Abstract

In this study, temporal trends in daily time series data of key climatic parameters were analyzed using Mann–Kendall and Sen's slope estimator. Sensitivity analysis of each climatic parameter on reference evapotranspiration (ETo) was performed to estimate the sensitivity coefficients and to evaluate the impact of global warming on ETo in the eastern Himalayan region of Sikkim, India. Results of trend analysis showed a significant increasing trend for minimum temperature and mean temperature. Mean relative humidity and sunshine duration showed decreasing trends. Reference evapotranspiration also showed a significant decreasing trend by 0.008 mm year–1 in Sikkim state of India. Sensitivity analysis revealed that the seasonal and annual ETo were most sensitive to maximum temperature followed by sunshine hours whereas wind speed, minimum temperature and relative humidity had a fluctuating effect on mean ETo. The sensitivity coefficient indicated that ETo changes positively with maximum and minimum temperature, sunshine hour, and wind speed, while it changes negatively with relative humidity. Analysis indicated that increase in relative humidity would decrease the ETo in the study area. The findings of this study would be useful for sustainable water resources planning and management of agriculture in hilly regions of the state and for development of adaptation strategies in adverse climatic conditions.

INTRODUCTION

The worldwide agriculture sector is the largest consumer of water therefore, effective planning and management of water use in agriculture is very important and needs special attention. Water demand in agriculture is represented by the crop water requirement (CWR) and is usually calculated from reference crop evapotranspiration (ETo) and crop coefficient (Patle et al. 2013). ETo is an important variable of the hydrologic cycle and plays a significant role in the estimation of the evaporative demand of the atmosphere independently of crop type, crop development, and management practices for a given location and time (Tabari & Talaee 2014). Temperature, relative humidity, wind speed and sunshine duration and rainfall are the key elements of climate and any change in climatic parameters would affect the CWR and future planning and management of water resources (Goyal 2004; Patle et al. 2013). Water demand is also increasing from the different sectors and is likely to increase due to climate change in the future (Patle et al. 2013). Air temperature is the most widely used indicator of climatic changes, both on global and regional scales (Bandyopadhyay et al. 2009). Therefore, precise estimation of ETo is essential for predicting actual crop evapotranspiration, water management, irrigation scheduling, and other practices of agricultural production (Ying et al. 2013; Patle & Singh 2015).

Several methods are available for the estimation of ETo and these vary from empirical- to physical-based models (Jabloun & Sahli 2008; Maina et al. 2014; Kulkarni et al. 2015; Mahida & Patel 2015). The selection of a method for the determination of ETo depends upon the type of meteorological data available for the given region and the accuracy desired in the computation of water needs (Joshani et al. 2015). FAO recommended the universal adoption of the Penman–Monteith combination method for estimation of reference evapotranspiration (Allen et al. 1998). Several researchers evaluated the different methods of reference evapotranspiration and compared them with the FAO Penman–Monteith method and reported that FAO-56 PM is more superior for estimating crop evapotranspiration (Chiew et al. 1995; Droogers & Allen 2001; Hamdi et al. 2008; He et al. 2013). ETo is expected to increase with temperature rise (Goyal 2004). Global warming and associated changes in climate are expected to affect the climatic parameters to varying degrees. This in turn would affect the reference ETo (Patle et al. 2013). Analysis of long term time series of climatic data is useful to detect the change in climatic variables and reference evapotranspiration using non-parametric methods, namely Mann–Kendall tests and Sen's slope estimator (Jhajharia et al. 2012; Wang et al. 2014; Gao et al. 2016).

Sensitivity analysis of climatic parameters plays an important role to understand the relative importance of each climatic variable in the Penman–Monteith formulation (Saltelli et al. 2004). The results from sensitivity analysis are of vital significance for determining the effect of climate change on ETo (Ying et al. 2013; Patle & Singh 2015). Generally, sensitivity analysis is carried out by plotting the relative changes in a dependent variable (ETo) against the relative changes in independent variables, namely temperature, wind speed, relative humidity and sunshine hours, etc. (Porter et al. 2012). The sensitivity coefficient derived from observed data may be used to estimate ETo for areas where detailed meteorological records are lacking (Liang et al. 2008; Schrier et al. 2011). According to Gong et al. (2006), sensitivity analysis is important in understanding the connection between climatic conditions and ETo variability and between data availability and estimation accuracy of ETo. The results of sensitivity analysis make it possible to determine the accuracy required when measuring climatic variables used to estimate ETo (Irmak et al. 2006; Vicente-Serrano et al. 2014).

Several studies were conducted on the sensitivity of reference crop evapotranspiration (ETo) to examine the expected change in ETo in response to a known change in one of the climatic variables while values of other climatic parameters remain unchanged (Rana & Katerji 2000; Irmak et al. 2006; Bormann 2011). The sensitivity of ETo to changes in meteorological parameters were studied by Kaboosi & Kaveh (2010), Ambas & Baltas (2012), Admasu et al. (2014), Sharifi & Dinpashoh (2014), Zhao et al. (2014), Dejonge et al. (2015) and Gao et al. (2016) under varying climatic conditions of different countries. Hou et al. (2013) carried out the sensitivity of ETo to key climatic variables in the Ejina oasis, northwest China, using a non-dimensional relative sensitivity coefficient. Sensitivity of the ETo to principal climatic parameters was studied by Goyal (2004) for the arid region of Rajasthan and for the semiarid region of Karnal Haryana in India by Patle & Singh (2015). Mosaedi et al. (2016) analyzed the sensitivity of the monthly ETo to climatic factors in Iran by using a qualitative detrained method. This shows that sensitivity analysis is required to gain a better understanding of the meteorological systems, particularly to indicate the physical meaning of each meteorological parameter used in the estimation of ETo.

