Climatic Change and its Impact on Wheat Production in Dinajpur District: Modeling and Quantitative Analysis

Abstract

Dinajpur is the leading district for wheat cultivation in Bangladesh and wheat is much sensitive to climatic variation and change. An attempt has made to study the climatic change and its impact on wheat production in Dinajpur district and both the secondary and primary data are collected in this respect. The historical climatic data are collected for 1948-2004 which contain a missing of 8 years during I 973-1980. Primary data are collected by a sample survey to know the people's perception regarding change of climatic behavior on wheat production over the period. In this study, climatic trend and variability of Dinajpur district are examined for 19 climatic variables in decadal, annual, seasonal [three crop seasons like Rabi ={November-February), Prekharif =(March-May) and Kharif =(June-October)], monthly and wheat growing period using exploratory data analysis (EDA) tools such as boxplot, stem-and-leaf display, median polish table etc. Different robust and nonrobust measures are used and outliers are detected. Residual's stationarity is checked by Box-Pierce test statistic and normality is tested by RM (rescaled moment) test. Upward trend is found in means of monthly total rainfall and average cloud, relative humidity, relative humidity at mourning, relative humidity at evening, wet bulb temperature, wind speed, maximum wind speed and minimum temperature (slight) and downward trend is observed in their coefficient of variations (CVs). Both the variations and means for average difference of dry and wet bulb temperature, difference in mourning and evening relative humidity, evaporation and range temperature demonstrate downward trend. The monthly means of average maximum temperature and sunshine hour indicate downward trend and their CVs indicate upward trend. The average dry bulb temperature follows upward trend in Rabi and Kharif season and downward trend in Prekharif season. The average soil temperature at the depth of 5cm follows upward trend during 1987-2000 and sunshine hour show downward trend in all the three seasons during 1989-2004. The total rainfall shows upward trend in February, April, May, September, October and December but it does not show any trend in January, March, June, July, August and November. Time series properties of climatic (monthly data) and wheat production data are investigated from fitted ARlMA models based on minimum root mean square forecasting error of 24 and the values for 2005 to 2008 and 2009 to 2012 are forecasted from selected models. The missing values are forecasted by fitting ARIMA models. The year 1981 of the variables AMNT, A WBT, A WS, ARH and ARH(0-12) are detected as outlier in fitting the models during 1981- 2004 and the outliers are replaced by the forecasted observations from the data during 1982-2004. The square root transformed monthly data of total rainfall during 1948-2004 follow ARIMA (100) (111) model and the ADBT follow AR.IMA (011) (IO1). The production of wheat during 1949-2001 in Dinajpur district primarily follow upward linear trend and residuals are departed from normality but support non-autocorrelated structure and one outlier is identified in 1982 from the plot of deletion Studentized residuals. We have got an opposite conclusions from outlier corrected data. Structural change is observed in 1976 and the data for 1949 to 1975 and 1976 to 2001 separately follow AR (1) models but not for the overall period. The corrected data set during 1949-2001 adequately fits the 1st order Piecewise Autoregressive Model PAR (l) [suggested by lmon (2007)]. It may be mentioned that the reason for not considering the wheat production during 1948-2004 is that the data during that period do not adequately fit PAR (l) model. To quantify the impact of climatic change on wheat production during the period of 1948-2004 we have tried to fit a multiple regression model of wheat production data on climatic data (Nov-Mar) where one dummy variable is taken for the structural change in wheat production data in 1976. Multicollinearity among the predictor variables and normality and stationarity for the residuals are checked. Strong evidence is not found for multicollinearity and first order autocorrelated structure is found in the residuals. The parameters have been estimated using Cochrane-Orcutt iterative procedure where residuals follow normality and stationrity after removing two hi-leverages and two outliers and we observe no significant coefficient except dummy variable and this may be due to technological change such as high yielding variety, improved physical inputs like irrigation, fertilizer, pesticide etc. In the survey, hundred percent (100%) respondents have opined that there is a change in climate than previous time and the changing behavior of climate has severely affected the crop production, acreage under crop and the total process of production as a whole. Insufficient rain in Rainy season, lesser coldness in Winter season, lesser temperature in Summer and lesser dew in Autumn are experienced by them. Stormy and dusty wind does not blow as it blew before and cold affects people at night in the month of Choitra (mid March-mid April). Due to the unusual behavior of climate wheat production has suffered a lot. According to them, Changes of temperature are affecting the production of wheat greatly and they think those to be the most potential reasons. Changing of climate may pose a big and devastating threat to the production of wheat and 88% people opine that the cultivation of wheat in future may fall under a serious threat.