LONG TERM CHANGES IN IRRIGATION WATER REQUIREMENT

Abstract

It is widely believed that the impact of climate change on agriculture has become one of the important issues in water resource management. The available water resource would be altered by change in rainfall pattern and rate of evaporation. Further, higher evapotranspiration (ET) would result in greater amount of irrigation water requirement (IWR). Despite availability of a number of ET estimation methods in literature, the accurate assessment of ET/IWR is a complicated task due to the limitations and assumptions associated with different methods. It is understood that the climate change may alter the demand for irrigation water in future on regional and the global scale. Hence, there is a need to study long term change in the key climatic variables (rainfall, minimum and maximum temperature, relative humidity, wind speed) which affect the ETo/CWR/IWR. Very few studies have been carried out in India on long term changes in irrigation water requirement. Present study is taken up to enhance the understanding of region specific changes in IWR on long term basis. Earlier studies have focused on assessment of climatic variables and crop water requirement based on perturbation method for scenario generation with GCM. With the development of statistical downscaling model (LS-SVM), the regional climate change assessment studies are becoming more accepted. Therefore, this study proposes to use LSSVM model to study the impact of climate change on IWR. This study has focus on quantification of future irrigation requirements on long term basis, which is necessary for sustainable management of basin water resources. This study has been carried out in the Seonath river basin (area = 30,860 sq. km), falling in Chhattisgarh State (India), is the longest (380 km) tributary of Mahanadi Basin, comprising 25% of the basin area. Agriculture is the main occupation of the people in the sub-basin. There are two cropping seasons viz, kharif (mid June to October) and rabi (November to mid April). The mean annual rainfall in the basin varies from 1005 mm to 1255 mm. The major part of rainfall occurs only within 3 monsoon months (July- September). It is also reported that the study area faces adverse effects of frequent droughts and thus crop production is adversely affected in drought years. The trends in annual and seasonal rainfall time series from 1960-2010 have been analyzed using Mann–Kendall test and the Sen’s Slope estimator for 24 stations in the Seonath river basin. The analysis has revealed that there is a significant decreasing trend in iii annual rainfall (-2.4 mm/yr) at 75% of the stations (northern part of basin) and nonsignificant decreasing trend in annual rainfall at 17% of the stations (southern part of basin). Moreover, the decreasing trends in seasonal rainfall are found significant for most of the stations. Decrease in monsoon rainfall at the rate of 2.79 mm/yr is likely to have significant adverse impact on rainfed agriculture in future. The conventional approach of planning for Rabi crop irrigation needs to be critically examined. Also, there is a need to examine supplemental irrigation requirements for Kharif (monsoon) season crops in the region. Rising trend has been observed in mean maximum temperature for monsoon and winter whereas there is decreasing trend in mean maximum temperature for summer season. The mean minimum temperature in monsoon, winter and summer seasons shows rising trend all over the basin. Few stations located in Northern part of the basin show nonsignificant rising trend in mean seasonal temperature. The minimum temperature has increased more as compared to maximum temperature over 51 years period of analysis. The percentage change in minimum temperature is highest for the month of November followed by December and January. The variability is observed to be more pronounced in minimum temperature ranging from 1.69% to 2.78%. For annual maximum and minimum temperature, the upper half of the basin shows more variability. The results of study indicate that the mean annual temperature is likely to increase by 1.98°C in next 100 years. Further, winter temperature may increase by 2.06°C, monsoon temperature may increase by 4.73°C and summer temperature may decrease by -0.528°C over the study area. The temperature changes may have significant impacts on rainfed crop cultivation due to increase in evapotranspiration. In the study area, the monsoon temperature is expected to increase by 4.73°C over 100 years. This rise in temperature may cause significant increase in the irrigation water requirements and may cause water shortages. Therefore conventional irrigation planning procedures for Rabi as well as Kharif crops need to be revised. Monthly trend analysis of Relative Humidity (RH) shows significant decreasing trend for months of July, September, October and November. Whereas, from March to June insignificant increasing trends are observed. The highest change in magnitude of RH is observed for July, September, October and November months. The inter-annual variability in RH of the basin ranges from 0.9 to 2.2%. iv The monthly and seasonal assessment of trend in wind speed (WS) and its variability is significant in order to quantify its effect on ET. On seasonal basis, significant increasing trend is obtained for WS in monsoon and winter season all over the basin. On monthly time scale, the highest rate of change is seen in August followed by July, June and September. The percentage change is highest for the entire basin ranging from 38% to 61%. The interannual variability (23%) is observed in monthly WS in northern part of the basin. Overall, there is increasing trend in monthly and seasonal WS for the entire basin. To measure the consistency and accuracy of ETo methods, the estimates obtained from six different methods (Hargreaves, Thornthwaite, Blaney-Criddle Method, Priestley- Taylor Method, Penman-Monteith Method and Turc Method) have been compared with pan evaporation data (Ep). According to statistical performance evaluation Penman-Monteith, Hargreaves and Thornthwaite methods have performed well. The radiation-based Priestley- Taylor and temperature based Blaney-Criddle method indicate lowest correlation values. The pan coefficient (Kp) has been estimated for the study region. The study illustrates that the Kp varies significantly from month to month (0.56 to 0.89) for the study area. The highest and lowest Kp value have been obtained for the month of July and November, respectively. Thus, if the standard average value of Kp (0.70) is used for assessment of ETo, it will provide erroneously large variation in ETo ranging from 11.8% to 56.3%. According to sensitivity analysis temperature