ANNUAL PROGRESS REPORT

FOR THE YEAR2015 (Jan - Dec)

Barani Agricultural Research Institute, Chakwal, Pakistan

CRP 1.1: Dryland Systems

CRP Activity Title: Improving Water and Land Productivity in Rainfed Systems Using Community- based Water and Agronomic Management Approaches.

Sub-Activity Title: Decision making and planning in agriculture by calibrating, validating and scenario simulating the crop model (DSSAT) according to rainfed climate

1 | P a g e

Contents1. INTRODUCTION...................................................................................................................................4

1.1 Objectives of the study......................................................................................................................4

2. METHODOLOGY...................................................................................................................................5

2.1 Study Area.........................................................................................................................................5

2.2 Meteorological data..........................................................................................................................6

2.3 Soil properties....................................................................................................................................7

2.4 Model Calibration for wheat and barley............................................................................................9

3 RESULTS AND DISCUSSIONS................................................................................................................9

3.1 Calibration of CERES-Wheat model...................................................................................................9

3.1.1 Initial conditions.......................................................................................................................10

3.1.2 Fertilizer application.................................................................................................................11

3.1.3 Performance evaluation of DSSAT for wheat............................................................................12

3.2 Validation of DSSAT for wheat trial data..........................................................................................14

3.2.1 Crop data..................................................................................................................................14

3.2.2 Initial conditions.......................................................................................................................15

3.3 Calibration of DSSAT for barley under non-stress conditions..........................................................16

3.3.1 Initial conditions.......................................................................................................................17

3.3.2 Fertilizer application.................................................................................................................18

3.3.3 Performance evaluation of DSSAT calibration for barley..........................................................18

3.4 Validation of DSSAT for barley.........................................................................................................20

3.4.1 Initial conditions.......................................................................................................................20

3.5 Scenario simulation.........................................................................................................................21

3.5.1 Model application: determining optimum sowing dates and Supplement irrigation for wheat...........................................................................................................................................................22

3.5.2 Model application: determining optimum sowing dates and Supplement irrigation for Barley...........................................................................................................................................................23

4 CONCLUSIONS...................................................................................................................................24

5 RECOMMENDATIONS........................................................................................................................24

6 REFERENCES......................................................................................................................................25

2 | P a g e

3 | P a g e

1. INTRODUCTIONClimate Change is a rapidly unfolding challenge of catastrophic global, regional and national proportions. Pakistan will be affected by the impacts far more seriously than is generally recognized by the policy makers and opinion leaders. Pakistan has continuously witnessed history’s worst disasters since 2001. According to the IPCC’s fifth Assessment Report (AR5),global surface temperature increase exceed 1.5°C and keep rising beyond 2100 in all scenarios except the lowest-emission scenario, in which actions are taken to nearly eliminate CO2

emissions in the second half of the 21st century. In scenarios with higher rates of emissions, warming is likely to exceed 2°C by 2100 and could even exceed 4°C. The temperature increases in both summer and winter are reported higher in northern Pakistan than in southern. Despite the fact that Pakistan has witnessed a number of natural disasters in recent past, the need to study severity and impact of the natural disasters was felt after the devastating flood in 2010 (Yu, Winston et. al., 2013). The flood in 2010 had a devastating effect on the lives and livelihoods of millions in the country. The cost of recovery was estimated at USD8.74-10.85 billion (ADB, WB, and GOP, 2010).

Rainfed agriculture is extremely vulnerable to climate change. Climate change is both an opportunity and potential threat for future agriculture and livestock production globally and in Pakistan particularly. Pakistan is most likely one of the worst hit countries due to adverse effects of climate change that could have far reaching implications for agriculture, livestock production and in turn for human beings.

To measure the climate change effect and impacts it’s necessary to assess changes in climate and effective ness of adaptive measures to these changes. In scientific community, there is no other tool/practice to check the climate change effect on agriculture except crop models. Decision makers need a risk analysis to identify and prioritize effective adaption and mitigation strategies. Climatic variations, continuously increasing population pressure and market infrastructures are additional driving forces to adversely affect agricultural productivity. New management options and appropriate genotypes are need of the day to be considered for sustainable agri-food production.

Crop simulation models have the potential to help understand the impact of climate change and climatic variability and agronomic practices on plant growth and development as the models integrate the soil-water-plant-atmosphere complex relationship. The Decision Support System for Agro-technology Transfer (DSSAT) (Hoogenboom et al. 2004) is a comprehensive decision support system (DSS) for assessing management options. The cropping system simulation model (CSM) included in DSSAT v4.5 (Jones et al. 2003; Hoogenboom et al. 2004), is process-oriented, dynamic and simulates growth, development and yield for more than 25 different crops.