Sikkim is an entirely hilly state of India situated in the foothills of the eastern Himalayan range which has varying climatic conditions due to altitudinal variations. A variety of field crops are grown in the region due to the available diverse agro-climatic conditions. Crop water requirements (CWRs) vary in the region on a temporal and spatial scale due to variation in the climate and the soil and crop physiology. Mountainous regions are more vulnerable to climate change. Therefore, the variation in the climatic parameters due to global warming may affect water availability and rural livelihoods based on agriculture. The water need of the state is mostly fulfilled by natural springs originating in the mountains and about 80% of the rural households depend on spring water for their rural water security. The effect of climate change may vary the water demand in hill agriculture due to changing CWRs.

Considering the above facts this study was undertaken to detect the temporal trends in the major climatic parameters which influence the reference crop evapotranspiration, and further to analyse the sensitivity of reference crop evapotranspiration (ETo) to climatic parameters for the east district of Sikkim which is under intensive organic cultivation. The sensitivity of ETo trend was analyzed quantitatively, i.e. by changing the value of each climatic variable within a specific range (e.g. ±20%, in ±5% steps) and keeping other parameters unchanged.

METHODS

Study area

The Sikkim state of India is a pioneer in organic agriculture. The study area comprises the east district of Sikkim and is situated between 27 °9′ to 27 °25′ N latitude and 88 °27′ to 88 °56′ E longitude with a total geographical area of 954 km2 (Figure 1). The topography of the district is completely hilly and elevation ranges from 300 to 5,000 m above mean sea level (MSL). The district is a part of eastern Himalayan region as per the classification of agro-climate zone. The average annual rainfall of the district is about 2,525 mm and consists of about 135 rainy days in a year. The district receives the major portion of rainfall from the southwest monsoon which contributes 61% of total rainfall.

Figure 1

Location of study area.

Figure 1

Location of study area.

As per the land use statistics, the district consists of 10,500 ha cultivated area, 9,112 ha area under forest and 3,277 ha area under non-agricultural use. The depth of soil at different places varies considerably because of differences in physiographic position and slope. The fertility of the soil also depends much on the geological formation of the rocks. Soils in the state are acidic in nature. Organic matter content of the soil varies between 2 and 5%. Major crops in the state are wheat, rice, maize, barley, buck wheat, cardamom, pulses, oil seeds, finger millets, etc.

Data used and methodology

The daily meteorological data (1985–2013) consisting of rainfall, temperature, relative humidity, wind speed and sunshine duration was collected from the observatory of India Meteorological Department located at ICAR-National Organic Farming Research Institute, east Sikkim. The observatory is located between 27 °19″13′ N latitude and 88 °36″10′ E longitude at a height of 1,322 m above MSL. Daily values of climatic parameters were used for trend analysis and computation of ETo. Temporal trends of major climatic parameter and ETo were analyzed for three seasons: summer (March–June), monsoon (July–October) and winter (November–February), on a seasonal and annual basis.

In this study, the method described by Mann (1945) for the non-parametric test of trend detection and the test statistic distribution given by Kendall (1975) for the testing of a nonlinear trend were adopted. The combined method is popularly known as the Mann–Kendall test. The change per unit time within time series was estimated by the non-parametric Sen's slope estimator (Sen 1968). Reference evapotranspiration (ETo) was estimated using the FAO's CROPWAT model.

Temporal trend analysis using Mann–Kendall test and Sen's slope estimator

The Mann–Kendall test is a non-parametric method used for trend analysis of time series data (Kendall 1975). The World Meteorological Organization (WMO) has recommended this method to assess the monotonic trend in hydro-meteorological time-series (Tian et al. 2012). In this study, the standard procedure described by Patle et al. (2013) and Pingale et al. (2016) was adopted for the Mann–Kendall test and Sen's slope estimator. The presence of a statistically significant trend is evaluated using the Zc value, where Zc test statistics and is given by Equation (1): 
formula
(1)
A positive value of Zc indicates an increasing trend and a negative value indicates a decreasing trend (Pingale et al. 2016). To test an increasing or decreasing monotone trend, a two-tailed test at α level of significance is used. Null hypothesis (Ho) is rejected if the absolute value of Zc is greater than Z1–α/2 where Z1–α/2 is obtained from the standard normal cumulative distribution tables. α is the significance level for the test and ±Z1−α/2 are the standard normal deviates. In this study, α and Z1−α/2 were taken as 5% and ±1.96, respectively (Patle et al. 2013). The magnitude of trend is predicted by the Sen's estimator. Slope is given by Equation (2): 
formula
(2)
where xj and xk are data values at times j and k (j > k) respectively. The median of these N values of Qi is represented as Sen's estimator. Qmed = Q(N+1)/2 if N is odd, and Qmed = [QN/2 + Q(N+2)/2]/2 if N is even (Patle et al. 2013). Positive value of Qi indicate an increasing trend and a negative value of Qi show a decreasing trend in the time series.

Reference crop evapotranspiration

In this study, the FAO Penman–Monteith (Allen et al. 1998) equation was used for the estimation of reference crop evapotranspiration. According to the Penman–Monteith combination equation, ETo can be expressed as: 
formula
(3)
where ETo is the reference evapotranspiration (mm day−1); RN is the net radiation at the crop surface (MJ m−2 day−1); G is the soil heat flux density (MJ m−2 day−1); T is mean daily temperature at 2 m height (°C); u2 is the wind speed at 2 m height (ms−1); es is the saturation vapor pressure (kPa); ea is the actual vapor pressure (kPa); es–ea is the saturation vapor pressure deficit (kPa); Δ is the slope of vapor pressure curve (kPa °C−1); γ is the psychometric constant (kPa °C−1) = 0.665 × 10−3P, where P is atmospheric pressure (kPa). P can be calculated by Equation (4): 
formula
(4)
Mean daily, monthly, seasonally and annual ETo for the period of 1985–2013 was estimated from mean values of climatic parameters using the FAO Penman–Monteith equation. The daily values of ETo were added to obtain the seasonal and annual values of ETo. Daily average long-term meteorological parameters of 29 years (1985–2013) were used as the reference point of study. Mean values of climatic parameters were referred to as baseline values. Sensitivity analysis was carried out by varying the magnitude of climatic parameters, namely temperature, wind speed, sunshine hours, and relative humidity one at a time by ±5, ±10, ±15 and ±20% over the baseline values of climatic parameters (Goyal 2004; Patle & Singh 2015).