is the most important driving parameter which effects ETo and next to that is relative humidity. Bilaspur station shows highest sensitivity coefficient of 1.77 in relation to temperature. It means ETo would increase by 17.7% in response to the 10% rise in maximum temperature if other meteorological variables remain constant. However Rajnandgaon station shows the highest value of sensitivity coefficient in relation to RH (-1.28) which means 10% decrease in RH causes increase in ETo by 12.8%. Hargreaves and Thornthwaite methods are therefore not recommended for this study area as these methods donot take into consideration the RH parameter. In this study, the Kc values recommended by FAO paper No. 56 have been adjusted according to climatic conditions of the study area. The average Kc values for major crops (kharif paddy, rabi wheat and summer paddy) for four crop growth stages viz, initial, v development, mid and late season have been computed. For Kharif paddy, percentage change in adjusted Kc value with respect to FAO recommended Kc values during different crop growth stages varies from -1% to -15% whereas for rabi crops (Wheat and Summer paddy) it range from -2% to -16% and -9% to -23% respectively. The CWR computed using FAO-56 Kc values gives significantly different (higher) values due to sub-humid climate of the basin. It is therefore, decided to use the adjusted Kc values for precise estimation of CWR and subsequently IWR. Trend and variability of annual and monthly ETo time series have been analyzed for 8 stations for which data are available. The increase in ETo is estimated as 13.4 mm/yr on annual time scale. On the seasonal scale, summer ETo trend is decreasing by -10.4 mm/yr. The winter and monsoon ETo show increase at the rate of 21 mm/yr and 22 mm/yr, respectively. The estimates of ETo for the months of December, January, February, July and August show non-significant increasing trend. However significant increasing ETo trend have emerged for the months of September October and November. The highest (3.4-3.6%) variability in annual ETo is seen in the stations located at southern part of the basin while rest of the stations exhibits inter-annual variability ranging from 1.0%-1.8%. The results of this study will be useful for the reliable estimation of supplemental irrigation water requirements. In order to detect trends in IWR, the MK-test and Sen's slope have been used for the 51-year period. For Kharif season increasing trend is detected at 88% of the stations, and remaining 12% of the stations show non-significant increasing trend. Further, significant positive slopes are dominant for wheat crop, (with 63% of the stations). For summer paddy, 50% of the stations show significant increasing trend and rest 50% shows non-significant increasing trend. The IWR for Kharif and Rabi seasons are increasing at the rate of 3.627 mm/yr and 1.264 mm/yr respectively. These changes are characterized by a relative increase in Kharif IWR by 47%, while Rabi IWR by 23%. Overall, the results of the study show an increase in IWR for agricultural crops it may be due to high variability of rainfall pattern, rise in temperature, wind speed and decrease in RH. These findings shall be helpful in more realistic planning and efficacious utilization of basin water resources. vi In a recent study by Mishra et al., (2014), developed a relationship between Soil Conservation Service Curve Number (SCS-CN) and ETo. Since ETo is a important parameter in estimation of IWR, therefore an attempt has been made to develop a relationship between IWR and CN. In this study, the CN derived from rainfall-runoff data on seasonal scale (Kharif and Rabi season) has been related to IWR of same scale and high R2 values of 0.970 and 0.926 for Kharif and Rabi seasons are found for calibration period. The results are validated with R2 values of 0.957 and 0.954 for Kharif and Rabi seasons, respectively; indicating the existence of a strong IWR-CN relationship. The supportive results of the proposed model assume to be a good substitute for complex IWR assessment, particularly in the area where meteorological parameters are not easily obtainable. The four statistical downscaling models viz., Artificial Neural Network (RBF), Multilayer Perception (MLP), Multiple Linear Regression (MLR), Model Tree (MT), Least Square Support Vector Machine (LS-SVM) are used for comparative study. The results of analysis indicate that for each climatic variable, LS-SVM model is performing best followed by MT and ANN (MLP). The annual rainfall is projected for the period of 2011-2100 and it is expected to increase from year 2020s upto 2090 in range varying from 2.74 mm/decade to 18 mm/decade. The annual rainfall is predicted to decrease for the period of 2091-2100. However for maximum temperature the increasing trend is predicted for the entire projected period and the highest temperature change is predicted for two decades i.e, 2021-2030 and 2031-2040. The rate of change may vary from 0.1°C/decade to 0.5°C/decade for monsoon and 0.01°C/decade to 0.3°C/decade for post monsoon season. For the minimum and mean temperature the overall increasing trend is observed but for Tmin the highest temperature rise is expected in the period of 2061-2070. The change in magnitude for minimum temperature for monsoon season is varies from 0.2°C/decade to 0.7°C/decade, whereas for post-monsoon season the minimum temperature may vary from 0.02°C/decade to 0.5°C/decade. It can be inferred that warming is expected to be more pronounced during the night than day. The relative humidity forecasts represent a significant decreasing trend for Kharif season and non-significant decreasing trend for rabi season for two decades i.e., 2020s and 2090s period. The projected wind speed shows non-significant increasing trend vii for the entire basin. Wind speed projections are highly uncertain with extremes in 2090s during Kharif season whereas for rabi season the uncertainty is for 2020s and 2050s period. The ETo have been predicted to increase in future for all months. Particularly, the change in ETo is more in the months of May to August due to the large projected changes in Tmax and Tmin variables. The peak is observed for the month of June 25 mm over 100 years. The monthly IWR in future have been estimated using the projections of rainfall (downscaled from LS-SVM model) and CWR projections. The IWR for Kharif paddy crop is projected to increase by 84%, 71% and 32% in the 2020s, 2050s and 2090s respectively whereas, for Rabi wheat crop IWR is predicted to increase by 201%, 163%, and 91%, for the three decades (2020s, 2050s, and 2090s). However for summer paddy the IWR may increase by 184%, 215% and 90% for 2020s, 2050s and 2090s periods respectively.