The model is parameterized using different agronomic parameters (phonological development, drymatter accumulation, leaf area index, physiological indices and grain yield) and climatic data. Efficiency of the model is tested using model validation skill scores including d-stat, RMSE and R2.

4 | P a g e

1.1 Objectives of the studyThe specific objective of the study was to develop crop models for wheat and barley for supporting in planning for improved resilience and sustainably stabilize yields and productivity under abnormal climatic conditions.  

2. METHODOLOGY2.1 Study AreaThe study was conducted at Barani Agricultural Research Institute (BARI) located within 72° longitude, 32° latitude and 575 m altitude in the district of Chakwal in Pakistan. The climate of Chakwal is semiarid subtropical and the annual rainfall varies from 500-1000 mm most of which falls during monsoon (Fig.1) season in the form of high intensity showers. The area also receives winter showers of lesser intensity during December to February.The map shows the location of the experimental area. Experimental site is located in Chakwal district and the total area of the district is 825578 hectares; 785,795 hectares is under cultivation and 39,783 hectares area is covered by the forests. About 8% of the total cultivated land is irrigated by canals, wells and tube wells (Government of Punjab 2013).About 70% of the population is engaged in farming or farm-related activities. Groundnut, barley and wheat are the main crops of the district. Sorghum, chickpea, canola, mustard, millet and gram are also grown by the farmers. Vegetables grown in the district include turnip, cauliflower, tomato, okra, onion and carrot. The main fruits are citrus and guava (Government of Punjab, 2009b). Percentage of major winter crops is shown in Figure. 2.

Figure 2: Winter cropping pattern for major crops in Chakwal. (Source: Akmal, M et al, 2014)

5 | P a g e

Figure 1: Location of study area with in Pakistan

In this study, two crops were selected - wheat and barley for modeling. Popular wheat and barley varieties of chakwal-50 and Joo-83 were used for model calibration, respectively. Previous three years (2010, 2012 and 2013) data for wheat and two years (2012 and 2013) data for barley were collected from the experimental site (BARI research station, Chakwal) including following parameters: planting date, emergence date, anthesis date, row spacing, plant height, planting method, plant population at seeding, plant population at emergence, planting depth, maturity date, harvesting date, and yield.

Crop simulation models are complementary tools to field trials. From this perspective, DSSAT (Decision Support System for Agro-technology Transfer) crop growth simulation model was calibrated and validated to predict alternative cropping pattern (shifting sowing date), growth and yield of wheat and barley under rainfed conditions for entire Pothwar region of Pakistan (Chakwal is within Pothwar region and is representative of the region).

The inputs required to run the DSSAT models include daily weather data (maximum and minimum air temperature, rainfall, and solar radiation); soil characterization data (physical, chemical, and morphological properties for each layer); a set of cultivar-specific coefficients characterizing the cultivar being grown in terms of plant development and grain biomass; and crop management information, such as plant population, row spacing, seeding depth, and application of fertilizer and irrigation. The soil-water balance is simulated to evaluate potential yield reduction caused by soil water deficits. The soil-water balance is determined on a daily basis as a function of precipitation, irrigation, transpiration, soil evaporation, runoff, and drainage from the bottom of the profile. The main components and module structure are shown in Figure3.

Figure 3: Overview of the components and modular structure of DSSAT v.4.5.0.0

2.2 Meteorological dataThe climate data, maximum and minimum temperature (°C), rainfall (mm) and solar radiation (MJ m-2 d-1) of last 31 years (1984-2015) were obtained from weather station of Soil and Water

6 | P a g e

Conservation Research Institute (SAWCRI), Chakwal. Data of the years (1984-2000) were in the raw form and contained missing data. The raw data was compiled in MS Excel and missing data gaps were filled by using climate data extrapolation software and were again converted to the format of DSSAT (Figure.4).

Figure 4: Climate data files and format in DSSAT

2.3 Soil propertiesFor the purpose of soil analysis, soil pit to depth of 100 cm was dug and the data of soil was measured as shown in Table 2.

