Evaluation of reference evapotranspiration models and determination of crop coefficient for Momordica charantia and Capsicum annuum

Agricultural Water Management 169 (2016) 77–89

Contents lists available at ScienceDirect

Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat

Evaluation of reference evapotranspiration models and determination of crop coefficient for Momordica charantia and Capsicum annuum Josilva M. Muniandy a , Zulkifli Yusop a,b,∗ , Muhamad Askari a a

Faculty of Civil Engineering, Universiti Teknologi Malaysia (UTM), Skudai, 81310 Johor, Malaysia Centre for Environmental Sustainability and Water Security, Research Institute for Sustainable Environment, Universiti Teknologi Malaysia (UTM), Skudai, 81310 Johor, Malaysia b

a r t i c l e

i n f o

Article history: Received 19 March 2015 Received in revised form 22 January 2016 Accepted 20 February 2016 Keywords: Penman Crop evapotranspiration Minilysimeter Momordica charantia Capsicum annuum

a b s t r a c t Studies on water balance and irrigation at agricultural areas require accurate values of reference evapotranspiration (ETo ). This study was conducted in the agricultural farm in the Modern Agriculture Centre in Kluang, Malaysia, to determine the crop coefficients of bittergourd (Momordica charantia) and chili (Capsicum annuum) by choosing the best ETo model. The experiment was conducted for two different crop cycles between October 2013 and May 2014. An automatic weather station was installed to record weather parameter at 30 min interval. Twenty six ETo models which were classified into four different groups were employed. The performance of the models was evaluated using Class A pan evaporation data from Kluang weather station. Eight statistical tests were used to assess and rank the accuracy of these 26 models. The ET values from the best ETo model of each group were then modeled with weather variables using multiple regression technique. Crop coefficient (Kc ) curves were developed as the ratio between actual crop evapotranspiration measured by minilysimeters and the ET values of the best model. The temperature based models tend to overestimate observed pan ET values, thus were not recommended at this site. Results of the mass transfer based Penman model show comparatively better ETo estimates among others. The total water requirement for bittergourd for the whole growing period is 153 mm while chili recorded 229 mm. The Kc values for the bittergourd were 0.58, 0.88 and 0.69 while for chili were 0.58, 0.95 and 0.73 for the initial, mid and end growth stages, respectively. The present results show very similar mid and end season chili Kc values with a study from Ghana while the mid and end season Kc of bittergourd is quite similar with a study in Florida, USA using cucumber. The values obtained can help farmers to determine the water requirement of these vegetable crops so that proper irrigation can be applied according to its growth stage and weather condition. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Bittergourd (Momordica charantia) and chili (Capsicum annuum) are some of the most popular vegetables that grow in tropical areas such as the Amazon, east Africa, Asia, and the Caribbean. Bittergourd also known as bitter melon or karella is a slender, climbing vine with long-stalked leaves and yellow. The fruit has a bitter taste and looks like a warty, oblong gourd that resembles a small cucumber. Chili is a spice crop that can be confused with other terminologies like pepper and capsicum. It is a small perennial shrub that can grow up to meter in height. It has a spicy hot taste unlike other vegetables and has been part of Asian and Mexican cuisine

∗ Corresponding author. E-mail address: [email protected] (Z. Yusop). http://dx.doi.org/10.1016/j.agwat.2016.02.019 0378-3774/© 2016 Elsevier B.V. All rights reserved.

for a very long time. Besides used for cooking, both of these plants have high nutrient values and medicinal properties, where bittergourd is used to cure diabetes, ulcer, gout and rheumatism, while chili is used as analgesic for arthritis pain, mastectomy pain and to cure fungal infection on skin (Crisan et al., 2009; Wang et al., 2014). The production of both crops has increased throughout the years. These crops use a large quantity of water, in which more than 90% of the water abstracted by roots is transpired back to the atmosphere causing increase in crop water requirement (Grace and Williams, 2004). All vegetable crops require adequate water management which strongly affect the plant growth and crop yield. Crop water use is a function of evaporation (E) and transpiration (T) that fluctuates daily. Allen et al. (1998) provides definition on evapotranspiration (ET) and reference evapotranspiration (ETo ). ET is defined as the sum of evaporation from water/soil surfaces and the amount of water transpired by plants. ETo is defined as evap-

78

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

otranspiration from an extensive surface of green grass of uniform height (0.08–0.15 m), an albedo of 0.23, fixed canopy resistance (70 s m−1 ), actively growing, completely shading the ground, and not short of water. There are many components that affect ETo , which include weather variables like solar radiation, air temperature, relative humidity (RH), and wind speed; crop factors such as type of vegetation, crop density and the growth stage; and other conditions such as soil type, salinity, fertility, cultivation level, crop disease, and pests (Allen et al., 1998). ET is one of the most difficult components to be determined in the water balance compared to other components like precipitation or irrigation (Fisher et al., 2005; Xu and Singh, 2005). Numerous models exist to estimate the ETo , but these models give inconsistent values due to their differences in modeling assumptions and input data requirements, or because the models have been developed for specific areas (Lu et al., 2005; Xu and Singh, 2005). FAO56 Penman-Monteith model is considered to be the best approach for estimating ETo and for the determination of crop coefficient because of its good approximation to lysimeter observations (Droogers and Allen, 2002; Popova et al., 2006; Xu and Singh, 2002). However, the FAO56 Penman-Monteith model requires many weather variables which can potentially introduce certain amounts of measurement and/or computational errors and cause cumulative errors in the calculated ETo (Meyer et al., 1989; Rahimikhoob et al., 2012). Due to this, other models that require less parameter should be considered for evaluation. Even though certain models such as Blaney-Criddle, Hargreaves, Makkink, Priestley Taylor and Turc, are developed under different weather variables, the models have been proven useful when applied at different climate regions (Federer et al., 1996; Kashyap and Panda, 2001; Trajkovic, 2007; Xu and Singh, 2001). Therefore, multiple ETo evaluation study for tropical regions is urgently required. The crop evapotranspiration (ETc ) is defined as the ET rate of crop under standard condition where there is no stress by water quality constraints, pests, or inadequate soil fertility (Allen et al., 1998). To determine the crop water requirement accurately, correct estimation of ETc is crucial as it provides the basis in determining water availability, crop water balance and crop water requirements (Pereira et al., 1999). Crop water requirement is defined as the depth of water required by plants to compensate the water loss via ET so plants are able to grow optimally while soil water balance shows amount of soil water added, removed or stored in volume of soil during a time period (Allen et al., 1998). Under-application of water can cause water deficit for plants and increase soil salinity, whereas over-application causes water wastage and nutrient leaching from the root zone (Merriam et al., 1999; Pereira et al., 2002). Thus, ETc can help farmers in order to decide when to start irrigation and how much water to apply to increase crop yields and profits while reducing costs, energy, and environmental impacts. Lysimeters are used to determine the ETc as actual ET depends on soil moisture. A lysimeter is basically a container that isolates soil from its surroundings but still replicates the adjoining soil, crop density and crop elevation as identical as possible. It provides a controlled soil-water environment for accurate measurement of water use and nutrient movement under defined bottom boundary condition (Dugas and Bland, 1989; Liu et al., 2002). The lysimeter can be categorized to either the weighing (measures ET directly by water mass balance) and non-weighing (measures ET indirectly by volume balance) type (Rana and Katerji, 2000). The usage of microlysimeter (non-weighing) is preferable since it does not require high level of expertise and special equipment compared to the high precision weighing lysimeter with continuous electronic data reading devices (to record water loss via ET and drainage) and other methods like Bowen ratio, remote sensing, surface energy balance, scintillometers and eddy covariance method (Allen et al.,

