SPATIAL AND TEMPORAL VARIATIONS OF CLIMATE VARIABLES OVER A RIVER BASIN

Variation in the climate acts as an important factor in managing the natural resources in order to meet the needs of human life for present and future generations. Future projections of the climate data obtained from the climate models help in developing the policies for the sustainable use of natural resources. In the present study, changes in the climate variables were assessed both spatially and temporally using Regional Climate Models (RCM) database under Coordinated Regional Downscaling Experiment (CORDEX) from Centre for Climate Change Research (CCCR), Pune, for Krishna river basin, India. Uncertainties in the climate variables were reduced by using Reliable Ensemble Averaging (REA) method. The results suggest that the ability of REA data performs well throughout the basin except in the upper region of the Krishna basin. First future period shows around 20 per cent decrease when compared to the historic period where the other two future periods show a less change in the precipitation.


Introduction
The local and global pressures on natural resources are increasing because of the external forces like high living standards, anthropogenic changes, land use and water management policies etc. In addition, climate change is also contributing pressure on natural resources externally. Generally, the long-term change in the properties of climate system due to natural and forced variability and the effects of anthropogenic activities is known as climate change. The variations in climate system help in Journal of Rural Development,Vol.37,No. (2), April-June:2018 altering water availability regionally, selection of the crop and vegetation based on the evapotranspirative water demands, salt-water intrusion in coastal regions, floods and drought extremes, groundwater recharge, water quality and other related processes. The additional stress developed by this climate change on the natural resources like water provides a clarity to the water managers and policymakers for efficient water supply for future periods (Mondal and Mujumdar, 2015). The future water demands will be more uncertain in addition to the uncertainty developed due to changes in demography and climate (Yang et al., 2008).
Global Climate Models (GCMs) are the coarse resolution climate models projected under increased global temperatures for large spatial scales, whereas finer spatial scales climate models for the better management of the resources at the basin level. Many studies have proved that the use of regional climate data for impact assessment is more reliable compared to the global climate model data (Chien et al., 2013;Deshpande, 2014;Demaria et al., 2016 (Giorgi and Mearns, 2003;Tebaldi and Knutti, 2007). The Reliability Ensemble Averaging (REA) is the method used to address the uncertainty developed using different RCMs (Giorgi and Mearns, 2003;Chandra et al., 2015).
The biases in the REA precipitation data are corrected statistically by Quantile-Quantile (Q-Q) mapping which improves the ability to project the future climate models data for the impact and vulnerable studies (Piani et al., 2010).
In India, Krishna river is categorised as the economical water-scarce and food-deficit basin (Amarasinghe et al., 2004;Gosain et al., 2006).
The main feature of the basin being high crop production, the seasonal or regular water predictions are likely to experience stressed conditions. It is also evident that the annual average renewable water availability per person is less than 500m3/cap/yr (Gosain et al., 2011) which emphasise the importance of water supply and demand in the basin. The main objective of the study is to assess the changes in the climate variables like precipitation, maximum and minimum temperatures both spatially and temporally. The climate model data obtained from five RCMs of Representative Concentration

Reliability Ensemble Averaging (REA) Method
The REA method proposed by Giorgi and Mearns, 2003, provides The statistical transformation is modeled using the non-parametric regression with the monotonic tricubic spline interpolation. The smoothing spline fits the fraction of the CDFcorresponding to observed wet days by assigning zero to the non-zero of the CDF-corresponding to observed wet days by assigning zero to the non-zero values of the modeled data. Figure 2 represents the REA precipitation data with observed data before and after the transformation for a grid.   The mean monthly maximum temperature for the future period 2 (2040-2070)

a) Precipitation data without bias correction b) Precipitation data with bias correction
projects highest values when compared to other three periods. Table 3 represents the maximum and minimum values of the REA data in comparison with the observed data projecting a decrease in the precipitation and increase in the temperature data.

Conclusion
In this paper, Reliability Ensemble Average Tungabhadra. Around 20 per cent decrease in the precipitation data in the Future 1 period is observed when compared to the Historic period.
Therefore, hydrology of the river basin simulated using the climate data obtained from REA for the Future periods help water managers and policymakers in developing the adaptation strategies for proper utilisation of resources.