Sensitivity coefficient

The sensitivity coefficient was calculated using Equation (5) which is a non-dimensional form of the sensitivity coefficient as suggested by Gong et al. (2006): 
formula
(5)
where SVI is the sensitivity coefficient for the ith variable and v is the climate variable analyzed. A positive or negative sensitivity coefficient indicates that ETo would increase or decrease with the increase or decrease of a climatic variable. The values of sensitivity coefficient (SVI) for a particular climatic parameter indicated the level of sensitivity of ETo calculation to variation in that parameter. The larger the absolute sensitivity coefficient, the larger the effect of a given variable on ETo.

The sensitivity coefficient was computed from the daily average values of 29 years of each climate variable. Thus, each sensitivity coefficient is a daily average value of 29 years for each climate parameter for percent increase or decrease in the respective climatic parameter, keeping the other climate parameters unchanged. For example, to determine the sensitivity coefficient for the maximum temperature (Tmax), the amount of increase or decrease in ETo at 5, 10, 15 and 20% increase or decrease in maximum temperature was calculated as the difference between the computed mean ETo and the new ETo values for each season (summer, monsoon and winter) and for the whole year. The difference between the computed ETo (with a 5% increase or decrease in base ETo) and base ETo was divided by the change in Tmax. Similarly, the sensitivity coefficient was determined for other climate parameters. The sensitivity coefficients on a seasonal and annual basis were determined for better understanding of the relative effect of each variable on ETo in different seasons.

RESULTS AND DISCUSSION

Trend analysis of major climatic parameters

Time series data of annual mean of daily climatic parameters, viz. temperature (minimum, maximum and average), relative humidity, sunshine hour, rainfall and reference evapotranspiration, were analyzed using the Mann–Kendall test and Sen's slope estimator for the east district of Sikkim. The findings of the Mann–Kendall statistics and Sen's slope estimated at 5% level of significance are presented in Table 1.

Table 1

Results of Mann–Kendall and Sen's slope estimator for meteorological parameters

Parameters Test period Mean SD Test Z Signific. Test Result 
Tmin 1985–2013 14.11 0.828 4.45 a 0.075 Increase 
Tmax 1985–2013 23.27 0.404 0.09  0.001 No trend 
Tmean 1985–2013 18.74 0.466 3.73 a 0.038 Increase 
RH 1985–2013 71.66 1.584 –0.39  –0.021 Decrease 
SSH 1985–2013 3.71 0.606 –5.53 a –0.070 Decrease 
Rainfall 1985–2013 3,084.5 300.18 0.43  2.987 Increase 
ETo 1985–2013 2.54 0.102 –3.66 a –0.008 Decrease 
Parameters Test period Mean SD Test Z Signific. Test Result 
Tmin 1985–2013 14.11 0.828 4.45 a 0.075 Increase 
Tmax 1985–2013 23.27 0.404 0.09  0.001 No trend 
Tmean 1985–2013 18.74 0.466 3.73 a 0.038 Increase 
RH 1985–2013 71.66 1.584 –0.39  –0.021 Decrease 
SSH 1985–2013 3.71 0.606 –5.53 a –0.070 Decrease 
Rainfall 1985–2013 3,084.5 300.18 0.43  2.987 Increase 
ETo 1985–2013 2.54 0.102 –3.66 a –0.008 Decrease 

aStatistically significant at 5% confidence level.

The value of Zc for minimum temperature (Tmin), maximum temperature (Tmax), average temperature (Tmean), relative humidity (RH), sunshine duration (SSH), rainfall and reference evapotranspiration (ETo) were 4.45, 0.09, 3.75, –0.39, –5.53, 0.43 and –3.66, respectively. The results of the Mann–Kendall analysis showed a significantly increasing trend for minimum temperature (Tmin) and average temperature (Tmean). Similarly, sunshine duration (SSH) and ETo showed a significantly decreasing trend. A non-significant decreasing trend was observed for relative humidity whereas rainfall showed a non-significant increasing trend over the period. The results of the Sen's slope estimator revealed that minimum temperatures increased by 0.075 °C/year and the mean temperature increased by 0.038 °C/year, whereas the maximum temperature increased by 0.001 °C/year over the period 1985–2013. Relative humidity decreased by 0.021% per year during 1985–2013. Results from the Sen's slope estimate also revealed that the sunshine hours and reference evapotranspiration decreased by 0.07 h/year and 0.008 mm/year respectively during 1985–2013. The negative sign signifies the decreasing trend in the slope and the positive sign signifies the increasing trend of slope.

Trend analysis revealed that the increase in minimum temperature and mean temperature may increase the CWR, particularly in the winter season which is mostly a dry period in the hill region of Sikkim and other hilly states of India. During this period, the scarcity of water exaggerates due to the drying of mountainous originating natural springs. An increasing trend in minimum temperature was also observed in Arunachal Pradesh and other parts of the northeastern hill states. The seasonal minimum temperature showed higher rates of warming in the post-monsoon season followed by monsoon and least in the pre-monsoon season whereas the mean annual rainfall showed an overall decreasing trend in the hill state of Arunachal Pradesh in India (Patle & Libang 2014). This may influence the seasonal CWR, especially for the vegetable and orchard crops grown in the region. In spite of the increase in mean temperature, annual ETo showed a decreasing trend in Sikkim. This is mainly due to the decreasing trend in sunshine duration and wind speed. Significant decreasing trends in ETo at annual and seasonal time scales was also reported by Jhajharia et al. (2012) for the different sites located in the northeast region of India. This indicates that the impact of an increase in temperature has been balanced by a steady decrease in wind speed and decline in net radiation which may be the reason for decreasing ETo over the humid region of northeast India (Jhajharia et al. 2012).