2011; Gebler et al., 2015; Rana and Katerji, 2000; Shuttleworth, 2008; Tomlinson, 1997; Wilson et al., 2001). Crop water use or water requirements is determined by multiplying ETo with crop coefficient, Kc . It is useful to determine the water requirement of crops according to their growth stage and environmental factors. The Kc value is sensitive and depends on several aspects such as type of crop, weather variables, canopy cover density, growth stage, soil moisture and agricultural operations (Allen et al., 1998). Previous studies have found that Kc for the same crop may vary from region to region depending on environmental factors such as climate and soil evaporation. Even though Allen et al. (1998) have compiled a list of Kc of various crops under different climates, Kc for a crop still has to be determined regionally as it may vary with factors like types of crop, growing stage, soil moisture, climate and agronomic techniques (Doorenbos and Pruitt, Doorenbos and Pruitt, 1977; Ko et al., 2009; Piccinni et al., 2009). In addition, some authors have reported differences between published and locally developed Kc (Kashyap and Panda, 2001; Tyagi et al., 2000). Due to this, more studies on determining different types of crop Kc at different climates should be conducted as it may help modelers and water resource engineers to provide more reliable water management schemes. Two approaches to determine crop coefficient are the single and dual crop coefficient approach. The dual crop coefficient approach divides the ET into E and T, whereas in the single crop coefficient approach, both E and T are combined into a single value (Allen et al., 1998). In the dual crop coefficient, the value of Kc is essentially composed of two terms: the basal coefficient (Kcb ) defined for a non-water-deficit condition with a “dry” soil surface; and Ke is a coefficient to account for soil or soil/crop surface evaporation from wetting by irrigation or precipitation (Allen et al., 1998). Developing Kc values involved determining the crop growing stages, lengths and Kc values for each stage. The Kc is different through the growing period due to differences in ET at various growth stages. According to the FAO methodology by Allen et al. (1998), the four growing stages of a crop are initial stage (Time from planting to approximately 10 percent of ground cover), crop development stage (from 10 percent of ground cover to effective full cover), mid-season stage (from full cover to beginning of maturity) and end-season stage (from start of maturity to end of harvest). Three Kc values are needed to create a Kc curve, which include the Kcini for the initial stage, Kcmid for the mid-season stage and lastly the Kcend for the end-season stage (Er-Raki et al., 2007; Mirzaei et al., 2011). The Kc value obtained from this experiment is compared to the values from the FAO publication that is adjusted to the local conditions/climate by these equations (Allen et al., 1998): K cini = f w K cini

(1)

where fw is the wetted perimeter. It is taken as 1.0 since the study site applies sprinkler irrigation. Kc mid = Kc mid + [0.04 (u2 − 2) − 0.004 (RHmin − 45)]

Kc end = Kc end + [0.04 (u2 − 2) − 0.004 (RHmin − 45)]

 h 0.3 3

 h 0.3 3

(2)

(3)

where RHmin is the mean minimum RH (20 < RHmin < 80%), h is the average crop height (0.1 < h < 10 m) and u2 is the mean daily wind speed at 2 m height (1 < u2 < 6m/s). The main objectives of this study are (i) to evaluate the performance of 26 ETo models by comparing with the pan evaporation data obtained from the nearest meteorological station and (ii) to determine the crop coefficients of bittergourd and chili crops.

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

79

Fig. 1. The study site location.

300

20

250

15 200 10

150

Rainy days

Rainfall (mm) Rainy days (day)

100 5

May

April

March

February

January

0

December

50

November

The experiment was conducted over eight months from August 2013 to May 2014 (227 days) at the Modern Agriculture Centre in Kluang located in the state of Johor, Peninsular Malaysia (1◦ 95 N, 103◦ 15 E, Fig. 1). The site elevation is about 20 m above mean sea level. Being close to the equatorial, the study site has a warm and humid climate. The weather conditions during the study period are presented in Figs. 2 and 3. Fig. 2 shows that the amount of rainfall during January and February is less than 2 mm. It is the lowest monthly rainfall during experimental period. Rainfall event in October is 270 mm which is the highest monthly rainfall. Higher amount of rainfall is recorded during October to December due to the effect of the North-East monsoon season (Suhaila et al., 2010; Wong et al., 2009). Fig. 3 shows that the mean monthly temperature ranges from 24 to 30 ◦ C throughout the study period. The higher solar radiation values were observed in October, February and March which contribute to higher air temperature due to greater efflux from ground surface. Due to earth revolution, the radiation are different at various latitudes and in different seasons. Monthly average solar radiation is 18 MJ/m2 (±2.14). This value is comparable with values ranging from 14.4 to 21.6 MJ/m2 for peninsular Malaysia (Sopian and Othman, 1992). Wind speed at the study site is classified as calm to light air based according to Beauford Wind Scale with monthly average value does not exceed 0.4 m/s (Oliver, 2005). The monthly RH tends to decrease with higher temperature.

October

2.1. Study area

The surface soil texture is sandy loam (Table 1). Soil samples were taken at every 10 cm depth using soil rings (height = 5 cm, diameter = 5 cm) at 3 different points at the study site. The soil particle distribution, initial soil water content ( i ) and hydraulic conductivity (Ksat ) were obtained from laboratory analysis using the hydrometer, oven dry method and constant head method respectively. Soil texture was determined using sand, silt and clay

Rinfall (mm)

2. Experimental method

Month Fig. 2. Monthly rainfall and number of rainy days at the study site.

80

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

Fig. 3. Monthly mean climatic patterns at the study site (Note: vertical bars represent standard deviations).

fractions according to USDA soil triangle (Shirazi and Boersma, 1984). Soil organic matter was determined using the loss on ignition method (Howard and Howard, 1990). Other parameters like ˛, n (empirical coefficient affecting the shape of the soil hydraulic function),  r (Residual water content) and ␪s (Saturated Water Content) were obtained using the optimization of Van Genuchten soil hydraulic function (Kool et al., 1987; Schaap et al., 2001; Van Genuchten, 1980). The crop planting was divided into two rotations that begin with bittergourd (October 2013–first week of January 2014) and followed by chilies (final week of January 2014–May 2014). Meteorological parameters were measured at the site using the ONSET RG2-M model automatic Rain Gauge and WatchDog 2550 model Weather Station to estimate daily rainfall and evapotranspiration. The combined sensors of air temperature and humidity were placed at 2 m above the ground (standard height). Parameters such as air temperature, humidity, wind speed and solar radiation were recorded every 30 min (mean value) and stored in a datalogger. Class A pan evaporation data from the nearest Meteorological station (Kluang station, 2◦ 01 N, 103◦ 19 E) which is about 18 km from the site was used as a standard value for comparison. It has been demonstrated that favourable results were achieved while using weather station data which is farther than 20 km (Camargo and Hubbard, 1999; Hashmi and Garcia, 1998).

2.2. Crop evapotranspiration determination Three lysimeters were filled with soil excavated from the study site. This was done carefully to resemble its original soil profile. The lysimeter, a cylindrical in shape, has an area of 550 cm2 and 60 cm depth for crop root development (Crisan et al., 2009). Since the lysimeter area is less than 1 m2 , it is considered as a minilysimeter (Dugas and Bland, 1989; Kong et al., 2012). The minilysimeters were located within crop planting area in a specific location to avoid disturbance to worker’s movement. It is also easy to install, maintain and not expensive to replicate. Daily crop evapotranspiration of each minilysimeter was calculated by determining the weight difference between two consecutive day readings using the following soil water balance equation: ETc = P + I − D − R − S

(4)

S = (W i+1 − W i )cf

(5)

cf =

water volume minilysimeter tank area

(6)

where P is the rain (mm), I is the irrigation depth (mm), D is the water loss through drainage from the lysimeter (mm), R is the runoff (mm), S is the change of water storage in the lysimeter (mm), Wi is the weight of the lysimeter on day I, Wi+1 is the weight of the lysimeter on the next day at a 24 hour interval and cf is the

Table 1 Soil physical properties at different soil depths at the study site. Soil depth(cm)

Sand (%)

Silt(%)

Clay (%)

Texture

SOM (%)

BD

i

(%)

0–10 10–28 28–70

53.1 51.8 40.5

27.7 26.9 35.2

19.3 21.3 24.3

Sandy loam Sandy clay loam Loam

6.6 (±0.1) 4.6 (±0.4) 5.6 (±0.1)

1.21 (±0.05) 1.53 (±0.09) 1.85 (±0.05)

35.2 (±3.3) 31.8 (±3.4) 30.6 (±2.4)

˛(cm−1 )

n

 r (cm3 /cm3

 s (cm3 /cm3

Ks(cm/h)

0.175 0.554 0.403

2.528 1.781 2.074

0.196 0.223 0.200

0.369 0.341 0.321

9.96 0.01 0.23

(Note: SOM = Soil organic matter, BD = bulk density, SWC = soil water content, ˛,n = empirical coefficient,  r = residual water content  s = saturated water content, Ks = hydraulic conductivity).

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

factor converting weight to equivalent depth of water (Igbadun, 2012). The average value of ETc was then used as daily ETc . ETopan was derived by multiplying the pan evaporation data with a coefficient (Kp ) value. Based on the lookup table developed by Doorenbos and Pruitt (1977), the appropriate Kp for the study site is 0.75.