7 | P a g e

Table 1: Soil properties of experimental site (BARI research station)

Soil depth (cm)

Drained lower limit

Drained upper limit

Saturation Root growth factor

Saturated hydraulic conductivity (cm/h)

Bulk density (g/cm3)

Organic carbon (%)

Clay (%)

Silt (%)

Nitrogen (%)

pH in water

18 0.06 0.2 0.42 1 0.75 1.52 0.45 6 16 0.04 9.161 0.065 0.18 0.39 1 0.6 1.7 0.35 14 8 0.02 9.198 0.07 0.17 0.35 1 0.8 1.6 0.2 6 20 0.02 8.9

151 0.06 0.17 0.38 1 0.83 1.39 0.02 8 22 0.02 8.9198 0.06 0.18 0.32 1 0.8 1.42 0.02 10 6 0.02 8.9

8 | P a g e

2.4 Model Calibration for wheat and barleyModel calibration is the adjustment of parameters so that simulated values compare well with observed data. The so-called 'genetic coefficients' that influence the occurrence of developmental stages in the DSSAT can be derived iteratively by manipulating the relevant coefficients to achieve the exactly match between the simulated and observed number of days to phenological events (Table 3).CERES-Wheat and CERES-Barley models were used for this study. For these CERES models, the plant life cycle is divided into several phases, which are similar among the two crops (Table 2).

Table 2: Growth stages simulated by DSSAT CERES

Wheat and Barley

Table 3: Cultivar coefficients for the DSSAT CERES-Wheat and CERES-Barley models

Cultivar coefficient

Definition

P1V Days, optimum vernalizing temperature, required for vernalizationP1D Photoperiod sensitivity coefficient (% reduction in rate/ h near threshold)P5 Grain filling (excluding lag) period duration(°C d)G1 Kernel number per unit canopy weight at anthesis (#/g)G2 Standard kernel size under optimum conditions (mg)G3 Standard , non-stressed mature tiller weight (including grain) (g dwt)PHINT Thermal time between the appearance of leaf tips (°C d)

3 RESULTS AND DISCUSSIONS3.1 Calibration of CERES-Wheat modelWheat variety of Chakwal-50 was selected for model calibration, because it is a high yielding, drought tolerant and disease resistant popular wheat cultivar for rainfed areas of Punjab, Pakistan. First of all,cultivar coefficients were adjusted for the variety chakwal-50 as shown in Table 4.

Table 4: Cultivar coefficients fitted for chakwal-50wheat variety

9 | P a g e

Cultivar coefficientsP1V 40P1D 55P5 550G1 24G2 26G3 1.3PHINT 70

Although, GENCALC software (Hunt andPararajasingham, 1994) does this type of adjustment automatically and, therefore, uses the observations of phenological events from one or several experiments from a range of environments, we chose the manual approach because there were relatively few experimental data percultivar, impeding the identification of optimal parameter values by such a mathematical algorithm. Godwin et al. (1989) suggestthat such a manual, iterative approach usually reaches reasonable estimates of the genetic coefficients.

The DSSAT model was calibrated by using the crop data of 2010, 2012 and 2013 (Table 5).

Table 5: Wheat crop data used for DSSAT CERES-Wheat model calibration

Parameter Years2010-2011 2012-13 2013-14

Planting date 10-Nov-2010 01-Nov-2012 05-Nov-2013Emergence date 16-Nov-2010 09-Nov-2012 12-Nov-2013Planting depth (cm) 10 10 10Planting method dry seed Dry seed Dry seedPlant population at seeding (plants/m2) 400 400 400

Plant population at emergence (plants/m2) 200 200 200

Row spacing (cm) 22.5 22.5 22.5Anthesis date 15-Mar-2011 11-Mar-2013 14-Mar-2014Plant height (cm) 97.8Maturity date 19-Apr-2011 10-Apr-2013 14-Apr-2014Harvesting date 30-Apr-2011 24-Apr-2013 28-Nov-2014Yield (kg/ha) 4209 4329 4046

3.1.1 Initial conditionsThe most important part of setting up a crop model is the initial soil moisture condition. It shows the level or conditions of moisture contents present in the soil before sowing, because by these moisture contents the model can predict yield precisely. Initial moisture contents for wheat for three years (2010, 2012 and 2013) are shown in Table 6.

10 | P a g e

Table 6: Initial soil moisture conditions

Measurement date Sampling depth Moisture content

(cm3/cm3)

09-11-2010

15 0.1035 0.1055 0.1075 0.1095 0.11115 0.12

30-10-2012

15 0.1435 0.1355 0.1275 0.1395 0.12115 0.13

04-11-2013

15 0.1535 0.1455 0.1475 0.1595 0.13115 0.13

3.1.2 Fertilizer applicationTable 7: Fertilizer application in wheat

Date of application Fertilizer type

Depth (cm)

N (kg/ha)

P (kg/ha)

K(kg/ha)

10-Nov-2010 Urea 20.00 120 100 601-Nov-2012 Urea 20.00 120 100 605-Nov-2013 Urea 20.00 120 100 60

The above mentioned data was entered in the DSSAT crop management data filecalled X-file, then we made the A-file for the DSSAT.

