Downscaling GCM data for climate change impact assessments on rainfall : a practical application for the Brahmani-Baitarani river basin

Downscaling GCM data for climate change impact assessments on rainfall: a practical application for the Brahmani-Baitarani river basin R. J. Dahm, U. K. Singh, M. Lal, M. Marchand, F. C. Sperna Weiland, S. K. Singh, and M. P. Singh Deltares, Boussinesqweg 1, P.O. Box 177, 2600 MH Delft, the Netherlands RMSI, A-8 Sector 16, NOIDA 201 30, India Central Water Commission, R.K. Puram, New Delhi 110 066, India

downscaling and bias-correction methods available to generate local climate projections that 10 also consider changes in rainfall extremes. Yet the applicability of these methods highly 11 depends on availability of meteorological observations at local scale. In this study, data from 12 a limited number of rain gauges with incomplete time-series and a lower resolution gridded 13 rainfall productthe APHRODITE dataset (Yatagai et al., 2012) -could be accessed and 14 used. Due to this data quality issue we concentrate on two widely used and easy to implement 15 bias correction procedures -Linear Scaling (LS) and Delta Change (DC). We evaluate the 16 influence of these procedures on resulting changes in rainfall indices. Instead of using a large 17 ensemble of GCMs we focus on three models that nearly span the full range of annual mean 18 rainfall projections available for India -analysing in more detail the locally relevant biases 19 within these three GCMs and the changes they project. 20 This paper is organized as follows. Section 2 presents an overview of the selected GCMs, 21 reference dataset, and bias correction methods. This is followed by a short description of the 22 impact analysis framework applied and the study area. In section 3 we report the main 23 findings of our study and present the range of rainfall indices under climate change. Here, we 24 are specifically concerned with an inter-and intra-comparison of the GCMs and downscaling 25 methods. Next the limitations experienced for bias-correction techniques experienced in this 26 study are discussed. Section 5 summarizes our main conclusions and provide outlook for 27 future work and practical applications in the study area. 28 29 Therefore, we restrict ourselves to two simple methods (1) Table 4 shows the classification of seasons according to IMD. 17

Study area 18
This study focusses on the Brahmani-Baitarani river basin located in the eastern part of India. 19 The The basin is an inter-state basin and spreads across the states of Chhattisgarh, Jharkhand, and 28 Odisha. The elevation ranges from >750 meter in the north-western part of the basin to 29 approximately 10 meter in the delta region. The Baitarani River enters the Brahmani River in 30 the deltaic region before it drains into the Bay of Bengal. The catchment area is 51,822 km 2 . 1 The region is characterized by a sub-tropical monsoon climate zone with mean annual rainfall 2 of approximately 1450 mm (CWC, 2011) most of which occurs during the SW monsoon 3 season (June to September). 4 The basin is exposed to orographic effects with a positive elevation-rainfall depth slope 5 during monsoon season and an inverse slope in the post-monsoon season when cyclones hit 6 the coast of Odisha (Deltares, 2015). Due to the size and shape, the Brahmani-Baitarani river 7 basin is captured by approximately one GCM grid cell in the deltaic region and one grid cell 8 in the Mountainous region. This allows studying possible orographic effects present in the 9 future projections (Prokop and Walanus, 2013). GCM and climatological observed rainfall 10 values were taken from the grid cell with center coordinates closest to these regions. It was 11 decided not to resample the climatological observed and GCM datasets to one common grid, 12 since up-scaling or spatial-averaging would result in spatial smoothening of rainfall extremes. 13 Figure 2 shows the locations of the grid cells used. We will focus our discussion of the results on the SW monsoon season only as mainly 18 changes in monsoon rainfall will affect water resources and flood risk management in the 19 Brahmani-Baitarani river basin. The figures contain results for all seasons. 20