2.3. Reference evapotranspiration The mass transfer methods estimate the ET by adopting the aerodynamic concept of water vapor movement from the evaporating surface to the air. All the models under this method are based on the Dalton’s gas Law. These models require parameters like air humidity and wind speed. The models under this family include the Rohwer 1936, Penman 1948 (Tabari et al., 2013), Meyer 1926, Brockamp & Wenner 1963, Albrecht 1950, Dalton 1802, WMO 1966 and Trabert 1896 (Bormann, 2011). The temperature methods are empirical models that use air temperature to represent the amount of energy available for evapotranspiration of crop. These methods provide poor result when compared to the combination methods as they do not consider parameters such as wind speed and solar radiation. This method can be very useful at places with limited weather data. Local calibration is required to improve the accuracy of ET determination since the models developed are site specific (Xu and Singh, 2002). Examples of the temperature based ET models are Hamon 1963 (Lu et al., 2005), Romanenko 1961 (Oudin et al., 2005), Szasz 1973 (Rácz et al., 2013), Hargreaves 1985 (Xu and Singh, 2002), Blaney Criddle 1977 (Niaghi et al., 2013), Linacre 1977 (Heydari et al., 2013) and Schendel 1967 (Bormann, 2011). The ET depends on the available energy and ability of water to be evaporated from land and vegetation surface. The transfer process is a function of wind speed and the amount of water vapor in the air closest to the surface. According to Priestley and Taylor (1972), the ET process is governed by net radiation, air temperature and pressure for a surface that extends over a large surface area. Radiation methods use solar radiation together with air temperature to determine ET. The solar radiation can also be used directly or indirectly to provide a measure of the net available radiation for ET estimation. Types of models that apply the radiation method include Makkink 1957, Turc 1962 (Lu et al., 2005), Priestley Taylor 1972 (Bormann, 2011), Abtew 1996 (Abtew, 1996), Rs (solar radiation) Based 2003, Rn (net radiation) Based 2003, McGuinness & Bordne 1972, Jones & Ritchie 1990 (Tabari et al., 2013), Caprio 1974 (Oudin et al., 2005) and Jensen Haise 1963 (Xystrakis and Matzarakis, 2010). The combination method consists of the radiation and aerodynamic components. This method requires more data than the temperature and radiation method, which includes net radiation (Rn ), air temperature (T), wind speed (U) and relative humidity (RH). This method gives the best result for a variety of vegetated surfaces and climates and it is applicable for locations where temperature, wind and radiation data are available. The FAO56 Penman Monteith 1998 (Rahimikhoob et al., 2012) is the most famous model that uses the combination method. In this study, 26 ETo models were evaluated without calibrating any model parameters which means that the original model were used without any changes on the coefficient/parameter. These models are Dalton, Trabert, Meyer, Rohwer, Penman, Albrecht, Makkink, Romanenko, Turc, Hamon, Jensen Haise, Brockamp & Werner, WMO, Schendel, Priestley Taylor, McGuinness & Bordne, Szasz, Caprio, Blaney Criddle, Linacre, Hargreaves Samani, Jones & Ritchie, Abtew, FAO56 Penman Monteith, Rs-Based and Rn-Based. The models are described in a chronological order and required weather data as follows (Tables 2 and 3).

81

2.4. Evaluation of model parameter The degree of the linear relationship between the observed and predicted model values can be evaluated from the coefficient of determination (R2 ) (Santhi et al., 2001). However, this approach is oversensitive to outliers and insensitive to additive and proportional differences between model predictions and observed data (Legates and McCabe, 1999). The d value is an enhancement of the R2 as it is more sensitive to the extreme values due to its squared differences (Legates and McCabe, 1999). A higher d value indicates a better agreement with the R2 value approach (Willmott, 1981). The MBE analysis describes the difference between the observations and the model predictions in the units of the variables divided with the number of observations. High RMSE value that exceeds the MBE is a good indicator of the presence and extent of outliers, or the variance of the differences between the modeled and observed values (Legates and McCabe, 1999). Therefore, the performance of the ET models was assessed using the following statistical approaches (Heydari et al., 2013; Legates and McCabe, 1999; Tabari et al., 2013; Ziegler et al., 2001): Sum Square Error (SSE) =



(Ot − Pt )

2

 Root Mean Square Error (RMSE) = Mean Bias Error (MBE) = Total Error (ETotal ) = Max Error (EMax ) = Min Error (EMin ) =

(7) E

(8)

N

(OTotal − PTotal ) N

(9)

PTotal − OTotal OTotal

(10)

PMax − OMax OMax

(11)

PMin − OMin OMin

(12)

N

Coefficient of Efficiency (E) = 1.0 −

t=1 (Ot

N  t=1

N Index of Agreement (d) = 1.0 −

N  t=1

− Pt )2

¯ Ot − O

t=1 (Ot

2

− Pt )2

(13)

2

¯ + |Ot − O| ¯ |Pt − O|

(14)

where E = error; P = predicted; O = observed, N = number of observation, Õ = mean observed value. The rank of ETo models within each group is determined by applying arithmetic mean to the combination of the above mentioned-statistical approaches. 3. Results and discussion The performance of each ETo model was evaluated by comparing its predicted values versus the pan evaporation derived evapotranspiration (ETopan ). Usually, statistics are made based on the Pearson correlation coefficient (r) but nowadays, models have become more complex as it includes other parameters. Therefore, other statistical measures like the goodness-of-fit test had been applied to describe the relationship between the actual and predicted ET. The total ETo values obtained for the entire experimental period are shown in Fig. 4. This figure shows that the Trabert, Meyer, Penman, Albrecht, Brockamp & Wenner and WMO models underestimated the observed pan value. The Schendel model overestimates the observed value highly compared to other models. 3.1. Mass transfer-based equation Table 4 summarizes the results of mass transfer-based models together with the observed ETopan values. The best results were

82

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

Table 2 ETo models and the required weather data. No

Model

Based

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

Dalton Trabert Meyer Rohwer Penman Albrecht Makkink Romanenko Turc Hamon Jensen Haise Brockamp & Wenner WMO Schendel Priestley Taylor McGuinness & Bordne Szasz Caprio Blaney Criddle Linacre Hargreaves Samani Jones & Ritchie Abtew FAO56 Penman Monteith Rn Based Rs Based

MT MT MT MT MT MT R T R T R MT MT R R T R R T T T R R C R R

Weather data R

T

Wind

Humidity

X X X X X X X X X X X X X X X X X X X X X X X X X

X X X X X

X X X X X

X

X X X

X X X

X X X X X X

X X

X

X X X

X

X

X

X X

X

X X X

(Note: MT = mass Transfer, T = temperature, R = radiation, C = combination).

obtained by the Penman equation, followed by the Rohwer and Dalton equations. From the MBE analysis, all the equations have underestimated the observed value except Meyer and Dalton. The Penman model recorded the lowest error (0.26 mm/day), followed by Rohwer (0.43 mm/day) and Dalton (0.72 mm/day). The highest MBE was recorded by Trabert model with 1.76 mm/day. The lowest SSE and RMSE were from the Penman model with 215 and 1.02 mm respectively. All models except Penman and Rohwer show E value less than zero, which is below the ‘minimum acceptable’ criteria. ETotal is the lowest in the Penman model while in EMax analysis, Rohwer model is the lowest.

1300

ET (mm)

From Table 5, the McGuinness & Bordne model is the best among the radiation based ET models. Unlike the mass transfer models, all the temperature based models overestimate the ET values compared to the observed ETopan values. The Makkink model recorded the lowest MBE while the Jensen Haise model recorded the highest MBE with 0.47 and 3.36 mm/day, respectively. From the EMax and EMin analyses, the Priestley Taylor method shows the lowest error value. Even though, Makkink model gave the lowest SSE, RMSE and MBE, it is ranked at the fifth place within this group as it performed poorly in other statistical parameters (ETotal , EMax , EMin , E and d).

Total ETo 3.3. Temperature-based equation

1200 1100 1000 900 800 700 600

Horizontal Line shows the observed (pan) value (587 mm)

500 400 300

The results of the statistical analysis for the temperature-based methods are given in Table 6. The Szasz method is the best option of the models to be applied at the study site. It has the lowest SSE and RMSE values. However, the MBE analysis shows that the Szasz model overestimates the pan ETopan values by 1.89 mm/day. The Schendel model shows the worst performance of ET estimation in this group.

3.4. Combination-based equation

Dalton Trabert Meyer Rohwer Penman Albrecht Makkink Romanenko Turc Hamon Jensen Haise B&W WMO Schendel PT McG & B Szasz Caprio BC Linacre HS J&R Abtew FAO56 Rs Based Rn Based

200 100 0

3.2. Radiation-based equation

ETo Models Fig. 4. Total ETo derived by various models compared with the ETopan (horizontal line) over eight months (October 2013– May 2014).

The FAO56 is the only combination based model used in this study as this model is widely used around the world. This model is the extension to the Penman 1948 model by incorporating the aerodynamic terms to the equation. FAO56 introduce empirical equations to replace the aerodynamic term of PM model. Therefore, the structure of FAO56-PM similarly related with Penman. This model gave a total SSE of 336 mm for the entire study period (Table 7). The RMSE and MBE values are 1.27 and 0.78 mm/day showing that the FAO56 overestimates the observed ET values.