The effect of percent change in each meteorological parameter on percent change in ETo is presented in Table 2. The computed mean annual ETo was 920 mm for the east district of Sikkim. The estimated mean ETo in the summer, monsoon and winter seasons were 380, 337, and 203 mm, respectively.

Table 2

Change in seasonal and annual ETo due to change in meteorological parameter

Percent change in ETo due to percent change in each meteorological parameter
 
Season 5% 10% 15% 20% –5% –10% –15% –20% 
Minimum temperature (Tmin
 Summer 382 (0.5) 384 (1.1) 387 (1.8) 388 (2.0) 380 (0.0) 378 (0.5) 378 (0.5) 375 (1.5) 373 (1.8) 
 Monsoon 339 (0.5) 341 (1.1) 344 (2.1) 347 (2.8) 337 (0.0) 336 (0.5) 333 (1.2) 331 (1.8) 323 (2.3) 
 Winter 203 (0.0) 204 (0.6) 204 (0.6) 205 (1.2) 203 (0.0) 202 (0.6) 202 (0.6) 202 (0.6) 200 (1.2) 
 Annual 923 (0.4) 927 (0.8) 934 (1.6) 938 (2.0) 920 (0.0) 913 (0.8) 913 (0.8) 898 (2.4) 894 (2.9) 
Maximum temperature (Tmax
 Summer 387 (1.8) 394 (3.7) 401 (5.6) 409 (7.5) 380 (0.0) 373 (1.8) 366 (3.8) 360 (5.6) 354 (7.4) 
 Monsoon 343 (1.7) 351 (3.9) 357 (5.7) 364 (7.9) 337 (0.0) 332 (1.6) 325 (3.9) 320 (5.5) 315 (7.1) 
 Winter 206 (1.8) 210 (3.6) 215 (5.9) 218 (7.7) 203 (0.0) 199 (1.8) 196 (3.7) 192 (5.6) 188 (7.6) 
 Annual 934 (2.0) 953 (4.0) 971 (6.0) 989 (8.0) 916 (0.0) 902 (1.6) 883 (3.7) 869 (5.5) 854 (7.3) 
Relative humidity (RH) 
 Summer 378 (–0.5) 376 (–1.1) 373 (–1.8) 371 (–2.4) 380 (0.0) 383 (–0.8) 386 (–1.4) 388 (–2.0) 397 (–4.2) 
 Monsoon 335 (–0.8) 331 (–1.9) 332 (–1.5) 323 (–4.1) 337 (0.0) 339 (–0.6) 343 (–1.7) 347 (–2.7) 359 (–6.1) 
 Winter 204 (0.6) 204 (0.6) 199.2 (–1.8) 194.4 (–4.1) 202.8 (0.0) 205.2 (–1.2) 206.4 (–1.7) 208.8 (–2.9) 216 (–6.1) 
 Annual 916 (–0.4) 909 (–1.2) 902 (–2.0) 887 (–3.6) 920 (0.0) 927 (–0.8) 934 (–1.6) 942 (–2.3) 971 (–5.3) 
Wind speed (WS) 
 Summer 382 (0.5) 383 (0.8) 384 (1.1) 386 (1.4) 380 (0.0) 379 (0.2) 378 (0.5) 376 (1.1) 376 (1.1) 
 Monsoon 338 (0.3) 338 (0.3) 339 (0.6) 341 (1) 337 (0.0) 336 (0.5) 335 (0.8) 333 (1.2) 333 (1.2) 
 Winter 204 (0.6) 205.2 (1.2) 206.4 (1.8) 207.6 (2.4) 202.8 (0.0) 201.6 (0.6) 199.2 (1.8) 198 (2.4) 196.8 (3) 
 Annual 923 (0.4) 923 (0.4) 927 (0.79) 931 (1.19) 920 (0.0) 916 (0.4) 909 (1.2) 909 (1.2) 905 (1.61) 
Sunshine hours (SSH) 
 Summer 387 (1.8) 392 (3.0) 397 (4.3) 403 (5.9) 380 (0.0) 379 (0.2) 378 (0.5) 376 (1.1) 376 (1.1) 
 Monsoon 341 (1.0) 344 (2.1) 348 (3.2) 352 (4.3) 337 (0.0) 336 (0.5) 335 (0.8) 333 (1.2) 333 (1.2) 
 Winter 205 (1.2) 206 (1.8) 209 (3.0) 211 (4.1) 203 (0.0) 202 (0.6) 199 (1.8) 198 (2.4) 197 (3.0) 
 Annual 931 (1.2) 942 (2.4) 953 (3.6) 964 (4.8) 920 (0.0) 916 (0.4) 909 (1.2) 905 (1.6) 905 (1.6) 
Percent change in ETo due to percent change in each meteorological parameter
 
