Comparative Evaluation of WOFOST and CERES-rice Models in Simulating Yield of rice Cultivars at Navsari

A field experiment was conducted during kharif season of 2015 to assess the prediction performance of CERES-Rice and WOFOST modelfor grain and straw yield of three rice cultivars viz., (V1:Jaya, V2: Gurjari and V3: GNR-2) sown under four different environments viz., (D1: 10/07/2015, D2: 25/07/2015, D3: 09/08/2015 and D4: 24/08/2015) with two nitrogen levels N1:75 and N2:100 kg NPK/ha-1.Results showed that the prediction of WOFOST model for grain yield of rice cultivars under different treatments more close to the corresponding observed value with percent error PE between (18.66%)as camper to CERES-rice model with PE (28.56%), but for straw yield CERES-rice model give moreclose prediction than WOFOST model with PE (20.99%) and (27.33%) between predicted and observed value.


INTRODUCTION
Crop Simulation model is a computer model used to simulate reality it can be used to simulate crop growth, development and estimate yield of crop as a function of weather condition and management practices.Before using any model, it needs proper calibration and validation.Once model is properly validated, it can be used for various application viz., predicting growth, development and yield of crops, to study response of various climatic parameters and management practices on crops.The crop growth simulation model has been defined as a simplified representation of the physical, chemical and physiological mechanisms underlying plant and crop growth processes.Crop simulation models are able to provide an answer for "if then" type of question.(Mote and Kumar, 2016) reported that crop simulation model accurately simulates crop growth development and yield for different rice cultivars under different environment.Sometimes model simulations were under/overestimated yield and yield attributing parameters, but the trend follows good association between actual and simulated response.Now a day's various crop simulation models are used like CERES, WOFOST, WTGROWS, INFOCROP etc.Among these models in present study CERES-rice and WOFOST model were used for prediction of grain and straw yield of different rice cultivars under different environment and nitrogen levels.

Minimum data requirement
To generate the required experimental and crop management data, a field experiment was conducted at Agronomy farm of N. M. College of Agriculture, Navsari Agricultural University, Navsari (20° 57' N, 72° 54' E and 16 m above mean sea level), Gujarat, India, during rainy season of the year 2015.Daily data on weather parameters, like sunshine hours, rainfall, minimum temperature and maximum temperature, required for the model were recorded at weather station located in the vicinity of the experimental field.The soil profile data needed for the model was retrieved from the publication of earlier research work on characterization of Navsari

G2
Single dry grain weight (g) under ideal growing 0.024 0.025 0.028 conditions.i.e., non-limiting light, water, nutrients, and absence of pests and diseases.

G3
Tillering coefficient (scalar value) relative to cultivars 1 1 1 under ideal conditions.A higher tillering cultivar would have coefficient greater than 1.

G4
Temperature tolerance coefficient.Usually 1.0 for 740 710 700 cultivars grown in normal environment.G4 for japonica type rice grown in warmer environments would be ≥ 1.0.Tropical rice grown in cooler environments or season will have G4 < 1.0 soils (Jalalpur series).The initial soil characteristics were determined by collecting samples from the profile depth extending up to 45 cm depth and analyzing in the college laboratory.All records of inputs applied were maintained.Treatments of the field experiment included three rice cultivars (V 1 :Jaya, V 2 : Gurjari and V 3 : GNR-2) sown under four different environments viz., (D 1 : 10/07/2015, D 2 : 25/07/2015, D 3 : 09/08/2015 and D 4 : 24/08/2015) with two nitrogen levels N 1 :75 and N 2 :100 kg NPK/ ha-1.Both The model were validated with 2015 year experimental data and calibration with past years 2012, 2013 and 2014 data.The main aim was to find out which model give accurate prediction of rice cultivars under different environment and nitrogen levels.Both the model was run separately for each variety under different dates of transplanting and nitrogen levels.The calibration of both models are based on certain Genetic coefficients.Genetic coefficients of all three rice cultivars used in the both model safter carefully adjustment.However, the adjusted Genetic coefficients for all three rice cultivars of CERES-rice model given in (Table 1) and WOFOST model in (Table 2).The different test criteria viz.mean of observed and simulated values, root mean square error (RMSE), mean bias error (MBE) and mean percent error (PE) were used to evaluate the performance of model for simulation of yield and yield attributes characters of allthree rice cultivars.

Grain yield
Among the dates of transplanting grain yield was higher D 1 transplanted crop (41.39 q ha -1 ) with compared to rest of the transplanting dates, with delay of transplanting after D 1 (10h July) grain yield was decreases.(Mote, 2013) also reported that the grain yield is decreases due to delayed of date of transplanting.The difference between the D 1 and D 4 transplanting was (19.15 q ha -1 ) Table 3.In case of models, prediction ofgrain yield by WFOST model were reasonably close to the observed values overall PE in all date of transplanting varied between (-7.13 to -20.33%) as compared to CERES-rice model (-44.03 to +9.66%).In case of cultivars both models under predicted but WOFOST give better prediction as compared to CERES-rice model, overall variation between observed and predicted values by WFOST and CERES-rice models for all cultivars varied between (-5.10 to -17.07%) and (-12.80 to -20.11%).But in case of Nitrogen levels prediction of CERES-rice model was closer at N 1 level with PE (+9.66) followed by N 2 (-10.07)compared to WOFOST model (-10.75%) and -12.84%).different test criteria viz., RMSE, MBE, and PE between the predicted and corresponding observed values for grain yield also showed that the performance of WOFOST is more better than CERES-rice model with root mean square error (RMSE), mean bias error (MBE) and percent error (PE) of (6.19,-3.92and 18.66) and CERES-rice model (9.48, -4.93 and 28.56) respectively.

Straw yield
In different dates of transplanting straw yield was maximum in D 1 transplanted crop (83.85 q ha -1 ) with compared to rest transplanting of date.The difference between the D 1 and D 4 transplanting was (22.19 q ha -1 ) Table 3.In case of prediction of straw yield CERES-rice model was predicted more accurately with PE varied between (-1.43 to +32.67%) in different date of transplanting as compared to WOFOST model (-8.09 to +23.66%).Similarly in case of cultivars also CERES-ricemodel give better prediction with PE between (-3.66 to 3.45%) as compared to WOFOST model (-13.72 to-19.54%).Also in different nitrogen levels prediction

CONCLUSION
Prediction of grain and straw yield of rice cultivars under different treatment by CERES and WOFOST with more or less same percent error but prediction for grain yield was more accurately by WFOST model.In case of straw yield, CERES model give close prediction for all cultivars under different transplanting dates but under nitrogen levels accuracy of WFOST wasbetter.

Table 3 : Comparison between observed values and simulated values by CERES-Rice and WOFOST model for grain yield and straw yield of different rice cultivars under different environment Treatments Grain yield (q ha -1 ) Straw yield (q ha -1 )
rice model was closer at N 1 level with PE (+1.21%) followed by N 2 (-3.52%) compared to WOFOST model (-15.5%) and -18.22%).different test criteria viz., RMSE, MBE, and PE between the predicted and corresponding observed values for straw yield also showed that the performance of CERES-ricew as accurately than WOFOST model with root mean square error (RMSE), mean bias error (MBE) and percent error (PE) of (15.31,-0.99 and 20.99) and WOFOST model with (19.94,-12.50 and 27.33) respectively.