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

83

Table 3 ETo model equations and its applications. No.

Model

Equation

Applications

Legend

1

Dalton (1802)

ETo = (0.3648 + 0.07223 (u)) (es − ea )

Bormann (2011), Tabari et al. (2013)

2

Trabert (1896)

ETo = (0.3075)

3

Meyer (1926)

ETo = (0.375 + 0.05026 (u)) (es − ea )

Bormann (2011), Tabari et al. (2013)

4

Rohwer (1936)

ETo = 0.44 (1 + 0.27 (u)) (es − ea )

Tabari et al. (2013)

5

Penman (1948)

ETo =

Tabari et al. (2013)

0.35 1 + 0.98/100u (es − ea )

6

Albrecht (1950)

ETo = (0.1005 + 0.297 (u)) (es − ea )

ea : actual vapor pressure (hPa); es : saturation vapor pressure (hPa); u: wind speed (m/s) ea : actual vapor pressure (hPa); es : saturation vapor pressure (hPa); u: wind speed (m/s) ea : actual vapor pressure (hPa); es : saturation vapor pressure (hPa); u: wind speed (m/s) ea : actual vapor pressure (mmHg); es : saturation vapor pressure (mmHg); u: wind speed (m/s) ea : actual vapor pressure (mmHg); es : saturation vapor pressure (mmHg); u: wind speed (miles/day) ea : actual vapor pressure (hPa), es : saturation vapor pressure (hPa); wind speed (m/s)

7

Makkink (1957)

ETo = 0.61

8

Romanenko (1961)

9

10

Turc (1962) Hamon (1963)

4.5 1 + ETo = 

0.013

Bormann (2011), Tabari et al. (2013)



ETo =



√ u (es − ea )



 +

 Rs 

 Tmean 2  25

Tmean Tmean +15



Bormann (2011), Tabari et al. (2013)

− 0.12

1−

ea es

Lu et al. (2005), Niaghi et al. (2013), Rahimikhoob et al. (2012)



Oudin et al. (2005)

Heydari et al. (2013), Lu et al. (2005), Xystrakis and Matzarakis (2010) Lu et al. (2005)

(Rs + 50)

ETo = 0.1651Ld RHOSAT × KPEC 216.7ESAT RHOSAT = Tmean +273.3 ESAT = 6.108exp



17.269 Tmean +273.3



Rs : solar radiation (MJ/m2 ); : slope of vapor pressure curve (KPa ◦ C-1); ␥: psychrometric constant (kPa ◦ C-1), : Latent heat of evaporation (MJ/kg) Tmean : mean temperature (◦ C); ea : actual vapor pressure (hPa), es : saturation vapor pressure (hPa) Tmean : mean temperature (◦ C); Rs : solar radiation(Cal/m2 day); Rs : solar radiation(Cal/m2 day) Ld: daytime length in multiples of 12 h; RHOSAT: saturated vapor density (g cm−3 ); Tmean : mean temperature (◦ C); ESAT = saturated vapor pressure (mbar), KPEC: calibration coefficient (1.2) Tmean : mean temperature (◦ C); Rs : solar radiation (MJ m−2 day−1 ); : latent heat of evaporation (MJ/kg)

Jensen Haise (1963)

ETRos =

Brockamp & Wenner (1963) WMO (1966)

ETo = 0.543 u0.456 (es − ea )

Bormann (2011), Tabari et al. (2013)

ETo = (0.1298 + 0.0934 (u)) (es − ea )

Bormann (2011), Tabari et al. (2013)

Schendel (1967)

ETo = 16

Bormann (2011)

Tmean : mean temperature (◦ C); RH: relative humidity (%)

15

Priestley Taylor (1972)

ETo = ˛

Bormann (2011), Niaghi et al. (2013), Rahimikhoob et al. (2012)

16

McGuinness & Bordne (1972)

(0.0082Tmean − 0.19)

: slope of vapor pressure curve (kPa ◦ C−1 ); Rn : net radiation (MJ/m2 ); G: soil heat flux density (MJ/m2 day), : latent heat of evaporation (MJ/kg), ˛: constant (1.26) Rs : solar radiation(Cal/m2 day). Tmean : mean temperature

11

12

13

14

17

Szasz (1973)

18

Caprio (1974)

19 20

Blaney Criddle (1977) Linacre (1977)



Xystrakis and Matzarakis (2010) (0.025Tmean + 0.08)





 Tmean 



RH (Rn ) (+)



ETo =



Rs 1500



Tabari et al. (2013) ×

2.54 ETo = 0.00536(Tmean + 21)2 (1 + RH)2/3 f (u)

u: wind speed (m/s); ea : actual vapor pressure (hPa), es : saturation vapor pressure (hPa); u: wind speed (m/s); ea : actual vapor pressure (hPa), es : saturation vapor pressure (hPa);

Rácz et al. (2013)

u: Wind Speed (m/s); f(u): Function of Wind Speed; RH: relative humidity (%) and T:mean temperature

Oudin et al. (2005)

Tmean : mean temperature (◦ C) and Rs : solar radiation(kJ/m2 )

ETo = p( 0.46Tmean) + 8

Niaghi et al. (2013)

ET o =

Heydari et al. (2013)

Tmean : mean temperature (◦ C) and p: constant (0.274) Z: local altitude (m); L: local latitude (degrees) and Td : dew point temperature (◦ C); Tmean : mean temperature (◦ C) Ra : extraterrestrial radiation (MJ/m2 .day); TD: maximum and minimum temperature difference( ◦ C) and Tmean : mean temperature (◦ C) Tmax : maximum temperature (◦ C); Tmin : minimum temperature (◦ C); ˛1 : constant (1.1)

f (u) =(0.0519u)  + 0.905 ETo =

6.1 106



Rs (1.8Tmean + 1.0)



700(Tmean ±0.006Z 100−L



+15(Tmean −Td)

80−Tmean

21

Hargreaves Samani (1985)

ETo = Ra TD0.5 (Tmean + 17.8) (0.0023 2.45

Heydari et al. (2013), Oudin et al. (2005), Rahimikhoob et al. (2012), Xu and Singh (2002); Xystrakis and Matzarakis (2010)

22

Jones & Richie (1990)

ETo=

Tabari et al. (2013)



˛1 ( 3.8710−3 (Rs (0.6Tmax + 0.4Tmin + 0.29))

84

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

Table 3 (Continued) No.

Model

Equation

Applications

Legend

23

Abtew (1996)

ETo = 0.53 Rs

(Abtew, 1996)

24

FAO56 (1998)

ETo =

Rs : solar radiation (MJ m−2 day−1 ); : latent heat of evaporation (MJ/kg) Rn : net radiation (MJ/m2 ); G: soil heat flux density (MJ/m2 .day); T: mean temperature ( ◦ C); u2 : wind speed at 2 m height (m/s); ␥: psychrometric constant (kPa/◦ C); : slope vapor pressure curve (kPa/◦ C);ea : actual vapor pressure (kPa) and es : saturation vapor pressure (kPa) Tmean : mean temperature (◦ C); Rs : solar radiation (MJ m−2 day−1 ) Tmean : mean temperature (◦ C); Rn : net radiation (MJ m−2 day−1 );

Heydari et al. (2013), Niaghi et al. (2013), Rahimikhoob et al. (2012), Tabari et al. (2013)

0.408(Rn −G)+ 900 (u2 )(es −ea ) T +273

+(1+0.34u2 )

25 26

ETo = −0.611 + 0.149Rs + 0.079Tmean ETo = 0.489 + 0.289Rn + 0.023Tmean

Rs Based (2003) Rn Based (2003)

Heydari et al. (2013), Niaghi et al. (2013), Tabari et al. (2013) Heydari et al. (2013), Niaghi et al. (2013), Tabari et al. (2013)

Table 4 Statistical performance of the mass transfer-based ET models versus the observed ET. Model

SSE

RMSE

Dalton Trabert Meyer Rohwer Penman Albrecht B&W WMO

320 847 342 242 215 562 494 565

1.24 2.02 1.29 1.08 1.02 1.65 1.54 1.65

MBE

ETotal

0.72 1.76 0.78 −0.43 −0.26 −1.34 −0.97 −1.34

0.28 0.67 0.30 0.16 0.10 0.51 0.37 0.51

EMax

EMin

0.09 0.53 0.09 0.16 0.39 0.44 0.13 0.60

0.84 0.82 0.88 0.69 0.29 0.41 0.61 0.29

E −0.29 −2.41 −0.38 0.02 0.13 −1.26 −0.99 −1.27

d

AM

Rank

0.51 −0.53 0.48 0.61 0.56 −0.08 0.47 −0.43

5.0 15.3 6.4 3.8 3.7 10.4 7.5 11.5

3 8 4 2 1 6 5 7

(Note: AM = Arithmetic Mean).