Table 8: A-file parameter for wheat

Year ADATMDAT

HWAM

2010 11074 11109 4209

11 | P a g e

2012 13070 13100 43292013 14070 14104 4046

ADAT = Anthesis date (YrDoY)MDAT = Physiological maturity data (YrDoY)HWAM = Yield at harvest maturity (kg (dry matter)/ha)

3.1.3 Performance evaluation of DSSAT for wheatThe performance of the DSSAT model after calibrationfor wheat trial data was evaluated. The variables tested included the key phenological dates (anthesis and harvest maturity) and the final yield. In general, the model gave good predictions of crop development and final grain yields, and for the cultivar Chakwal-50 (Table 9).For calibration, the simulated dates of anthesis and physiological maturity as well as yield were compared with the observed data (Table9).

Different statistical indices were employed, including Coefficient of Determination(r-square), absolute and normalized Root Mean Square Error (RMSE and NRMSE) and index of agreement (d-index). Normalized RMSE gives a measure (%) of the relative difference of simulated versus observed data. The simulation is considered excellent with a normalized RMSE is less than 10%, good if the normalized RMSE is greater than 10% and less than 20%, fair if normalized RMSE is greater than 20 and less than 30%, and poor if the normalized RMSE is greater than 30% (Jamieson et al.,1991). The index of agreement (d) proposed by Willmott et al.(1985), according to the d-statistic, the closer the index value is to one, the better the agreement between the two variables that are being compared and vice versa.

The good thing about DSSAT is that it performs the statistical analysis of theobserved and simulated values of the experiments in its Evaluate.outoutput file.

Table 9: Statistical indices of evaluating the performance of CERES-Wheat model in predicting phenological dates and simulating grain yield.

Cropping years

Anthesis (DAP) Physiological Maturity (DAP)

Grain yield (kg/ha)

Simulated

Measured

Simulated Measured Simulated

Measured

2010-2011 127 125 163 160 4478 42092012-13 130 128 167 165 4445 43292013-14 136 132 173 170 4112 4046

IndexRMSE (day)a 3 3 173.0NRMSE (%)b 2.0 2.0 4.0d-statc 0.84 0.90 0.68r-squared 0.98 0.98 0.75

aRMSE Rootmean square error

12 | P a g e

bNRMSE Normalized root mean square errorcd-stat Wilmot's index of agreementdr-square Coefficient of determination

The model was able to predict the anthesis date well as shown in Table 9, and Figure. 5. The values for RMSE, normalized RMSE, index of agreement (d) and r2 for anthesis date were3 d (days), 2%, 0.84 and 0.98, respectively. There was, also, a good match between predicted and observed physiological maturity dates. The values for RMSE, normalized RMSE, index of agreement (d) and r2for physiological maturity date were 3 d (days), 2%, 0.90 and 0.98, respectively.

126 128 130 132 134 136 138125

130

135

140

R² = 0.988416988416988

Simulated Data

Obs

erve

d D

ata

(a)

162 164 166 168 170 172 174160

165

170

175

R² = 0.986842105263158

Simulated Data

Obs

erve

d D

ata

(b)(b)

Figure 5: Comparison of predicted and measured days from sowing to anthesis (a) and to maturity (b).

All of the indices imply that there was a good agreement between simulatedand measured durations (days, d) from sowing to anthesis and from sowing to physiological maturity stages. Based on these results it can be concluded that the model is very robust in predicting the critical phonological growth stages.

Grain yield was very well simulated by the CERES-Wheat model. (Fig. 6; Table 9). The RMSE, normalized RMSE, d-index, and r2 were 173 kg ha-1, 4 %, 0.68 and 0.75, respectively. Aforementioned indexes imply the robustness of the model in simulating wheat grain yield.

13 | P a g e

400041004200430044004500460039003950400040504100415042004250430043504400

R² = 0.755101905172405

Simulated DataO

bser

ved

Dat

a

Figure 6: Comparison between simulated and measured grain yield (kg/ha)

3.2 Validation of DSSAT for wheat trial dataThe experiment for DSSAT validation for wheat variety chakwal-50 was conducted at Barani Agricultural Research Institute, Chakwal in 2014-2015 (Figure 7). Experimental conditions and results obtained (Table 14)from the field trial were used for validation/evaluation of CERES-Wheat model through DSSAT 4.5 software to simulate and predict wheat yield.

14 | P a g e

Figure 7: Field experiment for evaluation of DSSAT for wheat (Chakwal-50)

3.2.1 Crop dataCrop data for wheat variety Chakwal 50 for DSSAT validation was taken from the experiment conducted in the field (2014-15) as shown in Table 10.

