had an impact on s lowing the rate of decline in tea productivity (Rahman, 2017).It also proves that El Nino and La Nina play a role in reducing and increasing rainfall in Bandung Regency.Therefore, there is a need for research on the effect of rainfall anomalies due to (only) ENSO on tea productivity in West Java with case study of Bandung Regency).This means that the research consider only ENSO not the others and also be done for very limited area.

Data
In this study, Nino 3.

Methods
This study used methods to see the impact of ENSO on rainfall anomalies and rainfall anomalies on tea productivity by spectral analysis method (FFT), bandpass filter, IFFT, and correlation.First of all, there is a process called "calculating rainfall anomalies".The rainfall anomaly is the difference between the actual value of precipitation and the average value of precipitation for the long term (climatology).Rainfall anomalies are calculated using Equation 2 The methods used in this study are as follows.

2.2.1Spectral Analysis Method
The spectral analysis method is a method to transform from the time domain to the frequency domain, so it looks at periodicity to then determine the type of weather patterns involved as a peak spectral energy at a particular frequency.The spectral analysis method used is Fast Fourier Transform (FFT).This method is us ed in the data Nino 3.4 and rainfall anomalies to see the periodicity of ENSO events.

Butterworth Bandpass Filter
The butterworth Filter is one of the most widely used classical filter methods (Paarman, 2003).Butterworth Bandpass Filter method is used to separate the signal, weaken, eliminate certain frequency signals that are not wanted to escape the desired signal, then determined the value of the low cutoff, high cutoff, and the order of the bandpass.Determination of cutoff value in this study is reviewed from the results of Fast Fourier Transform (FFT).This method is used in Nino 3.4 index data, rainfall anomalies, and tea productivity to see the signal of ENSO events.

Invers Fast Fourier Transform (IFFT)
The inverse Fast Fourier Transform (IFFT) method is the opposite of the Fast Fourier Transform (FFT ) us ed in digital signal processing applications.IFFT is a method of converting the frequency domain into the time domain.In other words, in the IFFT stage, there will be a process of returning the signal f r om the f r equenc y domain to return to the time domain after the FFT-bandpass filter process.

Correlation
Correlation is one of the statistical analysis techniques used to find the relationship between two variables (independent variables with dependent variables) that are quantitative.Interpretation the value of correlation can be seen in the box below.The hypotheses used in a correlation analysis are as follows: 1.The significance value of t < α = 0,05 (5%), then H0 is rejected and H1 is accepted means that there is a significant influence between the independent variable to dependent variable.2. The significance value of t > α = 0,05 (5%), then H0 is accepted and H1 is rejectedmeans that there is no significant influence between the independent variable to dependent variable.

Rainfall Pattern in Bandung Regency
Based on the results of the calculation of the average monthly rainfall for 18 years (2003-2020) in Figure 3.1 shows that the rainfall pattern of four districts in Bandung Regency, West Java representing each estate (Rancabali, Pasirjambu, Pangalengan, and Kertasari) has a monsoon rainfall pattern that is unimodial because of the characteristics that are strongly influenced by the Asian-Australian monsoon system, these results November-December (Rancabali, Pangalengan, Pasirjambu) and the peak of the dry season occurs in June-August.Then, in Kertasari District, the peak of the rainy season occurs in December-February because it coincides with the west monsoon while the peak of the dry season occurs in June-August to coinc ide with the east monsoon.
The Western Monsoon occurs around October to April.The Western Monsoon occurs due to the position of the sun in the southern hemisphere.As a result, the Australian Continent experiences summer so that it has low pressure, while the Asian Continent is colder so that it has high pressure.Due to the difference in pressure between the two continents, the wind moves from the Asian Continent to the Australian continent.T he wind that blows from the Asian Continent through the vast ocean that is the Pacific Ocean and the South China Sea then the wind brings a lot of water vapor that causes rain in Indonesia and its surroundings Very strong the northern hemisphere.As a result, the Asian Continent experiences summer so that it has low pressure, while the Australian Continent is colder so that it has high pressure.Due to the difference in pressure between the two continents, the wind moves from the Australian Continent to the Asian Continent.The wind blows from the Australian Continent through the desert in northern Australia and only passes through a little ocean, the w ind is dry and only passes a little water vapor that causes drought in Indonesia and surrounding areas.

The Effect of Interannual Phenomena on Rainfall Anomalies
To determine the effect of interannual phenomena, namely El Nino and Southern Oscillation (ENSO) on rainfall anomalies in four sub-districts in Bandung Regency, West Java is done with steps as below.

FFT-Bandpass Nino 3.4
The results of FFT-bandpass and the original FFT data from Nino 3.4 are shown in Figure 3. 3, w hic h s hows that the results of FFT-bandpass and FFT are appropriate   The results of overlay FFT-bandpass and FFT original data in Figure 3.3 (b), selected 1/24 as high cutoff and 1/96 as low cutoff, the value is taken based on interannual phenomena that occur around 2-8 years .From the figure shows the results of spectral analysis (FFT) Nino 3.4 there is the strongest period (peak spectral energy) which is 36, 54 and 72 means that there is interannual variation corresponding to the period of the ENSO phenomenon cycle (El Nino and La Nina) which is 3-6 years where the period or phenomenon is quite influential at the point of study.