Table 5 Statistical performance of the radiation-based ET models versus the observed ET. Model Makkink Turc JH PT McG&B Caprio J&R Abtew Rn Based Rs Based

SSE 272 660 2336 747 400 2191 1051 598 872 765

RMSE

MBE

ETotal

1.15 1.79 3.36 1.90 1.39 3.25 2.25 1.70 2.05 1.92

0.47 1.42 3.07 1.57 0.88 2.96 1.90 1.29 1.78 1.63

17.95 0.54 1.17 0.60 0.34 1.13 0.72 0.49 0.68 0.62

EMax

EMin

E

d

AM

Rank

22.43 0.03 0.41 0.00 0.14 0.38 0.18 0.01 0.02 0.08

61.07 1.56 2.08 1.77 0.90 2.01 1.42 1.20 2.68 2.49

−0.10 −1.66 −8.41 −2.01 −0.61 −7.82 −3.23 −1.41 −2.51 −2.08

0.56 0.02 −1.13 −0.15 0.41 −1.03 −0.18 0.16 −0.64 −0.39

11.6 11.0 22.8 11.4 7.9 21.0 16.3 9.0 13.8 15.4

5 3 10 4 1 9 8 2 6 7

d

AM

Rank

−0.51 −3.71 −3.95 −0.53 −6.41 −2.91 −2.17

17.9 18.6 24.4 15.8 20.8 23.1 20.0

2 3 7 1 5 6 4

Table 6 Statistical performance of the temperature-based ET models versus the observed ET. Model

SSE

RMSE

MBE

ETotal

EMax

EMin

Roman Hamon Schendel Szasz BC Linacre HS

1238 1700 2346 947 2166 2309 1829

2.45 2.87 3.37 2.14 3.23 3.34 2.97

2.18 2.65 3.22 1.89 3.05 3.17 2.78

0.83 1.01 1.23 0.72 1.16 1.21 1.06

0.22 0.01 0.18 0.07 0.02 0.32 0.14

1.79 4.83 5.07 2.34 6.00 4.32 3.30

3.5. Overall Comparison of the group based ETo models The best model from each group was selected and ranked. These models are Penman (Mass Transfer), McGuinness & Bordne (Radiation), Szasz (Temperature) and the FAO56 Penman Monteith (Combination). From the arithmetic mean values analysis, it can

E −3.98 −5.85 −8.45 −2.81 −7.72 −8.30 −6.36

be observed that the Penman model shows the best performance among others (Fig. 5). McGuinness & Bordne model is a more preferred option for the application in the tropics compared to Penman or FAO56 model because the former uses less weather variables (Radiation and Temperature) but still provides satisfactory ETo estimate.

Table 7 Statistical performance of the combination-based ET model versus the observed ET. Model

SSE

RMSE

MBE

ETotal

EMax

EMin

E

d

AM

FAO56

336

1.27

0.78

0.30

0.17

1.18

−0.35

0.43

5.0

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

85

1000 800

2

SSE

RMSE

1.5 600 1

400

0.5

200 0

Penman FAO56 McG

Szasz

0 Penman FAO56 McG 0.4

ETotal 0.6

Szasz

EMax

0.3

0.4

0.2

0.2

0.1 Penman FAO56 McG

Penman FAO56 McG

Szasz

Szasz

0

EMin

E

2 -1

1

-2

Penman FAO56 McG

Szasz

-3 Penman FAO56 McG

Szasz

Fig. 5. Performance comparisons of the best models from each ET group.

Instead of the original equation, we have also propose an alternative Penman empirical equation as a function of radiation, temperature and RH using weather data of the present study site. ETo = 0.112T − 0.012Rs − 0.107RH + 7.666(R2 = 0.96,p = 0.00)(14) where T, Rs and RH indicate temperature, solar radiation and relative humidity respectively. As the p < 0.01, we concluded that the proposed equation is acceptable. The R2 value indicates that 96% of the variation of the predicted ETo Penman can be well explained by the proposed model. 3.6. Crop water requirement All the crops were grown on stalks reaching 1–2.5 m in height. Its effect on wind speed recording were minimized by locating the weather station at specific location which is 10 m apart from the plot. For comparison purpose, the Kc of cucumber was used since bittergourd is not found in the FAO publication by Allen et al. (1998). Cucumber Kc value is chosen since it belongs to the same Cucurbitaceae family with bittergourd (Crisan et al., 2009; Taylor, 2002).

For chili, sweet pepper (bell) was selected for comparison since chili does not appear in FAO publication since both are in the Solanacea family (Khan et al., 2012; Okada et al., 2011). The Kc values for the adjusted FAO were calculated based on the single crop coefficient approach method (Allen et al., 1998). The crop growing stage length is shown in Table 8. The Kc curve for both crops follows the classic Kc curve, where the value is small at the beginning of planting season and increases as the plant reaches maturity and decreases after harvesting begins (Fig. 6). In the present study, bittergourd has lower values of Kcini while chili has higher values of Kcini than the FAO56 adjusted values of cucumber and sweet pepper respectively (see Fig. 6). One possible reason is due to crop stress as more water has infiltrated into this sandy soil texture. Heavy rainfall in October decreases ET which in turn decreases the Kc values. Other factor that may affect Kc is soil salinity as reported by Ren et al. (2015). The high water tables and irrigation might maintain soil moisture in the root zone for crop growth, however it might cause high salinity level in some growth periods due to salt accumulation from fertilizer.

86

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

Table 8 Crop planted at the study site and its length of growing stage. Crop

Start date

Bittergourd (Momordica charantia) Chili (Capsicum annuum)

4 October 2013 27 January2014

End date

Growing stage (days)

16 January2014 30 May 2014

Total days (mm)

Initial

Development

Mid-season

End

20 (71a ;23b ) 20 (0a ;31b )

30 (359a ; 41b ) 35 (133a ;75b )

40 (226a ; 59b ) 45 (174a ;90b )

12 (1a ;30b ) 25 (108a ;33b )

102 (153) 125 (229)

(Note: irrigation date for Bittergourd:(12th October, 26th November, 26th December, 30th December, and Chili: 27th January, 30th January, 2nd February, 5th February, 8th February, 12th February, 17th February, 24th February, 1st March, 8th March 15th March). a Rainfall during growing stage. b ETc during growing stage.

During the initial stage for chili, the crops are no longer rain-fed since there is no rainfall recorded during January to February (see Fig. 2). Sprinkler irrigation which kept the soil wet most of the time had led to increase in ETc . During this stage, Kc decreases under high ETo values where more water is lost through soil evaporation since the crop canopy was not fully developed yet even under high soil water content. Table 8 shows that during the initial stage for chili, 31 mm (1.6 mm/day) of ETc was recorded. Later on, the Kc value increases linearly from the end of the crop initial stage until the start of the mid-growing stage where the Kc value reached its maximum. Overall, the Kc value increases in the development stage but surprisingly the Kc value for Bittergourd at 30 to 35 days of transplanting decreases. This might happen due to crop stress. During the mid-growing stage, this value remains constant. In this stage, the effect of soil surface evaporation reduced since more water was used for crop transpiration as the Leaf Area Index (LAI) of the crops increased (crop reached near or full ground cover). It is also possible that radiation was intercepted by the plants canopy due to high planting density that reduces the water consumption of the crop. Both Kcmid for bittergourd and chili Kcmid was lower than the FAO56 adjusted values (cucumber and sweet pepper). This could be due to different crop physical characteristics (height, LAI, fruit size) and climate conditions (higher ETo ) at the study site. During the end of growing season, Kc values of bittergourd were higher while Kc values of chili were lower than the adjusted FAO56 values (Fig. 6). The main reasons for the decrease in Kcend values are early drying of the crops (reduction of planting density and premature leaf ageing) and less irrigation. In addition, the LAI is also reduced due to leaf senescence that decreased the ET value (Karam et al., 2003; Karam et al., 2005). Kcend for bittergourd is higher since during this period (January), drought season has started and sprinkler irrigation has started causing less water loss by infiltration and

runoff. The differences in Kc values throughout the study period could be attributed to various factors, which include the tropical climate characteristics, crop planting operations and irrigation management. The percentages of difference in Kc values compared to the adjusted FAO values for bittergourd and chili ranged from 3% to 21% and from 3% to 17% respectively (Table 9). The highest difference between Kc and the adjusted FAO value occurred during the end-season stage. The difference between Kc from the present study and the values from Allen et al. (1998) suggests that there was difficulty in transferring the Kc value into different locations since the climate and agricultural management factors like irrigation method and irrigation frequency deviate from the adjusted Kc value, which is also claimed by Katerji and Rana (2014). Total crop water requirement for bittergourd and crop were found to be 153 mm and 229 mm respectively. Both crop recorded the highest crop water requirement at the mid-season stage and lowest at the initial stage. Chili may have recorded higher amount of cop water requirement due to the less rainfall recorded at the beginning stage of planting. The difference in crop water requirement of both crop may also be attributed to the differences in crop root structure, canopy development and crop height. Table 10 shows the summary of Kc value determination for vegetables from the Solanacea and Cucurbitae genus. The countries involved in this study are from the subtropical climate (USA, Cuba and Brazil) and warm humid climate (Ghana, South Africa and Nigeria). In most study, the ETo is estimated by the FAO56 equation except by the study of Owusu-Sekyere et al. (2010) that used the Class A pan method. In Ghana, Owusu-Sekyere et al. (2010) found a longer mid-season stage (56 days) for hot pepper crop in comparison with the present study (45 days) even though the present study have longer total growing days (125 days) compared to the former (121 days). Owusu-Sekyere et al. (2010) recorded the high-