Table 10: Wheat crop data forvalidation of DSSAT

Parameters 2014-15Planting date 14-Nov-14Emergence date 20-Nov-14Planting depth (cm) 10Planting method Dry seedPlant population at seeding (plants/m2) 400Plant population at emergence (plants/m2) 200Row spacing (cm) 22.5Anthesis date 20-Mar-15Plant height (cm) 95Maturity date 22-Apr-14Harvesting date 07-May-15Yield (kg/ha) 4100

3.2.2 Initial conditionsInitial moisture contents were recorded before the sowing of wheat crop from 6 different depths (15, 35, 55, 75, 95, 115 cm). It is important to take this data because with these initial conditions model predict the yield and yield components.

Table 11: Initial moisture contents

Year Date Sampling depth Moisture content

15 | P a g e

(cm) (cm3/cm3)

2014-15 04-11-2014

15 0.235 0.255 0.1875 0.295 0.18115 0.18

For DSSAT validation for wheat we added only one year dataset in the model. DSSAT ran and simulated satisfactory results (Table 12), which shows our calibration is good. With one year dataset, it is not possible for the DSSAT to calculate r2 and d-stat factor. But DSSAT calculated RMSE and NRMSE which were in the satisfactory range.

Table 12: Statistical indices for evaluating the performance of DSSAT model in predicting phenological dates and grain yieldduring validation.

Cropping years

Anthesis (DAP) Physiological Maturity (DAP)

Grain yield (kg/ha)

Simulated Measured Simulated Measured Simulated Measured2014-15 128 126 161 159 4295 4100

IndexRMSE (day)a 2 2 195NRMSE (%)b 2 1 5

aRMSE Rootmean square errorbNRMSE Normalized root mean square error

Overall results indicated that the model performance indicators have closer agreement between observed and simulated wheat anthesis day, physiological maturity and grain yield data for the variety chakwal-50.

3.3Calibration of DSSAT for barley under non-stress conditionsModel calibration should be done under non-stressed conditions and as Barani Agricultural Research Institute, Chakwal, worked on barley crop in purely rainfed conditions (without applying irrigation), the data of 3 years (2010, 2012 and 2013) on barley crop under irrigated conditions was taken from Ayub Agricultural Research Institute (AARI), Faisalabad for barley model calibration.

In the soils where wheat, maize, rice and sugarcane cannot be cultivated easily, we can cultivate barley because it is a highly stress-tolerant crop and can resist harsh environments. Not only barley gives good yields with lesser agriculture inputs but it can also replenish the damaged soils. Barley, like all other crops is affected by the soil salinity, but due to its high resistance it can give better yields than other crops. Barley is adaptive to a wide range of soils such as less fertile, fertile, or sandy soils.

Barley variety Joo-83 was selected for the model calibration.16 | P a g e

Table 13: Cultivar coefficients fitted for Joo-83

(Barley variety)Cultivar coefficients

P1V 12P1D 30P5 625G1 13G2 21G3 3PHINT 70

The DSSAT model was calibrated by using the Barley crop data of 2010, 2012 and 2013 (Table 14).

Table 14: Crop data used for DSSAT calibration for Barley crop

Parameter Year2010-2011 2012-13 2013-14

Planting date 29-Oct-2011 04-Nov-2012 01-Nov-2013Planting depth (cm) 10 10 10Emergence date 04-Nov-2011 09-Nov-2012 07-Nov-2013Plant population at seedling (plants/m2) 350 350 350

Plant population at emergence(plants/m2) 230 230 230

Planting method Dry seed Dry seed Dry seedRow spacing (cm) 30 30 30Anthesis date 07-Feb-2011 08-Feb-2013 05-Feb-2014Plant height (cm) 81.4 73.2 81.4Tillage date 06-Oct-2011 06-Oct-2012 06-Oct-2013Tillage implement Cultivator Cultivator CultivatorTillage depth (inch) 8 to 12 8 to 12 8 to 12Harvest area/m2 1 1 1Harvest method Manual Manual ManualMaturity date 23-Mar-2011 28-Mar-2013 4-Apr-2014Harvesting date 22-Apr-2012 20-Apr-2013 23-Apr-2014Yield (kg/ha) 1480 1190 1250

3.3.1 Initial conditionsThe most important part of running any crop model is the initial soil moisture contents.

Table 15: Initial conditions for barley cv. Joo-83.

