Assessment and evaluation of potential climate change impact on monsoon flows using machine learning technique over Wainganga River basin, India

Abstract

In this study, classification-and regression-based statistical downscaling is used to project the monthly monsoon streamflow over the Wainganga basin, India, using 40 global climate model (GCM) outputs and four representative concentration pathways (RCP) scenarios. Support vector machine (SVM) and relevance vector machine (RVM) are considered to perform downscaling. The RVM outperforms SVM and is used to simulate future projections of monsoon flows for different periods. In addition, variability in water availability with uncertainty and change point (CP) detection are accomplished by flow-duration curve and Bayesian analysis, respectively. It is observed from the results that the upper extremes of monsoon flows are highly sensitive to increases in temperature and show a continuous decreasing trend. Medium and low flows are increasing in future projections for all the scenarios, and high uncertainty is noticed in the case of low flows. An early CP is detected in the case of high emissions scenarios.

Direct statistical downscaling of monthly streamflow from atmospheric variables in catchments with differing contributions from snowmelt

“…We applied the train() function from the caret package in R (http://topepo.github.io/caret/index.html) to fit linear regression with backward selection, which starts with all predictors in the model, iteratively removes the least contributive predictors until all predictors are statistically significant (Kassambara, 2018). The RVM method was chosen because it was applied in several previous direct downscaling studies and always showed good model performance (Ghosh and Mujumdar, 2008; Okkan and Inan, 2015; Das and Nanduri, 2018). Both models used 10‐fold cross‐validation to estimate the mean squared error and to select the best model based on the smallest error.…”

“…By directly linking GCM outputs and hydrological variables, uncertainty introduced by different RCMs, post‐processing methods and hydrological models can, in principle, be avoided. In addition, such a method is simple and computationally efficient, so it can be easily applied to a full ensemble of GCM outputs to account for the uncertainty contributed by the GCMs (Das and Nanduri, 2018). However, the direct downscaling method has also been criticized for its over‐simplification of the hydrological cycle and its lack of water stores and transfers within the soils and groundwater of a catchment (Xu, 1999).…”

“…Nevertheless, recent studies applying direct downscaling methods using advanced statistical models show reasonable results for monthly streamflow (Noori et al ., 2011; Sachindra et al ., 2013; Tofiq and Guven, 2014; Okkan and Inan, 2015; Sachindra et al ., 2015) and seasonal flows (Landman et al ., 2001; Ghosh and Mujumdar, 2008; Das and Nanduri, 2018) across the world. Some studies have even succeeded in simulating streamflow with a higher temporal resolution, such as daily or 5‐day streamflow (Cannon and Whitfield, 2002; Tisseuil et al ., 2010).…”

“…In general, the training data accounted for 70–80% of the time series and the rest was used as the testing subset (Tisseuil et al ., 2010; Joshi et al ., 2013; Tofiq and Guven, 2014; Okkan and Inan, 2015). Das and Nanduri (2018), however, used 90% of the data as the training subset in order to maximize the number of data available for training the nonlinear models. Detailed information about these studies is listed in Appendix .…”

A Machine Learning Approach for Improving Near-Real-Time Satellite-Based Rainfall Estimates by Integrating Soil Moisture

“…The elevation of the catchment varies from 144 to 1036 m above sea level [53]. Agricultural lands and forests are the major land use over the catchment [54]. Figure 1 represents the location of the catchment in India along with the observed monsoonal average rainfall during the study period.…”