1.2

1.2

Chili

1

1.0

0.8

0.8

7

0.6

0.6 0.4

0.4

8

Present Study Adjusted FAO (Allen et al., 1998) for cucumber Original FAO (Allen et al., 1998) for cucumber

0.2

0

Kc Value

Kc Value

Bittergourd

20

40

60

80

Days after Transplanting

100

Present Study Adjusted FAO (Allen et al., 1998) for sweet pepper Original FAO (Allen et al., 1998) for sweet pepper

0.2

0

20

40

60

80

100

Days after Transplanting

Fig. 6. Crop coefficient curves during growing period for bittergourd and chili.

120

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

87

Table 9 Kc obtained in this study and its comparison with the FAO and adjusted FAO values.

Kcini Kcmid Kcend

Kcini Kcmid Kcend

Kc (present study)

Adjusted FAO Kc (Allen et al., 1998)

FAO Kc (Allen et al., 1998)

Bittergourd 0.58 0.88 0.79

Cucumber 0.6 (3.3) 1.12 (21.4) 0.70 (1.4)

Cucumber 0.6 (3.3) 1.15 (23.5) 0.75 (8.0)

Chili

Sweet pepper

Sweet pepper

0.67 0.95 0.76

0.6 (3.3) 1.14 (16.7) 0.87 (16.1)

0.6 (3.3) 1.15 (17.4) 0.9 (18.9)

(Note: values in parentheses indicate the percentage of difference between the experimental and adjusted values).

est Kcmid value (1.42) for hot pepper when crop is applied with full water supply (100 percent crop water requirement). This shows that higher percentage of crop water requirement supplied leads to higher ETc , thus increasing the Kc value of crop in each growth stage. In addition, the height of the hot pepper in the Ghanaian study is shorter (35 cm) compared to the chili in the present study which can reach up to 1 m height. Table 10 also highlights the factors that contribute to Kc value determination which include the crop type, irrigation type, planting length and location. From the summary, it can be concluded that the Kc value can be different from one location to another due to different environmental conditions. The Kc developed from other regions is not applicable at other region and may lead to over irrigation (increased cost) or deficit irrigation (reduced yield). Difference in Kc values within the FAO56 and literature values may also differ due to the temporal scale (daily, monthly or seasonal) adopted in their study. In addition, the Kc values proposed by the FAO publication depends on the length of the four growing stage. Later studies have shown greater accuracy in computing the Kc as a function of LAI, canopy cover percentage expressed as growing degree days (GDD) that serves as an improvement to the FAO publication method (Lazzara and Rana, 2010). No studies have been found on bittergourd from literature thus cucumber is used for comparison. The study by Simonne et al. (2004) have 100 growing days (drip irrigation) while the present study recorded a 102 growing day (sprinkler irrigation). Kisekka et al. (2010) did not mention the total growing days of its cucumber crop. Simonne’s study have a close value with the present study Kc except it recorded much lower Kcini (0.25). Kisekka et al. (2010) shows a much higher Kcmid and Kcend than the present study but have lower Kcini (0.4). The influence of agronomic management factor like irrigation method can also be seen in crop water requirement determination.

Shukla et al. (2013) reported a high Kc value for bell pepper under seepage irrigation (from below the crop root zone and absorbed upwards) even though plastic mulching was applied. ET from crops with mulch should be lower since mulching is applied to reduce water loss due to runoff and evaporation. High soil moisture (near to saturation) causes higher soil evaporation and resulting in higher ETc especially in no mulching condition. Shukla et al. (2013) had also mentioned that pepper crop is highly dependent on the moisture content of soil affected by the irrigation method, rainfall and high water table. Interestingly, their Kc value did not decrease after the mid-season but increased further from 1.21 to 1.28. The Kcini of crops using sprinkler irrigation is much higher than the Kcini of crops using drip irrigation. In drip irrigation, only the root area is wetted while sprinkler irrigation wets an area in a circular shape. Since most crops have no leafs during this stage, the soil evaporation will increase, leading to higher Kcini . The effectiveness of drip irrigation is higher for crops since runoff and evaporation are higher in the sprinkler method. Comparison within the Solanacea genus shows that crops with longer growing season (Miranda et al., 2006) have higher Kc values regardless of where it is grown. This situation occurs as there is more than one harvest cycle. In the study involving hot pepper, Miranda et al. (2006) recorded the longest growing days (165 days) followed by Adams et al. (2014) with 130 days and present study (125 days). Miranda et al. (2006) applied drip irrigation while Adams et al. (2014) applied furrow irrigation. Durruthy et al. (2010) applied sprinkler irrigation to its sweet pepper crop and have 120 total growing days in their study conducted in Havana, Cuba. The Kc values from the current study shows almost identical value with the Kc value from Sam-Amoah et al. (2006) except for Kcini where the value is higher by 41 percent. Compared to the current study, most of the studies conducted showed a Kc mid value of higher than 1.0. This maybe the result of

Table 10 Crop coefficient (Kc ) values used for Cucurbitae and Solanacea genus crops in countries with tropical/humid climate. Genus

Crop

Kc Initial

Mid

End

Cucurbitae

Cucumber

0.4 0.25 0.58

1.05 0.9 0.88

Solanacea

Sweet pepper Bell pepper Green pepper Hot pepper

0.8 0.86 0.4 0.41 0.6 0.75 0.3 0.47 0.58

1.22 1.21 1.05 0.94 1.08 1.0 1.08 1.42 0.95

Bittergourd

a b

Applies plastic mulching as cover crop. Applies shrub as cover crop.

Study location

Irrigation

Total crop length (days)

Soil type

Source

0.9 0.7 0.79

Florida, USA Florida, USA Kluang, Malaysia

Drip Drip Sprinkler

– 100 102

– Sandy Sandy loam

Kisekka et al. (2010) Simonne et al. (2004) This study

0.62 1.28 0.9 0.74 0.96 0.89 0.65 0.9 0.76

Havana, Cuba Florida, USA Florida, USA Cape Coast, Ghana Tono, Ghana Kampe, Nigeria Parapaiba, Brazil Cape Coast, Ghana Kluang, Malaysia

Sprinkler Seepage Drip Drip Furrow – Drip Drip Sprinkler

120 100 – 118 130 97 165 121 125

Sandy loam Loamy sand-sandy – Sandyloam Sandy loam Loamy-sandy loam Sandy Sandy loam Sandy loam

Durruthy et al. (2010) Shukla et al. (2013)a Kisekka et al. (2010) Sam-Amoah et al. (2006) Adams et al. (2014) Adeniran et al. (2010) (Miranda et al., 2006) Owusu-Sekyere et al. (2010)b This study