17 | P a g e

Date of measurement

Depth of sampling (cm)

Moisture content (cm3/cm3)

29-Oct-2010

15 0.1135 0.155 0.0975 0.195 0.11

4-Nov-2012

15 0.1535 0.1355 0.1175 0.1295 0.15

1-Nov-2013

15 0.1435 0.1355 0.1275 0.1195 0.11

3.3.2 Fertilizer applicationTable 16: Fertilizer application in barley

Date of application Fertilizer type N (kg/ha) P (kg/ha)29-Oct-2010 Urea 110 4004-Nov-2012 Urea 110 4001-Nov-2013 Urea 110 40

3.3.3 Performance evaluation of DSSAT calibration for barleyThe performance of the DSSAT model after calibration for barley was evaluated. The variables tested included the key phenological dates (anthesis and harvest maturity) and the final harvested yield. In general, the model gave good predictions of crop development and final grain yieldsfor the cultivar Joo-83 (Table 17).For calibration, the simulated dates of anthesis and physiological maturity as well as yield were compared with the observed data (Table 18).

Different statistic indices were employed, including Coefficient of Determination(r-square), absolute and normalized Root Mean Square Error (RMSE and NRMSE) and index of agreement (d-index). Normalized RMSE gives a measure (%) of the relative difference of simulated versus observed data. The simulation is considered excellent with a normalized RMSE is less than 10%, good if the normalized RMSE is greater than 10% and less than 20%, fair if normalized RMSE is greater than 20 and less than 30%, and poor if the normalized RMSE is greater than 30% (Jamieson et al.,1991). The index of agreement (d) proposed by Willmott et al.(1985), according to the d-statistic, the closer the index value is to one, the better the agreement between the two variables that are being compared and vice versa.

18 | P a g e

Table 17: Statistical indices of evaluating the performance of DSSAT CERES-Barley model in predicting phenological dates and simulating grain yield.

Cropping years

Anthesis (DAP) Physiological Maturity (DAP)

Grain yield (kg/ha)

Simulated

Measured Simulated Measured Simulated

Measured

2010-2011 103 101 151 145 1704 14802012-13 97 96 146 144 1341 11902013-14 100 96 156 154 1459 1250

IndexRMSE (day) 3 4 174NRMSE (%) 3 3 13d-stat 0.783 0.843 0.737r-square 0.75 0.82 0.99

RMSE Rootmean square errorNRMSE normalized root mean square errord-stat Wilmot's index of agreementr-square Coefficient of determination

Spring barley was calibrated in the same way as wheat - the difference between the simulated and observed anthesisdate as well as physiological maturity date varied between 2 and 3 d; the simulated yield was within 20 % of the measured values for each year (R² = 0.99; RMSE = 174 kg ha–1).

90 92 94 96 98 100 102 10490

95

100

105

R² = 0.75

Simulated Data

Obs

erve

d D

ata

(a)

19 | P a g e

140

144

148

152

156

160

140142144146148150152154156158160

R² = 0.824175824175824

Simulated Data

Obs

erve

d D

ata

(b)

1000 1200 1400 1600 18001000

1200

1400

1600

1800

2000

R² = 0.983948914751992

Simulated DataO

bser

ved

Dat

a

(c)

Figure 8: Comparison of simulated and measured (a) anthesis day, (b) maturity day and (c) yield of barley cv. Joo-83.

All of the indices imply that there is a good agreement between simulated and measured durations (days, d) from sowing to anthesis and from sowing to physiological maturity stages. Based on these results it can be concluded that the model is very robust in predicting the critical phenological growth stages.

3.4 Validation of DSSAT for barleyValidation of DSSAT for barley (Joo-83) was done from the data set of 2014-2015 obtained from BaraniAgricultural Research Institute, Chakwal experimental site.

Table 18: Crop data for DSSAT validation for barley (Joo-83)

Parameter 2014-15Planting date 07-Nov-2014Planting depth (cm) 10Emergence date 12-Nov-2014Plant population at seedling (plants/m2) 330Plant population at emergence(plants/m2) 200Planting method Dry seedRow spacing (cm) 30Anthesis date 06-Feb-2014Plant height (cm) 87Tillage date 06-Oct-2014Tillage implement CultivatorTillage depth (inch) 8 to 12Harvest area/m2 1Harvest method ManualMaturity date 31-Mar-2015Harvesting date 27-Apr-2015

20 | P a g e

Yield (kg/ha) 800

3.4.1 Initial conditionsInitial moisture contents were recorded before the sowing of barley crop from 5 different depths. It is important to take this data because with these initial conditions model predict the yield and yield components.

Table 19: Initial moisture contents

Date of samplingDepth of sampling

(cm)Moisture content

(cm3/cm3)

7-Nov-2014

15 0.1335 0.1255 0.1275 0.1195 0.12

For DSSAT validation for barley we only added one year dataset in the model. DSSAT run and simulatedsatisfactory results (Table 20), which shows our calibration is good. With one year data set it is impossible for the DSSAT to calculate R2 and d-stat factor. But DSSAT calculated RMSE and NRMSE which were in the range.

