Determination of crop-coefficients and estimation of evapotranspiration of rapeseed using lysimeter and different reference evapotranspiration models ABHIJIT SARMA* and KRISHNA BHARADWAJ

Accurate estimation of evapotranspiration of rapeseed is essentially required for irrigation scheduling and water management. The present study was undertaken during 2015-16 and 2017-18 in ICR Farm, Assam Agricultural University, Jorhat to determine the crop coefficients (Kc) and estimate evapotranspiration of rapeseed using lysimeter and eight reference evapotranspiration models viz. Penman-Monteith, Advection-Aridity (Bruitsaert-Strickler), Granger-Gray, Makkink, Blaney-Criddle, Turc (1961), Hargreaves-Somani and Priestly-Tailor models. During 2015-16, the crop coefficients were developed by these models. Actual evapotranspiration was determined by three weighing type lysimeters. During 2017-18, evapotranspiration was estimated by multiplying reference evapotranspiration with Kc derived by different models and compared with actual evapotranspiration estimated by lysimeter during similar growing periods. All the models except Turc (1961) showed less than 10% deviation between actual and estimated ET. The estimated evapotranspiration using Penman-Monteith and Priestly-Tailor reference evapotranspiration recorded the lowest MAE and RMSE. The study revealed that estimated evapotranspiration using Penman-Monteith reference evapotranspiration gave the best estimate of evapotranspiration of rapeseed followed by Priestly-Tailor. The crop coefficients for initial, mid and end stages were 0.83, 1.20 and 0.65, respectively for Penman-Monteith and 0.70, 1.05 and 0.55, respectively for Priestly-Tailor.These results can be used for efficient management of irrigation water for rapeseed.

amount of water and correct timing of application is very essential for scheduling irrigations to meet the crop's water demands and for optimum crop production (Mehta and Pandey, 2016). On average, obtaining a better understanding of the actual crop water requirement based on modern technologies could save at least 50% of irrigation water (Ragab et al., 2017). Among the empirical models, the Food and Agricultural Organization hasrecommended the Penman-Monteith equation (FAO-PM) asa standard method for ET estimation (Allen et al.,1998). FAO-PM equation requires meteorological parameters such astemperature, humidity, wind speed, sunshine hours and net radiation to determine ET. Empirical models like Hargreaves-Somani, Turc,Blaney-Criddle etc., have also been used by several working as they require less number of meteorological parameters Phad et al., 2019). As such, it is required to develop Kc values for different models for the estimation of evapotranspiration. Based on the above, this experiment was undertaken in order to determine the crop coefficients (Kc) and estimate evapotranspiration of rapeseed using eight reference evapotranspiration models.

Location of experiment
The experiment was conducted at Instructional-cum-Research (ICR) Farm, Assam Agricultural University, Jorhat-13 during 2015-16 and 2017-18. The ICR Farm is situated at 26°47' N latitude, 94°12' E longitude and at an altitude of 87.0 metres above mean sea level. The climatic condition of Jorhat is subtropical humid with hot summer and cold winter. The average annual rainfall is 1864.8 mm. Out of this, 1194.8 mm, 467.1 mm, 151.4 mm and 51.5 mm are received during monsoon, pre-monsoon, post-monsoon and winter, respectively. The minimum monthly temperature of 9.7 °C and maximum monthly temperature of 32.4 °C are observed in January and August, respectively. During January and March, maximum (morning) and minimum (evening) monthly relative humidity of 94.8% and 61.1%, respectively are observed (Sarma and Das, 2017).

Measurement of actual evapotranspiration
The components of the water balance equations were measured by 3 weighing type lysimeters with dimensions of 1.3 m × 1.3 m × 0.9 m each. Each lysimeter was filled up with soil and rapeseed was grown. Fertilizers were applied as basal @ 60-40-40 kg N-P 2 O 5 -K 2 Oha -1 in the form of urea, SSP and MOP, respectively.The texture of the soil was sandy loam and acidic(pH 5.1) in nature.The field capacity of the soil was found to be 25.9% and the permanent wilting point was 8.92%. The rapeseed variety TS 38 was sown on 30 October, 2015 and 30 October 2017 during the first and second year, respectively maintaining a spacing of 30 cm between row to row and 5-7 cm between plant to plant. During 2016, the experiment could not be conducted as the lysimeters were used for another experiment. The crop was harvested on 31 January, 2016 and 31 January, 2018. The same variety of rapeseed crop was sown inside and outside the lysimeters to eliminate boundary effects. Fluctuations in weight of lysimeters were recorded at 8.30 a.m. everyday and daily loss of weight was replenished by irrigation. During 2015-16 and2017-18, 81.10 mm and 24.40 mm rainfalls were received.
Actual evapotranspiration of rapeseed was measured using the soil water balance equation. The water balance equation can be expressed as follows: ET = P+ (I-D)+ S Where, ET = Evapotranspiration; P = Precipitation; I =Irrigation water; D = Excess water drained from the bottom; S = Increase or decrease in the storage of soil moisture Change in soil moisture (S) is the difference in the moisture content of each consecutive days and it was calculated by deducting the moisture content of the day from the previous day starting from sowing to the last harvest. The drained outwater accumulated at the bottom tank of the lysimeter. This water was pumped out with the help of a pedal pump and the volume was measured. Dividing the volume by the area of the lysimeter, drainage depth of water was calculated.

Estimation of evapotranspiration (Est Etc)
Evapotranspiration of rapeseed under the climatic condition of Jorhat was estimatedby multiplying the calculated reference evapotranspiration of 2017-18 with Kcderived in 2015-16 by different models during the similar growing period.