88

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89

continuous irrigation pattern, water evaporation from crop canopy due to differences in crop roughness that lead to different interception capacity since hot pepper has many varieties in its species. 4. Conclusion The Penman (1948) from the mass transfer group was found to be the best model for estimating daily ETo of bittergourd and chili and followed by the FAO56, Makkink and Szasz from the combination, radiation and temperature group respectively. Among these 26 models, Jensen Haise, Linacre and Schendel ranked as the least accurate models as it tends to overestimate the observed ETopan value. All these models belong to the temperature-based group that tends to overestimate the ETo values. Therefore, these models may not be appropriate for the application in humid tropical site. Overall, it shows that mass transfer models have the best performance. The use of simple models like the Rohwer and Penman models helps to predict the ET at tropical areas. Instead of the original equation, we have also propose an alternative Penman empirical equation as a function of radiation, temperature and RH using weather data of the present study site. The total water requirement for bittergourd for the whole growing period is 153 mm while chili recorded 229 mm. The Kc values for the bittergourd were 0.58, 0.88 and 0.69 while for chili were 0.58, 0.95 and 0.73 for the initial, mid and end growth stages, respectively. Findings of this study can be applied for other humid tropical environments with similar irrigation scheduling, crop pattern selection and optimum crop water use. The Kc values obtained can help farmers to determine water requirement of these vegetable crops. Further studies should be conducted by calibrating the ET models using longer term data to improve its performance. Acknowledgements This work is part of the research project (J130000 7322 4B041) sponsored by the Ministry of Natural Resources and Environment through the Humid Tropic Centre, Kuala Lumpur (HTCKL). We thank the Research Management Centre of Universiti Teknologi Malaysia (UTM) for managing the project and the All Cosmos Sdn Bhd for their support to carry out the field experiment in their property. We are grateful for the MyBrain scholarship from the Ministry of Higher Education (MOHE) to the first author. We acknowledge Water Research Alliance, UTM for sponsoring first author to attend a 3-day publication workshop held in November 2014. The workshop facilitators helped in improving the earliest draft of the manuscript. This study is also partly supported by the Ministry of Higher Education, Malaysia and the Japanese Society for the Promotion of Science (JSPS) under the Asian Core Program. References Abtew, W., 1996. Evapotranspiration measurements and modeling for three wetland systems in south Florida. J. Am. Water Resour. Assoc. 32, 465–473. Adams, S., Quansah, G.W., Issaka, R.N., Asamoah, E.A., Nketia, K.A., Amfootu, R., 2014. Water requirements of some selected crops in Tono irrigation area. J. Biodivers. Environ. Sci. 4, 246–257. Adeniran, K., Amodu, M., Amodu, M., Adeniji, F., 2010. Water requirements of some selected crops in Kampe dam irrigation project. Australian. J. Agric. Eng. 1, 119. Allen, R.G., Pereira, L.S., Howell, T.A., Jensen, M.E., 2011. Evapotranspiration information reporting: I: factors governing measurement accuracy. Agric. Water Manage. 98, 899–920. Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration-Guidelines for computing crop water requirements-FAO Irrigation and drainage paper 56, FAO, Rome 300, 6541. Bormann, H., 2011. Sensitivity analysis of 18 different potential evapotranspiration models to observed climatic change at German climate stations. Clim. Change 104, 729–753.

Camargo, M.B.P., Hubbard, K.G., 1999. Spatial and temporal variability of daily weather variables in sub-humid and semi-arid areas of the United States high plains. Agric. For. Meteorol. 93, 141–148. Crisan, S., Campeanu, G., Halmagean, L., 2009. Study of Momordica Charantia L. species grown on the specific conditions of Romania’s western part. In: Zbornik Radova 44. Hrvatski i 4 Medunarodni Simpozij Agronoma, Opatija, Hrvatska, 16–20 Veljaˇce 2009. Poljoprivredni Fakultet Sveuˇciliˇsta Josipa Jurja Strossmayera u Osijeku, pp. 425–428. Doorenbos, J., Pruitt, W., 1977. Crop water requirements. In: FAO Irrigation and Drainage Paper 24. Land and Water Development Division, FAO, Rome. Droogers, P., Allen, R.G., 2002. Estimating reference evapotranspiration under inaccurate data conditions. Irrig. Drain. Syst. 16, 33–45. Dugas, W., Bland, W., 1989. The accuracy of evaporation measurements from small lysimeters. Agric. For. Meteorol. 46, 119–129. Durruthy, Y.C., Díaz, C.E.D., Seijas, T.L., Paredes, P., Pereiras, L.S., 2010. Determining of crop coefficients for horticultural crops in Cuba through field experiments and water balance simulation. Revista Ciencias Técnicas Agropecuarias 19, 50–56. Er-Raki, S., Chehbouni, A., Guemouria, N., Duchemin, B., Ezzahar, J., Hadria, R., 2007. Combining FAO-56 model and ground-based remote sensing to estimate water consumptions of wheat crops in a semi-arid region. Agric. Water Manage. 87, 41–54. Federer, C., Vörösmarty, C., Fekete, B., 1996. Intercomparison of methods for calculating potential evaporation in regional and global water balance models. Water Resour. Res. 32, 2315–2321. Fisher, J.B., DeBiase, T.A., Qi, Y., Xu, M., Goldstein, A.H., 2005. Evapotranspiration models compared on a Sierra Nevada forest ecosystem. Environ. Model. Softw. 20, 783–796. Gebler, S., Hendricks Franssen, H.-J., Pütz, T., Post, H., Schmidt, M., Vereecken, H., 2015. Actual evapotranspiration and precipitation measured by lysimeters: a comparison with eddy covariance and tipping bucket. Hydrol. Earth Syst. Sci. 19, 2145–2161. Grace, J., Williams, M., 2004. Understanding and parameterizing the soil–water–atmosphere transfer through vegetation. In: Feddes et al. (Org.) Unsatured-Zone Modeling: Progress, Challenges and Applications. Springer, Wageningen, Netherlands, 73–94. Hashmi, M.A., Garcia, L.A., 1998. Spatial and temporal errors in estimating regional evapotranspiration. J. Irrig. Drain. Eng—ASCE 124, 108–114. Heydari, M.M., Noushabadi, R.N., Vahedi, M., Abbasi, A., Heydari, M., 2013. Comparison of evapotranspiration models for estimating reference evapotranspiration in arid environment. Middle-East J. Sci. Res. 15, 1331–1337. Howard, P., Howard, D., 1990. Use of organic carbon and loss-on-ignition to estimate soil organic matter in different soil types and horizons. Biol. Fertil. Soils 9, 306–310. Igbadun, H.E., 2012. Estimation of crop water use of rain-fed maize and groundnut using mini-lysimeters. Pac. J. Sci. Technol. 13, 527–535. Karam, F., Breidy, J., Stephan, C., Rouphael, J., 2003. Evapotranspiration, yield and water use efficiency of drip irrigated corn in the Bekaa Valley of Lebanon. Agric. Water Manage. 63, 125–137. Karam, F., Masaad, R., Sfeir, T., Mounzer, O., Rouphael, Y., 2005. Evapotranspiration and seed yield of field grown soybean under deficit irrigation conditions. Agric. Water Manage. 75, 226–244. Kashyap, P.S., Panda, R., 2001. Evaluation of evapotranspiration estimation methods and development of crop-coefficients for potato crop in a sub-humid region. Agric. Water Manage. 50, 9–25. Katerji, N., Rana, G., 2014. FAO-56 methodology for determining water requirement of irrigated crops: critical examination of the concepts, alternative proposals and validation in Mediterranean region. Theor. Appl. Climatol. 116, 515–536. Khan, M., Hoque, M., Farooque, A., Habiba, U., Rahim, M., 2012. Physio-morphological features of chilli accessions under moisture stress conditions. Bangladesh J. Agric. Res. 37, 263–269. Kisekka, I., Migliaccio, K.W., Dukes, M.D., Crane, J.H., Schaffer, B., 2010. Evapotranspiration-based irrigation for agriculture: crop coefficients of some commercial crops in Florida. Ko, J., Piccinni, G., Marek, T., Howell, T., 2009. Determination of growth-stage-specific crop coefficients (Kc ) of cotton and wheat. Agric. Water Manage. 96, 1691–1697. Kong, Q., Li, G., Wang, Y., Huo, H., 2012. Bell pepper response to surface and subsurface drip irrigation under different fertigation levels. Irrig. Sci. 30, 233–245. Kool, J., Parker, J., Van Genuchten, M.T., 1987. Parameter estimation for unsaturated flow and transport models—a review. J. Hydrol. 91, 255–293. Lazzara, P., Rana, G., 2010. The use of crop coefficient approach to estimate actual evapotranspiration: a critical review for major crops under Mediterranean climate. Ital. J. Agrometeorol. 15, 25–39. Legates, D.R., McCabe, G.J., 1999. Evaluating the use of goodness-of-fit measures in hydrologic and hydroclimatic model validation. Water Resour. Res. 35, 233–241. Liu, C., Zhang, X., Zhang, Y., 2002. Determination of daily evaporation and evapotranspiration of winter wheat and maize by large-scale weighing lysimeter and micro-lysimeter. Agric. For. Meteorol. 111, 109–120. Lu, J., Sun, G., McNulty, S.G., Amatya, D.M., 2005. A comparison of six potential evapotranspiration methods for regional use in the southeatern United States. J. Am. Water Resour. Assoc. 41, 621–633.