Season 5% 10% 15% 20% –5% –10% –15% –20% 
Minimum temperature (Tmin
 Summer 382 (0.5) 384 (1.1) 387 (1.8) 388 (2.0) 380 (0.0) 378 (0.5) 378 (0.5) 375 (1.5) 373 (1.8) 
 Monsoon 339 (0.5) 341 (1.1) 344 (2.1) 347 (2.8) 337 (0.0) 336 (0.5) 333 (1.2) 331 (1.8) 323 (2.3) 
 Winter 203 (0.0) 204 (0.6) 204 (0.6) 205 (1.2) 203 (0.0) 202 (0.6) 202 (0.6) 202 (0.6) 200 (1.2) 
 Annual 923 (0.4) 927 (0.8) 934 (1.6) 938 (2.0) 920 (0.0) 913 (0.8) 913 (0.8) 898 (2.4) 894 (2.9) 
Maximum temperature (Tmax
 Summer 387 (1.8) 394 (3.7) 401 (5.6) 409 (7.5) 380 (0.0) 373 (1.8) 366 (3.8) 360 (5.6) 354 (7.4) 
 Monsoon 343 (1.7) 351 (3.9) 357 (5.7) 364 (7.9) 337 (0.0) 332 (1.6) 325 (3.9) 320 (5.5) 315 (7.1) 
 Winter 206 (1.8) 210 (3.6) 215 (5.9) 218 (7.7) 203 (0.0) 199 (1.8) 196 (3.7) 192 (5.6) 188 (7.6) 
 Annual 934 (2.0) 953 (4.0) 971 (6.0) 989 (8.0) 916 (0.0) 902 (1.6) 883 (3.7) 869 (5.5) 854 (7.3) 
Relative humidity (RH) 
 Summer 378 (–0.5) 376 (–1.1) 373 (–1.8) 371 (–2.4) 380 (0.0) 383 (–0.8) 386 (–1.4) 388 (–2.0) 397 (–4.2) 
 Monsoon 335 (–0.8) 331 (–1.9) 332 (–1.5) 323 (–4.1) 337 (0.0) 339 (–0.6) 343 (–1.7) 347 (–2.7) 359 (–6.1) 
 Winter 204 (0.6) 204 (0.6) 199.2 (–1.8) 194.4 (–4.1) 202.8 (0.0) 205.2 (–1.2) 206.4 (–1.7) 208.8 (–2.9) 216 (–6.1) 
 Annual 916 (–0.4) 909 (–1.2) 902 (–2.0) 887 (–3.6) 920 (0.0) 927 (–0.8) 934 (–1.6) 942 (–2.3) 971 (–5.3) 
Wind speed (WS) 
 Summer 382 (0.5) 383 (0.8) 384 (1.1) 386 (1.4) 380 (0.0) 379 (0.2) 378 (0.5) 376 (1.1) 376 (1.1) 
 Monsoon 338 (0.3) 338 (0.3) 339 (0.6) 341 (1) 337 (0.0) 336 (0.5) 335 (0.8) 333 (1.2) 333 (1.2) 
 Winter 204 (0.6) 205.2 (1.2) 206.4 (1.8) 207.6 (2.4) 202.8 (0.0) 201.6 (0.6) 199.2 (1.8) 198 (2.4) 196.8 (3) 
 Annual 923 (0.4) 923 (0.4) 927 (0.79) 931 (1.19) 920 (0.0) 916 (0.4) 909 (1.2) 909 (1.2) 905 (1.61) 
Sunshine hours (SSH) 
 Summer 387 (1.8) 392 (3.0) 397 (4.3) 403 (5.9) 380 (0.0) 379 (0.2) 378 (0.5) 376 (1.1) 376 (1.1) 
 Monsoon 341 (1.0) 344 (2.1) 348 (3.2) 352 (4.3) 337 (0.0) 336 (0.5) 335 (0.8) 333 (1.2) 333 (1.2) 
 Winter 205 (1.2) 206 (1.8) 209 (3.0) 211 (4.1) 203 (0.0) 202 (0.6) 199 (1.8) 198 (2.4) 197 (3.0) 
 Annual 931 (1.2) 942 (2.4) 953 (3.6) 964 (4.8) 920 (0.0) 916 (0.4) 909 (1.2) 905 (1.6) 905 (1.6) 

Values in parentheses show the percent change in ETo.

An increase in mean minimum temperature (Tmin) by 20% increased the summer ETo by 2% (from 380 to 387 mm), whereas a 20% decrease in Tmin would reduce the ETo by 1.8% (from 380 to 373 mm) in the summer season. This indicates that an increase in temperature would have a greater effect on ETo than a decrease in Tmin. Variation in ETo due to change in mean minimum temperature was greater in the monsoon season (+2.8% for a 20% increase in temperature and –2.30% for a 20% decrease in temperature). This suggests that ETo is more sensitive to the mean minimum temperature in the monsoon season than in other seasons. In the winter season, the effect of an increase or decrease in Tmin on ETo was almost similar.

Annual ETo increased by about 2% (from 920 to 938 mm) for a 20% increase in Tmin, whereas ETo decreased by 2.9% for a decrease in Tmin by 20%. Similarly for an increase in Tmin by 10%, ETo would increase by 1.6% and for a 10% decrease in Tmin, ETo would decrease by 2.4%.

The sensitivity analysis revealed that the effect of change in maximum temperature (Tmax) on ETo was more prominent as compared to other parameters. A 20% increase in the value of Tmax would increase the seasonal and annual ETo by approximately 8% whereas a 20% decrease in the value of Tmax would decrease the ETo by about 7.4%. The percent change in relative humidity influenced ETo but the effect was less than the mean maximum temperature. The maximum change of ETo was about 4.1% in the monsoon and winter seasons, whereas a 20% decrease in relative humidity increased the ETo by 6.1%, respectively. Reference evapotranspiration was more sensitive to relative humidity during winter, followed by the monsoon and summer seasons. Annual ETo increased from 920 to 971 mm for a 20% decrease in relative humidity. However, if the relative humidity was increased by 20%, ETo decreased to 916 mm. ETo was more sensitive to wind speed in winter, followed by the summer and monsoon seasons. The annual ETo would increase by 1.18% for a 2% increase in wind speed. However, if wind speed was decreased by 20%, ETo would decrease by 1.61% in the study area. After maximum temperature, percent change in duration of sunshine hours affected the ETo most. The maximum change of 5.9% was observed in summer followed by monsoon (4.3%) and winter (4.1%). Annual ETo increased from 920 to 964 mm for a 20% increase in duration of sunshine hours. However, when the duration of sunshine hours is decreased by 20%, ETo decreased to 905 mm which is only 1.6% from the baseline ETo.