Comparison of different models
During 2017-18, actual evapotranspiration (ETc) was measured using lysimeter and estimated evapotranspiration (Est ETc) for the same period using different models was compared with average error (AE), mean absolute error (MAE), mean bias error (MBE) and root mean square error (RMSE). These were calculated as follows: Where, n = number of observation ( ) L prev 

Crop coefficients of different models
The crop coefficients of rapeseed developed by lysimeter experiment during 2015-16 for initial (Kc ini), mid (Kc mid) and end period (Kc end) by different models using Fig. 1 are presented in Table 2. The Kc values integrate the effect of characteristics that distinguish a typical field crop from the grass reference, which has a constant appearance and a complete ground cover. The changing characteristics of the crop over the growing season have an effect on the Kc.In the study, the crop coefficients were lowest in the early crop growth period, gradually increased and reached a peak during 34-65 DAS and then decreased.At the last, the crop coefficient value decreased steadily due to maturity and senescence of leaf. Allen et al. (1998) also found that the crop coefficient depended on the type of crop, its stageof growth, canopy cover and crop density.
In all the models except Blaney-Criddle, Kc mid recorded the highest values indicating highest evapotranspiration during the mid-period of growth followed by Kc ini. The Kc mid values for Penman-Monteith, Granger-Gray and Makkink models were higher than 1.0 which indicated higher ETc than estimated ET0 under these models during the mid-period. The leaf area index, wind turbulence and leaf temperature are possible reasons for the increase in crop requirement above reference evapotranspiration (Kokilavani et al., 2018).The lowest value was recorded for Kc end. The developed Kc ini, Kc mid and Kc end values for Penman-Monteith were 0.83, 1.20 and 0.65. Granger-Gray model recorded higher Kc ini value than the Penman-Monteith model; however, Kc midvalue was equal to Penman-Monteith. Other models recorded lower Kc ini values than Penman-Monteith. Granger-Gray, Makkink and Blaney-Criddle models recorded slightly higher Kc end values. Kc end values for the rest of the models were smaller than the Penman-Monteith model.This variation is due to the differences in the estimation of reference evapotranspiration by different models. Tahashildar et al. (2017) also observed wide variations of reference evapotranspiration estimated by different empirical models inthe mid-hill region of Meghalaya.

Comparison of estimated evapotranspiration derived from different methods
Estimated evapotranspiration (Est ETc) of rapeseed during 2017-18 using the reference evapotranspiration and crop coefficients developed by different models during 2015-16 indicated slight average error from actual evapotranspiration (ETc) ( Table 3). The ETc of the crop during the initial, development, mid and late periods were 27.90 mm, 36.90 mm, 61.86 mm and 42.41 mm, respectively with a total of 169.13 mm.The total ET loss estimated throughthe Penman-Monteith method for rapeseed during the entire season of the crop was 168.86 mm with -0.16% average error.In case of other models, the sum total ET losses were found to be 159. 54 mm, 158.17 mm, 170.28 mm, 159.37 mm, 151.59 mm, 167.45 mm and 171.44 mmwith average error of -5.67%, -6.47%, 0.68%, -5.77%, -10.37%, -0.99% and 1.37% usingAdvection-Aridity (Bruitsaert-Strickler), Granger-Gray, Makkink, Blaney-Criddlel,Turc (1961), Hargreaves-Samani and Priestly-Tailor models, respectively.Thus all the models except Turc (1961) showed less than 10% deviation between actual and estimated ET. Penman-Monteith slightly underestimatedand Makkink and Priestly-Tailor models slightly overestimated the evapotranspiration. Contrary to it, Prajapati and Subbaiah (2019) found that adjusted FAO Kc overestimated evapotranspiration in Bt cotton in Junagadh. It suggested that local variability of meteorological conditions is important for estimation of reference evapotranspiration by different models. In the present study, the Penman-Monteith, Makkink and Priestly-Tailor models were also very consistent in different growth stages (except for developmental period) with average error within + 1.5%. Bhat et al. (2017) found that Makkink model fits best with the Penman-Monteith model and it was followed by Priestley-Taylor. Khavse et al. (2017) also found the Penman-Monteith model to be more appropriate as this method is rationalizing the weightage factor of different meteorological parameters.
The performance of the model was evaluated in termsof error analysis (Table 4). The ET estimated using Penman-Monteith and Priestly-Tailor reference evapotranspiration recorded the lowest MAE and RMSE indicating the lowest magnitude of average error. This can be attributed to the fact that the Penman-Monteith model takes into consideration both radiation as well as aerodynamic components in the estimation of evapotranspiration (Allen et al., 1998). Tomar (2016) found that the Priestley-Taylor method could estimate compatible ET0 values as estimated by the Penman-Monteith method.

CONCLUSION
The investigation was carried out to determine the crop coefficients (Kc) and estimate evapotranspiration of rapeseed using lysimeter and eight reference evapotranspiration models. The study revealed that estimated evapotranspiration using Penman-Monteith reference gave the best estimate of evapotranspiration of rapeseed followed by Priestly-Tailor. During the initial stage of the crops, the evapotranspiration was less and increased during the development stage, reached its maximum values during mid-season and reduced during crop maturation stages. The crop coefficients for initial, mid and end stages were 0.83, 1.20 and 0.65, respectively for Penman-Monteith and 0.70, 1.05 and 0.55, respectively for Priestly-Tailor.The information generated can be used in scheduling irrigation for rapeseedin Jorhat.