Table 20: Statistical indices of evaluating the performance of DSSAT model in predicting phenological dates and simulating grain yield for evaluation (validation).

Cropping years

Anthesis (DAP) Physiological Maturity (DAP)

Grain yield (kg/ha)

Simulated Measured Simulated Measured Simulated Measured2014-2015 91 88 146 143 992 800 Index RMSE (day) 3 3 192NRMSE (%)

3 2 24

RMSE Rootmean square error NRMSE normalized root mean square error

Overall results indicated that the model performance indicators have closer agreement between observed and simulated barleyanthesis day, physiological maturity and grain yield data for the variety Jo0-83.

3.5 Scenario simulationPlanting dates treatments:

21 | P a g e

An analysis of the effect of different sowing dates and supplement irrigation on yield of wheat and barley was conducted using long-term 30 year historic (1984-2014) daily weather data of BARI. Six different sowing dates (20 Oct, 30 Oct, 10 Nov, 20 Nov, 30 Nov and 10 Dec) were simulated using the seasonal analysis (Figure 9)tool of DSSAT Version 4.5 under rainfed and irrigated conditions. This period is the typical sowing window in the region, however, the early and late sowing dates are not suitable to obtain high grain yields, but due to the limitation of theavailable water, wheat may be sown early and due to delay in harvesting previous crops may be sown at end of the window.

Figure 9: Scenario simulation window of seasonal analysis tool of DSSAT

22 | P a g e

Seasonal analysis tool of DSSAT was provided with dates of sowing for rainfed, SI at the time of sowing and SI 30 DAS. DSSAT model processed the provided information for previous 30 years and gave results which are presented in figure 10.

3.5.1 Model application: determining optimum sowing dates and Supplement irrigation for wheat

The analysis of the effects of different sowing dates on wheat and barley was conducted based on 30 years of historical weather data from the BARI location. At BARI Chakwal location, long-term simulated yield ranged from 877 to 6422 kg/ha depending upon the sowing date. The highestyield was attained for sowing on 10-30 November with supplemental irrigation either at time of sowing or 30 DASand the lowest yield through sowing on 20 October under rainfed conditions. Delay in sowing date from 20 October to 30 November has resulted in increase in yield. Grain yield decreased with delay in sowing from 30 November to 10 December (Figure 10).

Figure 10: Box graph showing increase or decrease in maturity yield kg/ha(* in the boxes represent the 50th percentile (probability).Treatment 1 – 6 = Rainfed, Treatment 7-12 = SI at sowing, Treatment 13-18 = SI at 30 DAS).

The above graph is showing the yield trend with dates of sowing and supplements irrigations (Figure 10). 1-6 are the 6 sowing dates under rainfed conditions, 7-12 are the 6 sowing dates with 30 mm supplement irrigation at the time of sowing (if initial moisture content in the soil is less for plant germination) and 13-18 are the 6 sowing dates with 30 mm supplement irrigation

23 | P a g e

30 DAS. It is clearly seen in the graph that the highest yield with 50% probability could be attain if sowing date is 10 Nov to 30 Nov.

Effect of auto-irrigation on potential yield of wheat:

When we gave DSSAT the condition of auto irrigation with the same planting dates (20 Oct, 30 Oct, 10 Nov, 20 Nov, 30 Nov and 10 Dec ) then model gave the following trend (Figure 11). The highest yield (6400kg/ha) could be attained if wheat sow on 20 October if irrigation amount of 315 mm should apply to wheat. Potential yield could be 6400 kg/ha with 315 mm supplement irrigation when sow on 20 October and farmer could attain 5000 kg/ha with 30 mm supplement irrigation when sow wheat on 20-30 November.

(a) (b)

Figure 11: (a) Irrigation amount (b) potential yield of wheat that DSSAT suggested

3.5.2 Model application: determining optimum sowing dates and Supplement irrigation for Barley

For barley, same treatments were considered except the sowing dates (figure 12). Sowing dates considered for barley were 15 Oct, 25 Oct, 05 Nov, 15 Nov, 25 Nov and 05 Dec.

24 | P a g e

Figure 12: Box graph showing increase or decrease in mat yield kg/ha of barley. (* in the boxes represent the 50th percentile (probability).Treatment 1 – 6 = Rainfed, Treatment 7-12 = SI at sowing, Treatment 13-18 = SI at 30 DAS).

Higher yield could be attained if barley was sown on 5 December in case of rainfed condition and if there is a 30 mm supplement irrigation then it should have been applied on 25 Nov at the time of sowing or 30 DAS.