FFT-Bandpass Rainfall Anomaly
The results of FFT-bandpass and FFT rainfall anomalies in four districts in Bandung Regency showed the suitability of the pattern as shown in Figure 3

The effect of ENSO Phenomenon on Rainfall Anomaly
The results of the IFFT-bandpass Nino 3.4 and the rainfall anomaly in each sub-district are presented in the plot of Figure 3.5.The black and blue horizontal solid wave lines are Nino 3.4 and rainfall anomalies, respectively.The red (greater than 0.5) and blue (smaller than -0.5) dotted lines show the thresholds for El Nino and La Nina events, respectively.The red and blue vertical lines indicate El Nino and La Nina occurrences, respectively.

The effect of Rainfall Anomalies due to ENSO on Tea Productivity
Analysis of the effect of rainfall anomalies due to ENSO on tea productivity was conducted in the dis tric ts of Rancabali, Pasirjambu, Pangalengan, and Kertasari from the results of IFFT-bandpass Nino 3.4, rainfall anomalies, and tea productivity.Based on Figure 3.6, during the moderate El Nino event in 2009 there was a decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by an increase in tea productivity.During the strong El Nino event in 2015 there was a decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by an increase in tea productivity.

Rancabali District
During the strong La Nina event in 2008 there was an increase in rainfall which was indicated by a positive rainfall anomaly.This was accompanied by a decrease in tea productivity.During the strong La Nina event in 2011 there was an increase in rainfall which was marked by a positive rainfall anomaly.This was accompanied by a decrease in tea productivity.Based on Figure 3.7, during the moderate El Nino event in 2009 there was a decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by an increase in tea productivity.During a strong El Nino event in 2015 there was a slight decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by a decrease in tea productivity.

Pasirjambu District
During the strong La Nina event in 2008 there was an increase in rainfall which was indicated by a positive rainfall anomaly.This was accompanied by a decrease in tea productivity.During the strong La Nina event in 2011 there was a slight increase in rainfall which was indicated by a positive rainfall This was accompanied by a decrease in tea productivity.Based on Figure 3.8, during the moderate El Nino event in 2009 there was a slight decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by an increase in tea productivity.During the strong El Nino event in 2015 there was a decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by an increase in tea productivity.

Pangalengan District
During the strong La Nina event in 2008 there was an increase in rainfall which was indicated by a positive rainfall anomaly.This was accompanied by a decrease in tea productivity.During the strong La Nina event in 2011 there was a slight increase in rainfall which was indicated by a positive rainfall anomaly.This was accompanied by a decrease in tea productivity.Based on Figure 3.9, during the moderate El Nino event in 2009 there was a slight decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by an increase in tea productivity.During the strong El Nino event in 2015 there was a decrease in rainfall which was marked by a negative rainfall anomaly.This was accompanied by an increase in tea productivity.

Kertasari District
During the strong La Nina event in 2008 there was an increase in rainfall which was indicated by a positive rainfall anomaly.This was accompanied by an increase in tea productivity.During the strong La Nina event in 2011 there was a slight increase in rainfall which was indicated by a positive rainfall anomaly.This was accompanied by a decrease in tea productivity.

Tea Productivity Pattern in Bandung Regency
The monthly average tea productivity in the study area is shown in Figure 3.10.In May generally shows an increase in productivity while September is the month with the lowest tea productivity.Although other fac tors such as management also have an effect, in this study this was ruled out.When compared to the rainfall pattern chart with the tea productivity pattern chart in Figure 3.10 shows there is a one-month time lag between rainfall patterns and tea productivity in the four districts.For example, when rainfall begins in May, productivity begins to decline 1 month later (lag 1).

Correlation of ENSO with Rainfall Anomalies
The results of IFFT-bandpass index Nino 3.4 correlated to IFFT-bandpass rainfall anomalies.The correlation values are low in the four districts with a significance value of less than 0.05 (5%) which means that there is a significant influence between the Nino 3.4 index (ENSO) on rainfall anomalies.
All study points have a negative correlation showing an inverse relationship between rainfall anomalies and the Nino 3.4 index.This means that if the Nino 3.4 index increases (higher positive) above 0.5 then the rainfall anomaly will decrease (more negative).Rising sea surface temperatures above normal conditions in the central and eastern Pacific Ocean (El Nino) which will weaken the trade winds blowing from east to west (inc reas ing temperatures in the eastern Pacific) causing a decrease in rainfall in Indonesia, including in four districts of Bandung Regency.Meanwhile, if the Nino 3.4 index decreases (more negative) below -0.5 then the rainfall anomaly will increase (more positive).The decrease in sea surface temperature below normal conditions in the central and eastern Pacific Ocean is known as the La Nina phenomenon.The trade winds blowing from east to west are strengthening (cold temperatures in the eastern Pacific) causing increas ed rainfall in Indonesia including four sub-districts, Bandung Regency.