Analysis revealed that in all the seasons, ETo was more sensitive to mean maximum temperature followed by duration of sunshine hours. This suggests that the maximum temperature and duration of sunshine hours are major climatic parameters that control the ETo in the semi-humid region of Sikkim. However, change in wind speed, relative humidity and minimum temperature had a fluctuating effect on mean ETo. After the maximum temperature and sunshine hours, ETo was more sensitive to wind speed, followed by relative humidity and minimum temperature in summer. In the monsoon season, after maximum temperature and sunshine hours, ETo was more sensitive to minimum temperature followed by relative humidity and wind speed. However, in the winter season, after maximum temperature and sunshine hours, ETo was more sensitive to relative humidity followed by wind speed and minimum temperature. In the winter season, ETo was less sensitive to minimum temperature compared to other seasons.

The changes in ETo (mm/day) with respect to percent change in each climate variable are presented in the Figure 2(a)–2(d). Four separate lines are shown in each figure, which denote the change in ETo for percent increase or decrease in each climate variable annually and seasonally. It is observed that the ETo response was linear with respect to percent change in all climate parameters except for minimum temperature in the winter season.

Figure 2

(a–d) Percent change in ETo with respect to percent change in climatic parameter.

Figure 2

(a–d) Percent change in ETo with respect to percent change in climatic parameter.

Table 3 shows the seasonal and annual changes in sensitivity coefficient. Ratios of change in ETo (mm day–1) to change in climate variable were reflected as sensitivity coefficient. All sensitivity coefficients showed a large degree of fluctuations in all three seasons. The sensitivity of ETo to the same climate variables showed significant differences within the seasons. In general, maximum temperature (Tmax) had a large sensitivity coefficient throughout the seasons and year.

Table 3

Seasonal and annual sensitivity coefficient

Season + 5% + 10% + 15% + 20% –5% –10% –15% –20% 
TMIN 
 Summer 0.10 0.10 0.11 0.11 0.10 0.10 0.09 0.09 
 Monsoon 0.13 0.12 0.14 0.14 0.12 0.11 0.26 0.11 
 Winter 0.05 0.00 0.06 0.01 0.07 0.01 0.06 0.06 
 Annual 0.10 0.07 0.10 0.09 0.10 0.07 0.14 0.09 
TMAX 
 Summer 0.36 0.37 0.38 0.38 −0.36 0.35 0.35 0.34 
 Monsoon 0.37 0.38 0.39 0.39 −0.36 0.37 0.36 0.37 
 Winter 0.36 0.38 0.39 0.39 −0.37 0.36 0.35 0.35 
 Annual 0.36 0.38 0.38 0.39 −0.36 0.36 0.35 0.36 
RHMEAN 
 Summer –0.16 –0.14 –0.14 –0.15 –0.12 –0.13 –0.18 –0.22 
 Monsoon –0.19 –0.19 –0.19 –0.21 –0.16 –0.18 –0.07 –0.32 
 Winter –0.14 –0.18 –0.19 –0.20 –0.28 –0.21 –0.25 –0.36 
 Annual –0.16 –0.17 –0.18 –0.18 –0.19 –0.17 –0.17 –0.30 
WS 
 Summer 0.06 0.03 0.06 0.06 0.07 0.07 0.07 0.07 
 Monsoon 0.05 0.02 0.05 0.05 0.08 0.07 0.06 0.07 
 Winter 0.13 0.08 0.14 1.28 0.15 0.15 0.16 0.13 
 Annual 0.08 0.04 0.08 0.46 0.10 0.10 0.10 0.09 
SSH 
 Summer 0.25 0.28 0.26 0.29 0.27 0.29 0.27 0.21 
 Monsoon 0.24 0.24 0.23 0.22 0.23 0.22 0.23 0.22 
 Winter 0.24 0.26 0.25 0.25 0.25 0.26 0.25 0.22 
 Annual 0.24 0.26 0.25 0.25 0.25 0.26 0.25 0.22 
Season + 5% + 10% + 15% + 20% –5% –10% –15% –20% 
TMIN 
 Summer 0.10 0.10 0.11 0.11 0.10 0.10 0.09 0.09 
 Monsoon 0.13 0.12 0.14 0.14 0.12 0.11 0.26 0.11 
 Winter 0.05 0.00 0.06 0.01 0.07 0.01 0.06 0.06 
 Annual 0.10 0.07 0.10 0.09 0.10 0.07 0.14 0.09 
TMAX 
 Summer 0.36 0.37 0.38 0.38 −0.36 0.35 0.35 0.34 
 Monsoon 0.37 0.38 0.39 0.39 −0.36 0.37 0.36 0.37 
 Winter 0.36 0.38 0.39 0.39 −0.37 0.36 0.35 0.35 
 Annual 0.36 0.38 0.38 0.39 −0.36 0.36 0.35 0.36 
RHMEAN 
 Summer –0.16 –0.14 –0.14 –0.15 –0.12 –0.13 –0.18 –0.22 
 Monsoon –0.19 –0.19 –0.19 –0.21 –0.16 –0.18 –0.07 –0.32 
 Winter –0.14 –0.18 –0.19 –0.20 –0.28 –0.21 –0.25 –0.36 
 Annual –0.16 –0.17 –0.18 –0.18 –0.19 –0.17 –0.17 –0.30 
WS 
 Summer 0.06 0.03 0.06 0.06 0.07 0.07 0.07 0.07 
 Monsoon 0.05 0.02 0.05 0.05 0.08 0.07 0.06 0.07 
 Winter 0.13 0.08 0.14 1.28 0.15 0.15 0.16 0.13 
 Annual 0.08 0.04 0.08 0.46 0.10 0.10 0.10 0.09 
SSH 
 Summer 0.25 0.28 0.26 0.29 0.27 0.29 0.27 0.21 
 Monsoon 0.24 0.24 0.23 0.22 0.23 0.22 0.23 0.22 
 Winter 0.24 0.26 0.25 0.25 0.25 0.26 0.25 0.22 
 Annual 0.24 0.26 0.25 0.25 0.25 0.26 0.25 0.22 