In case of auto irrigation model gave the highest yield with 50% probability on 05 December with 170 mm supplement irrigation (Figure 13). But a farmer can have maximum yield of 1200 kg/ha with 30 mm supplement irrigation (Figure 12)

25 | P a g e

Figure 13: Potential yield of Barley with auto-irrigation at 25%, 50% and 70% probability

These are the DSSAT suggested results for barley and wheat. But to check the model compatibility with actual field conditions, a trialsare in field of BARI with the wheat and barley sowing on 20 November and 5 December, respectively under rainfed, SI at sowing and SI 30 DAS

4 CONCLUSIONS

DSSAT model was applied to rainfed area of chakwal district to simulate the growth stages of wheat (Chakwal-50) and barley (Joe-83) for the period of 1984-2014. The model was first calibrated using the locally measured crop yield and initial soil moisture contents and later on this well calibrated model was used to simulate the best planting dates and supplement irrigation scenarios under different climatic conditions.

5 RECOMMENDATIONS/POLICY BRIEF

For wheat

If there is enough rainfall in August and September, then for rainfed conditions best sowing date for wheat is 20-30 November to get highest yields.

If there is not enough rainfall in July - September and a farmer have water for only one irrigation then he must go for either applying it at the time of sowing or 30 DAS and plant wheat on 30 November to get high yields.

For barley

In rainfed conditions sowing around 05 December to have highest yield.

26 | P a g e

If irrigation is applied at sowing or 30 DAS then the best sowing date could be 25 Nov to have high yields.

In case of dry year apply 30 mm of irrigation either at the time of sowing or 30 DAS. In case of wet year do not go for supplement irrigation but sow wheat on 20 Nov and barley on 30 Nov or 05 Dec.

Note:

Dry year means if the rainfall is less than 200 mm, normal year if rainfall is 500-600 mm and wet year means if rainfall is above 1000 mm.(This wet, medium and dry category is based on last 30 year rainfall data set)

This policy is based on the DSSAT suggested results for barley and wheat. But to check the model compatibility with actual field conditions, trials must be conducted in the field.

6 REFERENCESADB, WB and GoP (2010). Pakistan Floods Damage and Need Assessment. Islamabad, Pakistan. Asian

Development Bank, World Bank, and Government of Pakistan.

Akmal, M., et al. (2014). Climate Change and Adaptation– Farmers’ Experiences from Rainfed areas of Pakistan. Study conducted by Climate Change Centre, The Agricultural University Peshawar.

Godwin, DC, Ritchie JT, Singh U, Hunt L (1989). A user’s guide to CERES-Wheat v2. p. 10.

Government of Punjab (2009b). Pre-Investment study on Chakwal District 2009. Directorate of Industries Government of Punjab, Lahore, Pakistan.

Government of Punjab (2013). Punjab Development Statistics 2013, Bureau of Statistics, Government of Punjab. Lahore.

Hoogenboom, G., Jones, J.W., Wilkens, P.W., Porter, C. H., Batchelor, W. D., Hunt, L. A., Boote, K. J., Singh, U., Uryasev, O., Bowen, W.T., Gijsman, A. J., Du Toit, A., White, J.W. & Tsuji, G.Y. (2004). Decision Support System for Agrotechnology Transfer Version 4.0 (CD-ROM). Honolulu, HI: University of Hawaii.

Hunt, LA, Pararajasingham S (1994). Genotype coefficient calculator. In: IBSNAT (International Benchmark Sites Network for Agrotechnology Transfer) (eds.), DSSAT Manual Version 3, vol. 3, pp. 201-223.

Jamieson, P.D., Porter, J.R., Wilson, D.R., 1991. A test of computer simulation model ARC- WHEAT1 on wheat crops grown in New Zealand. Field Crops. Res. 27, 337–350.

Jones, J.W., Hoogenboom, G., Porter, C. H., Boote, K. J., Batchelor, W. D., Hunt, L. A., Wilkens, P.W., Singh, U., Gijsman, A.J. & Ritchie, J. T. (2003). DSSAT cropping system model. European Journal of Agronomy 18, 235–265.

Willmott, C.J., Akleson, G.S., Davis, R.E., Feddema, J.J., Klink, K.M., Legates, D.R., Odonnell, J., Rowe, C.M., 1985. Statistic for the evaluation and comparison of models. J. Geophys. Res. 90,8995–9005.

27 | P a g e

Yu, Winston et.al. (2013). The Impacts of Climate Risks on Water and Agriculture. The Indus Basin of Pakistan. World Bank Washington DC.

28 | P a g e