Correlation of Rainfall Anomalies with Tea Productivity
The results of IFFT-bandpass rainfall anomalies correlated to IFFT-bandpass tea productivity.The correlation values at lags 0 and 1 are low in the all districts with a significance value of less than 0.05 (5%), meaning that the rainfall anomaly has a significant effect on tea productivity in the four districts.T he highes t correlation value of all districts is Kertasari District at lag 0 and lag 1 which are -0.304 and -0.285 with a significance value of 0.000 respectively.
The results of this correlation and significance level show that the effect of rainfall anomaly on tea productivity is relatively weak so it can be said that rainfall has a small effect on tea productivity in the region.T his is als o following the results of research conducted by Hendro Prawiro (2009) and Lou et al (2020), who said that a decrease in rainfall is not always followed by a decrease in tea productivity.Climate conditions with high rainfall can cause fungi activity resulting in a decrease in production which results in a decrease in tea productivity.Meanwhile, when rainfall decreases, sunlight becomes more increase.Thus, the process of photosynthesis runs faster to produce more energy and growth goes fast. .

Summary
Based on the results of this study, several conclusions were obtained, which is the correlation of ENSO to rainfall anomalies in each district with the highest value of Pangalengan District (-0.309) with a significance value of 0.000.The correlation of rainfall anomalies to tea productivity in each district is low with the highes t value of Kertasari District for lag 0 and lag 1 are -0.304 and -0.285 with a significant value of 0.000 respectively.Rainfall anomalies due to ENSO on tea productivity in each district of Bandung Regency can be said to have small effect, this is evidenced by the low correlation value of -0.148 to -0.309 and has a significance value of less than 5% of 0.000 to 0.030 so that there is a significant relationship.So although rainfall anomalies impact on tea productivity is small but this value is significant.We suggest to investigate more closely this relationship for larger scale and for the other phenomenons such as Monsoon, Dipole Mode, Madden J ulian Oscillation, Semi Annual Oscillation, etc. .1.

Figure 1 . 1
Figure 1.1 Graph of tea productivity in Bandung Regency and West Java Province in 2011-2020.
4 index data from 2003-2020 from KNMI Climate Explorer is the result of data 1245 (2023) 012033 IOP Publishing doi:10.1088/1755-1315/1245/1/0120333 processing taken from NOAA's SST Optimum Interpolation (OI) version 2 (v2) with monthly temporal resolution.Then, rainfall data from ERA5-Land reanalysis data with a spatial resolution of 0.1° × 0.1° and monthly temporal resolution from 2003-2020.In addition, monthly tea productivity data from the Sinumbra plantation in Rancabali District, Rancabolang plantation in Pasirjambu District, Pasirmalang plantation in Pangalengan District, and Sedep plantation in Kertasari District in Bandung Regency in 2003-2020 were obtained from PT Perkebunan Nusantara VIII, Bandung, West Java.The study area was in Bandung Regency, West Java, located at 6 o 49 '-7 o 18' LS and 107 o 14' -107 o 56 ' BT with an area of 176.238,67 ha.

Figure 3 .
Figure 3.3 (a) and (b) show that the red area is an ElNino event, and the blue area is a La Nina event that s hows the pattern produced by FFT-bandpass already represents the original data (made overlay plots index Nino 3. 4 and FFT-bandpass).It is because the peaks and valleys of the bandpass filter are in the places that correspond to the original data.El Nino and La Nina events according to the Ocean Nino Index (ONI) in the region of 3.4 in 2003-2020.There were 7 El Nino events and 8 La Nina events from 2003-2020 shown in Table 3.1.

4
Gambar 3.4 The results of time series plots, (a), (c), (e), (g) comparison of IFFT-bandpass rainfall anomalies and original rainfall anomaly data of each sub-district in Bandung Regency, (b), (d), (f), (h) FFT of bandpass data and original rainfall anomaly data of each sub-district in Bandung Regency.

Figure 3 . 5
Figure 3.5 The result of the IFFT-bandpass Nino 3.4 time series plot and rainfall anomalies in each sub-

Figure 3 . 6
Figure 3.6 The results of the IFFT-bandpass Nino 3.4 times series plot, rainfall anomalies, and tea productivity in Rancabali District in 2003-2020.

Figure 3 . 7
Figure 3.7 The results of the IFFT-bandpass Nino 3.4 times series plot, rainfall anomalies, and tea productivity in Pasirjambu District in 2003-2020.

Figure 3 . 8
Figure 3.8 The results of the IFFT-bandpass Nino 3.4 times series plot, rainfall anomalies, and tea productivity in Pangalengan District in 2003-2020.

Figure 3 . 9
Figure 3.9 The results of the IFFT-bandpass Nino 3.4 times series plot, rainfall anomalies, and tea productivity in Kertasari District in 2003-2020.

Table 3 .
2 shows the results of the correlation of the Nino 3.4 index with the rainfall anomaly in 2003-2020.

Table 3 .
3 Correlation and significance level between rainfall anomalies and tea productivity

Table 3 .
3shows the results of the correlation of rainfall anomalies with tea productivity in 2003-2020.