This study makes use of temperature and precipitation from CMIP5 climate model output projections for climate change applications (agricultural application). Bias correction of temperature and precipitation from CMIP5 GCM simulation results with respect to observation is discussed in detail. The non-linear statistical bias correction is a suitable bias correction method for climate change data because it is simple and does not add artificial uncertainties to the impact assessment of climate change scenarios for application studies (agricultural production) in future projections. The simple statistical bias correction uses observational constraints on the GCM baseline, and the projected results are scaled with respect to the changing magnitude in future scenarios, varying from model to model. Two types of bias correction techniques are shown here. (1) A simple bias correction using a percentile-based quantile-mapping algorithm and (2) a simple but improved bias correction method, a cumulative distribution function (CDF; Weibull distribution function) based quantile-mapping algorithm, are done. This study shows that the percentile-based quantile matching method gives results similar to the CDF (Weibull) based quantile matching method. The usefulness of the quantile matching method for climate change application is elucidated with the regression-based agricultural yield projection scenarios using CMIP5 precipitation and temperature data sets.

1. INTRODUCTION

2. DATA SOURCES

2.1 Rainfall data

2.2 Temperature data

2.3 CMIP5 model data set

2.4 Kharif food grain yield data set for India

3. METHODOLOGY

3.1 Bias correction methodology

4. RESEARCH RESULTS

4.1 Percentile-based bias correction method results

4.2 Time series of percentile-based quantile-mapping bias correction

4.3 Weibull distribution function (CDF) based bias correction method results

4.4 Climate change impacts and projection of total food grain production over South

5. CONCLUSIONS