244 M. Jawed Iqbal et al 2.2. Methodology To examine the impact <strong>of</strong> agro-climatic parameters season mean <strong>of</strong> crops from November to April. It is important to understand the role <strong>of</strong> agro-climatic parameters in modelling for wheat crop. Albeit techniques for identifying parameters range from simple correlation analysis to advanced procedures such as canonical correlation and neural network [13, 14], we used the Pearson correlation analysis calculated between crop yields and agro-climatic parameters (e.g. rainfall, GDD, NVDI, SOI, etc.). Secondly, multiple linear regression equation was constructed between wheat yields <strong>of</strong> Punjab variance associated with each parameter designates the relative importance <strong>of</strong> the index in modulating the crop variability. Moreover, the total variance inter-annual variations. The linear regression model Y = b 0 + b i X i , (i {0, 1, 2, 3, ... ,p}), where Y is the estimated wheat yield in standardised unit, b 0 is the intercept, X i is a value <strong>of</strong> the i th predictor (agro-climatic parameter) in the model, b i is the model parameters representing the linear relationship between the predictor and predicted, p the number <strong>of</strong> the predictors variable in the model. 3. RESULTS AND DISCUSSION Percentage departure <strong>of</strong> wheat yield in Punjab province from median 599.2 is plotted in Fig. 3. Percentage departure started from 65.5% and there is a fall initially up to minimum value <strong>of</strong> 132.2% (in 1952). After that it rose linearly with some ups and downs. It had the maximum value <strong>of</strong> 46.6% in the year 2006. Wheat yield was below the median till year 1977. Since 1978 the yield was above the median, except in 1983. In our study we also observed that during La Nina years (i.e., 1947, 1948, 1949, 1954, 1955, 1956, 1964, 1967, 1970, 1971, 1973, 1975, 1988, 1998, 2000, and 2007) the average yield <strong>of</strong> wheat was 558.32 kg/acre. During El-Nino Years (i.e., 1951, 1957, 1963, 1965, 1969, 1972, 1976, 1982, 1986, 1987, 1991, 1997, 2002 and 2009) the average yield was 621.68 kg/acre, which clearly indicated that warm phase <strong>of</strong> ENSO had positive effect on wheat yield. . Sarma et al [9] have also obtained similar results for rice production in Andhra Paradesh, a province <strong>of</strong> India. 3.1 Variation in Wheat Yield with Meteorological Parameters increasing trend (Fig. 4). The annual rainfall and wheat crop yields are plotted in Fig. 5. It shows that the wheat yield increased linearly and its trend is linear from 1987 to 1998. But overall its trend did not described the variation in wheat yield. The highest wheat yield was recorded in the year 1999 when the rainfall was observed 453 mm and the minimum wheat yield was recorded in the year 1983 when the rainfall was 766 mm. Pearson correlation 0.16, wheat area is irrigated by tube wells, this might be relationship <strong>of</strong> crop yield with the rainfall was also observed by Sarma et al [9]. Fig. 6 depicts the plot <strong>of</strong> wheat yield with Nino Table 1. Parameters/Months Nov Dec Jan Feb Mar Apr Nov-Apr Selected Months SOI 0.38 * 0.36 * 0.12 0.30 0.25 0.14 0.28 0.38 * Nov-Dec GDD 0.45 ** 0.42 * -0.12 0.24 0.09 0.37 * 0.41 * 0.50 ** Nov-Dec NDVI -0.08 -0.29 0.40 * 0.58 *** 0.46 ** 0.03 0.22 0.52 ** Jan-Mar
Agroclimatic Modelling for Estimation <strong>of</strong> Wheat Production 245 Fig. 3. Percentage departure <strong>of</strong> wheat yield in the Punjab. Fig. 4. Five-year moving average <strong>of</strong> wheat yield in the Punjab. Fig. 5. Variability <strong>of</strong> wheat yield with rainfall in the Punjab.