J.M. Muniandy et al. / Agricultural Water Management 169 (2016) 77–89 Merriam, J.L., Burt, C.M., Clemmens, A., Solomon, K., Howell, T., Strelkoff, T., 1999. Irrigation performance measures: efficiency and uniformity. J. Irrig. Drain. Eng. 125, 97–99. Meyer, S.J., Hubbard, K.G., Wilhite, D.A., 1989. Estimating potential evapotranspiration: the effect of random and systematic errors. Agric. For. Meteorol. 46, 285–296. Miranda, F.R., Gondim, R.S., Costa, C.A.G., 2006. Evapotranspiration and crop coefficients for tabasco pepper (Capsicum frutescens L.). Agric. Water Manage. 82, 237–246. Mirzaei, M., Sohrabi, T., Jahanbania, H., Faghih, M., Shu, L.T., 2011. Evaluation of evapotranspiration coefficient and daily crop reference evapotranspiration in a semi-arid region based on field water balance and FAO method. Aust. J. Basic Appl. Sci., 5. Niaghi, A.R., Majnooni-Heris, A., Haghi, D.Z., Mahtabi, G., 2013. Evaluate several potential evapotranspiration methods for regional Use in Tabriz, Iran. J. Appl. Environ. Biol. Sci. 3, 31–41. Okada, R., Kiyota, E., Sabanadzovic, S., Moriyama, H., Fukuhara, T., Saha, P., Roossinck, M.J., Severin, A., Valverde, R.A., 2011. Bell pepper endornavirus: molecular and biological properties, and occurrence in the genus Capsicum. J. Gen. Virol. 92, 2664–2673. Oliver, J.E., 2005. The Encyclopedia of World Climatology. Springer Science & Business Media. Oudin, L., Hervieu, F., Michel, C., Perrin, C., Andréassian, V., Anctil, F., Loumagne, C., 2005. Which potential evapotranspiration input for a lumped rainfall–runoff model?: part 2—towards a simple and efficient potential evapotranspiration model for rainfall–runoff modelling. J. Hydrol. 303, 290–306. Owusu-Sekyere, J., Asante, P., Osei-Bonsu, P., 2010. Water requirement, deficit irrigation and crop coefficient of hot pepper (Capsicum frutescens) using irrigation interval of four (4) days. ARPN J. Agric. Biol. Sci. 5, 72–78. Pereira, L.S., Oweis, T., Zairi, A., 2002. Irrigation management under water scarcity. Agric. Water Manage. 57, 175–206. Pereira, L.S., Perrier, A., Allen, R.G., Alves, I., 1999. Evapotranspiration: concepts and future trends. J. Irrig. Drain. Eng. 125, 45–51. Piccinni, G., Ko, J., Marek, T., Howell, T., 2009. Determination of growth-stage-specific crop coefficients (Kc ) of maize and sorghum. Agric. Water Manage. 96, 1698–1704. Popova, Z., Kercheva, M., Pereira, L.S., 2006. Validation of the FAO methodology for computing ETo with limited data: Application to south Bulgaria. Irrig. Drain. 55, 201–215. Priestley, C., Taylor, R., 1972. On the assessment of surface heat flux and evaporation using large-scale parameters. Mon. Weather Rev. 100, 81–92. Rácz, C., Nagy, J., Dobos, A.C., 2013. Comparison of several methods for calculation of reference evapotranspiration. Acta Silvatica Lignaria Hungar. 9, 9–24. Rahimikhoob, A., Behbahani, M.R., Fakheri, J., 2012. An evaluation of four reference evapotranspiration models in a subtropical climate. Water Resour. Manage. 26, 2867–2881. Rana, G., Katerji, N., 2000. Measurement and estimation of actual evapotranspiration in the field under Mediterranean climate: a review. Eur. J. Agron. 13, 125–153. Ren, D., Xu, X., Hao, Y., Huang, G., 2015. Modeling and assessing field irrigation water use in a canal system of Hetao, upper Yellow River basin: application to maize, sunflower and watermelon. J. Hydrol. Sam-Amoah, L., Darko, R.O., Owusu-Sekyere, J., 2006. Water requirement, deficit irrigation and crop coefficient of hot pepper (Capsicum frutescens var legon18) using irrigation interval of two (2) days. Santhi, C., Arnold, J.G., Williams, J.R., Dugas, W.A., Srinivasan, R., Hauck, L.M., 2001. Validation of the SWAT Model on a Large River Basin with Point and Nonpoint Sources. Wiley Online Library.

89

Schaap, M.G., Leij, F.J., van Genuchten, M.T., 2001. ROSETTA: a computer program for estimating soil hydraulic parameters with hierarchical pedotransfer functions. J. Hydrol. 251, 163–176. Shirazi, M.A., Boersma, L., 1984. A unifying quantitative analysis of soil texture. Soil Sci. Soc. Am. J. 48, 142–147. Shukla, S., Jaber, F.H., Goswami, D., Srivastava, S., 2013. Evapotranspiration losses for pepper under plastic mulch and shallow water table conditions. Irrig. Sci. 31, 523–536. Shuttleworth, W.J., 2008. Evapotranspiration measurement methods. Southwest Hydrol. 7, 22–23. Simonne, E., Dukes, M., Haman, D., 2004. Principles and practices of irrigation management for vegetables. In: Vegetable Production Guide for Florida. University of Florida, Gainesville, FL, pp. 33–39. Sopian, K., Othman, M.Y.H., 1992. Estimates of monthly average daily global solar radiation in Malaysia. Renew. Energy 2, 319–325. Suhaila, J., Deni, S.M., Zin, W.Z.W., Jemain, A.A., 2010. Trends in peninsular Malaysia rainfall data during the southwest monsoon and northeast monsoon seasons: 1975–2004. Sains Malays. 39, 533–542. Tabari, H., Grismer, M.E., Trajkovic, S., 2013. Comparative analysis of 31 reference evapotranspiration methods under humid conditions. Irrig. Sci. 31, 107–117. Taylor, L., 2002. Technical data report for bitter melon (Momordica Charantia). Herbal Secrets of the Rainforest, 1–103. Tomlinson, S.A., 1997. Comparison of Bowen-ratio, Eddy-correlation, and Weighing-lysimeter Evapotranspiration for Two Sparse-canopy Sites in Eastern Washington. US Department of the Interio, US Geological Surveyr. Trajkovic, S., 2007. Hargreaves versus Penman-Monteith under humid conditions. J. Irrig. Drain. Eng. 133, 38–42. Tyagi, N.K., Sharma, D.K., Luthra, S.K., 2000. Evapotranspiration and crop coefficients of wheat and sorghum. J. Irrig. Drain. Eng.—ASCE 126, 215–222. Van Genuchten, M.T., 1980. A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Soc. Am. J. 44, 892–898. Wang, L., Waltenberger, B., Pferschy-Wenzig, E.M., Blunder, M., Liu, X., Malainer, C., Blazevic, T., Schwaiger, S., Rollinger, J.M., Heiss, E.H., Schuster, D., Kopp, B., Bauer, R., Stuppner, H., Dirsch, V.M., Atanasov, A.G., 2014. Natural product agonists of peroxisome proliferator-activated receptor gamma (PPAR-gamma): a review. Biochem. Pharmacol. 92, 73–89. Willmott, C.J., 1981. On the validation of models. Phys. Geogr. 2, 184–194. Wilson, K.B., Hanson, P.J., Mulholland, P.J., Baldocchi, D.D., Wullschleger, S.D., 2001. A comparison of methods for determining forest evapotranspiration and its components: sap-flow, soil water budget, eddy covariance and catchment water balance. Agric. For. Meteorol. 106, 153–168. Wong, C., Venneker, R., Uhlenbrook, S., Jamil, A., Zhou, Y., 2009. Variability of rainfall in peninsular Malaysia. Hydrol. Earth Syst. Sci. Discuss. 6, 5471–5503. Xu, C.-Y., Singh, V., 2002. Cross comparison of empirical equations for calculating potential evapotranspiration with data from Switzerland. Water Resour. Manage. 16, 197–219. Xu, C.-Y., Singh, V., 2005. Evaluation of three complementary relationship evapotranspiration models by water balance approach to estimate actual regional evapotranspiration in different climatic regions. J. Hydrol. 308, 105–121. Xu, C.Y., Singh, V.P., 2001. Evaluation and generalization of temperature-based methods for calculating evaporation. Hydrol. Process. 15, 305–319. Xystrakis, F., Matzarakis, A., 2010. Evaluation of 13 empirical reference potential evapotranspiration equations on the island of Crete in southern Greece. J. Irrig. Drain. Eng. 137, 211–222. Ziegler, A.D., Giambelluca, T.W., Sutherland, R.A., 2001. Erosion prediction on unpaved mountain roads in northern Thailand: validation of dynamic erodibility modelling using KINEROS2. Hydrol. Process. 15, 337–358.