Sensitivity coefficient of relative humidity was lower during the monsoon and higher during the winter season. The magnitude of the sensitivity coefficient was ranked as STmax > SSSH > STmin > SWS > SRH in summer season, STmax > SSSH > STmin > SWS > SRH in monsoon season, STmax > SSSH > SWS > STmin > SRH in winter, and STmax > SSSH > STmin > SWS > SRH annually, where S stands for sensitivity coefficient and Tmax is maximum temperature, SSH is sunshine hours, WS is wind speed, RH is relative humidity, and Tmin is minimum temperature. The sensitivity coefficient indicated that ETo changes positively with maximum and minimum temperature, sunshine hours, and wind speed while reversely with the relative humidity. A plot of the sensitivity coefficient versus climate variables for seasonal and annual time scales are shown in Figure 3(a)–3(d). The value of sensitivity coefficient was found to be negative for relative humidity and indicated that an increase in relative humidity would decrease the ETo potential.

Figure 3

(a–d) Plot of sensitivity coefficient versus climate variable.

Figure 3

(a–d) Plot of sensitivity coefficient versus climate variable.

Reference crop evapotranspiration (ETo) is an integrated effect of principal meteorological parameters and these parameters vary in a set pattern throughout the year. The crop water demand also varies season to season, therefore, to minimize the effect of variation of meteorological parameters, trend analysis and sensitivity analysis was carried out on a summer, monsoon and winter season and annual basis. The climatic variables included in this study showed either increasing or decreasing trends over the past 29 years. The influence of each variable on ETo was estimated using the sensitivity analysis method. Of the four climatic variables evaluated in this study, mean temperature was the primary factor driving changes in ETo over the past 29 years. Generally, the sensitivity of ETo to climatic variables varies among regions due to their different climatic characteristics. Tabari & Talaee (2014) also reported that sensitivity of ETo to wind speed and air temperature decreased from arid to humid climates, whereas its sensitivity to sunshine hours increased from arid to humid environments. In the arid region of northwest China, ETo was found to be most sensitive to wind speed, followed by relative humidity, temperature, and radiation (Hou et al. 2013). Goyal (2004) reported that in the arid zone of Rajasthan in India, evapotranspiration was more sensitive to temperature followed by wind speed, solar radiation, and vapor pressure, whereas for the semi-arid region of Karnal, Haryana, seasonal and annual ETo was most sensitive to maximum temperature followed by sunshine hours (Patle & Singh 2015). In our study area, which is situated at the foothills of the eastern Himalayan region, seasonal and annual ETo was found to be more sensitive to maximum temperature followed by duration of sunshine hours. The arid zone of Rajasthan is in the famous Thar Desert and therefore temperature and wind speed are relatively higher as compared to the eastern Himalayan Region. Vapor pressure, which represents relative humidity, had the least effect on ETo in Rajasthan in contrast to the eastern Himalayan region where wind speed and minimum temperature had the least effect.

CONCLUSIONS

In this paper, the impact of climate change on reference evapotranspiration (ETo) was observed by performing a trend analysis of long-term time series of meteorological parameters and sensitivity of ETo to major climatic parameters for the east district of Sikkim located in the eastern Himalayan region of India. Sensitivity of ETo was investigated in terms of change in temperature, wind speed, relative humidity and sunshine hours within a possible range of ±20% from the normal long term climatic variables and sensitivity coefficients were determined to evaluate the impact of principal climatic parameters on ETo. Trend analysis showed significant increasing trends for minimum temperature and mean daily temperature, whereas mean daily relative humidity and sunshine duration showed decreasing trends. Reference evapotranspiration also showed a decreasing trend. The results indicated that the seasonal and annual ETo were most sensitive to maximum temperature followed by sunshine hours for seasonal and annual ETo in the east district of Sikkim. However, wind speed, minimum temperature and relative humidity had a varying effect on mean ETo. Results revealed that a 20% increase in maximum temperature would increase the seasonal and annual ETo by about 8% whereas a 20% decrease in maximum temperature would decrease the ETo by about 7.4%.

The sensitivity coefficient indicated that ETo changes positively with maximum and minimum temperature, sunshine hours, and wind speed, while reversely with the relative humidity. Analysis indicated that an increase in relative humidity would decrease the ETo in the study area. The study suggests that the climate variability would affect reference evapotranspiration, however its impact on ETo would be different for different parameters. This study was carried out for the East district of Sikkim using the data of only one station. However, the findings of the study will be applicable in regions having a similar climate.

REFERENCES

REFERENCES
Admasu
W.
,
Tadesse
K.
,
Hordofa
T.
&
Habte
D.
2014
Sensitivity analysis of climatic parameters on reference evapotranspiration (ETo) calculated by FAO Penman–Monteith Equation (CROPWAT MODEL 8.0) for Sub-humid to humid areas of Arsi Zone, Ethiopia
.
Int. J. Agric. Sci. Res.
3
(
4
),
39
51
.
Allen
R. G.
,
Pereira
L. S.
,
Raes
D.
&
Smith
M.
1998
Crop Evapotranspiration – Guidelines for Computing Crop Water Requirements
.
Irri. Drain. Paper 56
.
UN-FAO
,
Rome
, pp.
1
300
.
Ambas
V. T. H.
&
Baltas
E.
2012
Sensitivity analysis of different evapotranspiration methods using a new sensitivity coefficient
.
Global NEST J.
14
(
3
),
335
343
.
Bandyopadhyay
A.
,
Bhadra
A.
,
Raghuwanshi
N. S.
&
Singh
R.
2009
Temporal trends in estimates of reference evapotranspiration over India
.
J. Hydrol. Eng.
4
(
5
),
508
515
.
Chiew
F. H. S.
,
Kamaladasa
N. N.
,
Malano
H. M.
&
McMahon
T. A.
1995
Penman–Monteith, FAO-24 reference crop evapotranspiration and class-A pan data in Australia
.
Agric. Water Manage.
28
,
9
21
.
Dejonge
K. C.
,
Ahmadi
M.
,
Ascough
J. C.
&
Kinzli
K. D.
2015
Sensitivity analysis of reference evapotranspiration to sensor accuracy
.
Comput. Electron. Agric.
110
,
176
186
.
Hamdi
M. R.
,
Bdour
A. N.
&
Tarawneh
Z. S.
2008
Developing reference crop evapotranspiration time series simulation model using Class a Pan: a case study for the Jordan Valley/Jordan
.
Jordan J. Earth Environ. Sci.
1
(
1
),
33
44
.
He
D.
,
Liu
Y.
,
Pan
Z.
,
An
P.
,
Wang
L.
,
Dong
Z.
,
Zhang
J.
,
Pan
X.
&
Zhao
P.
2013
Climate change and its effect on reference crop evapotranspiration in central and western Inner Mongolia during 1961–2009
.
Front. Earth Sci.
7
(
4
),
417
428
.
Irmak
S.
,
Payero
J. O.
,
Martin
D. L.
,
Irmak
A.
&
Howell
T. A.
2006
Sensitivity analysis and sensitivity coefficients of standardized daily ASCE Penman–Monteith equation
.
J. Irrig. Drain. Eng.
132
(
6
),
564
578
.
Jhajharia
D.
,
Dinpashoh
Y.
,
Kahya
E.
,
Singh
V. P.
&
Fard
A. F.
2012
Trends in reference evapotranspiration in the humid region of northeast India
.
Hydrol. Process.
26
,
421
435
.
Joshani
A.
,
Dastjerdi
J. K.
&
Abyaneh
H. Z.
2015
Evaluating different estimation methods of reference evapotranspiration with FAO evaporation pan in both dry and humid climate of Caspian Sea Basin
.
Bull. Environ. Pharmacol. Life Sci.
4
(
2
),
190
200
.
Kaboosi
K.
&
Kaveh
F.
2010
Sensitivity analysis of Doorenbos and Kassam (1979) crop water production function
.
Afr. J. Agric. Res.
5
(
17
),
2399
2417
.
Kendall
M. G.
1975
Rank Correlation Methods
,
4th edn
.
Charles Griffin
,
London
,
UK
.
Mahida
H. R.
&
Patel
V. N.
2015
Impact of climatological parameters on reference crop evapotranspiration using multiple linear regression analysis
.
SSRG Int. J. Civil Eng.
2
(
1
),
22
25
.
Maina
M. M.
,
Amin
M. S. M.
,
Rowshon
M. K.
,
Aimrun
W.
,
Samsuzana
A. A.
&
Yazid
M. A.
2014
Effects of crop evapotranspiration estimation techniques and weather parameters on rice crop water requirement
.
Aust. J. Crop Sci.
8
(
4
),
495
501
.
Mann
H. B.
1945
Non-parametric test against trend
.
Econometrica
13
,
245
259
.
Mosaedi
A.
,
Sough
M. G.
,
Sadeghi
S. H.
,
Mooshakhian
Y.
&
Bannay
M.
2016
Sensitivity analysis of monthly reference crop evapotranspiration trends in Iran: a qualitative approach
.
Theor. Appl. Climatol.
128
(
3–4
),
857
873
.
Patle
G. T.
,
Singh
D. K.
,
Sarangi
A.
,
Rai
A.
,
Khanna
M.
&
Sahoo
R. N.
2013
Temporal variability of climatic parameters and potential evapotranspiration
.
Indian J. Agric. Sci.
83
(
4
),
518
524
.
Pingale
S. M.
,
Khare
D.
,
Jat
M. K.
&
Adamowski
J.
2016
Trend analysis of climatic variables in an arid and semi-arid region of the Ajmer District, Rajasthan, India
.
J. Water Land Dev.
28
(
1–3
),
3
18
.
Porter
D.
,
Gowda
P.
,
Marek
T.
,
Howell
T.
,
Moorhead
J.
&
Irmak
S.
2012
Sensitivity of grass- and alfalfa-reference evapotranspiration to weather station sensor accuracy
.
Appl. Eng. Agric.
28
(
4
),
543
549
.
Saltelli
A.
,
Chan
K.
&
Scott
M.
2004
Sensitivity Analysis
.
John Wiley and Sons
,
New York
.
Schrier
G. D.
,
Jones
P. D.
&
Briffa
K. R.
2011
The sensitivity of the PDSI to the Thornthwaite and Penman-Monteith parameterizations for potential evapotranspiration
.
J. Geophys. Res.
116
,
1
16
.
Vicente-Serrano
S. M.
,
Azorin-Molina
C.
,
Sanchez-Lorenzo
A.
,
Revuelto
J. S.
,
an-Tejeda
E. M.
,
Opez-Moreno
J. I. L.
&
Espejo
F.
2014
Sensitivity of reference evapotranspiration to changes in meteorological parameters in Spain (1961–2011)
.
Water Resour. Res.
50
,
8458
8480
.
Wang
M.
,
Liu
H.
,
Zhang
L. W.
&
Zhang
R. H.
2014
Climate change trend and its effects on reference evapotranspiration at Linhe Station, Hetao Irrigation District
.
Water Sci. Eng.
7
(
3
),
250
266
.