Comparison of climatological and gauge rainfall data 21
The climatological dataset covers the full GCM baseline period whereas the gauge 22 observations are more recent and only available for a limited time-spam, they can therefore 23 not be used as baseline reference. We here verify the quality of the climatological dataset 24 compared to gauge observations for the overlapping period 1990-2007. Overall the 25 climatological dataset resembles the CWC rain gauges quite well (see figure 3). per year (a frequency of less than 1 and changes therein will affect the analysis of extreme 2 value return periods). This underestimation is most likely caused by the smoothening 3 introduced when interpolating point observations to the grid. Differences in rainfall extremes 4 between the two datasets can clearly be seen from their respective CDFs (see figure 4). Still, 5 based on (1) this analysis, (2) the period covered by the climatological dataset and (3) the 6 work of Rajeevan and Bhate (2008) we conclude that the dataset is suitable as a gridded 7 reference dataset for the Brahmani-Baitarani river basin. When quantifying the impact of 8 climate change to strategic flood management studies in practice, the rainfall deviations 9 between climatological and gauge extremes should be considered. 10 long-year observations. Their conclusion that MIROC-ESM is 'quite wet' does not hold 28 throughout the year for the baseline period for the Brahmani-Baitarani river basin. MIROC-29 ESM underestimates rainfall in February-May in the delta region and during the monsoon in 30 the Mountainous region. GFDL-CM3 overestimates rainfall during the monsoon season in the 1 deltaic region but overall performs best. 2 3

Impact of downscaling method on future GCM projections 4
We made a modification to the LS method for the month June as the underestimation of 5 monsoon rainfall in June by HadGEM2-ES and MIROC-ESM would result in extreme large 6 correction factors, unrealistic rainfall amounts (up to 900 mm/day) and consequently an 7 unrealistic large increase in heavy rain events occurring in June when applying the LS 8 method, see Table 6. Therefore, we decided to apply the July multiplier also for June. By 9 doing so, the GCM total monsoon rainfall amount of MIROC-ESM and HadGEM2-ES 10 become closer to the climatological dataset. Only for GFDL-CM3 total monsoon rainfall 11 becomes slightly lower than that of climatological values. By applying the multiplier of July 12 in June the annual mean baseline rainfall amounts will be somewhat different between the LS 13 and DC method. Although the underestimation of rainfall by MIROC-ESM for the winter 14 season in the Delta region leads to high multipliers as well, we decided not to apply a tailored 15 correction here as it would only be applicable for one of the GCMs. For the LS method we conclude that for a reliable projection of change in WD, the WD in the 23 GCM baseline period should be close to observed, otherwise the correction method and bias 24 from observations will disturb the change signal too much. For GFDL-CM3 we have seen that 25 the number of WD during the baseline period is 103, which already nearly corresponds to the 26 entire Monsoon season which explains the small increase found for this GCM. With the DC 27 method the WD only increase based on the monthly linear increase in total precipitation. 28 amounts for both the Delta and Mountainous regions (see Figure 11). In general changes 15 derived after applying the DC method are more modest as a result of the unchanged day-to-16 day variability -changes will most likely be within the present day climate variability. 17 Unfortunately, we find unrealistic high rainfall extremes for MIROC-ESM due to the high 18 February-May multipliers that correct for the underestimation of winter and pre-monsoon 19 rainfall in the Delta region. The possible increases in return periods for the delta region, 20 obtained after applying the LS method to GFDL-CM3, is inconsistent with all other 21 projections and is likely a result of correction multipliers below one that also decrease the 22 extremes. The overall reduction in return times of annual extremes of daily rainfall for the late 23 21 st century, which we find here, was also found by Ramesh and Goswami (2014). 24 The return period analysis once more illustrates the importance of focussing on relative 25 changes. The precipitation amounts for a 100-year return period of a one-day rainfall event for 26 the period 1990-2013 are 378 mm, 328 mm, and 229 mm for Akhuapada, Anandpur and Tilga 27 rainfall gauge respectively, whereas the climatological data and the baseline rainfall for the 28 LS method applied to all GCMs and regions show intensities of 100-200 mm for the same 29 return period. Therefore, translating absolute future climate results of extreme rainfall events 30 15 based on the climatological dataset to rainfall intensities recognizable for local authorities 1 familiar with the rainfall gauges remains a challenge. We introduced an additional correction to the LS method by applying the multiplier for July 27 to the data of June; this is to avoid the occurrence of extreme high rainfall amounts in the 28 corrected data for June by compensating for the variable monsoon onset date. This introduces 29 a deviation in the corrected baseline GCM data from observations and rainfall amounts for 30 June in both the corrected baseline and also the future climate GCM data and will remain too 31 low. In future work the selection of GCMs could be extended with more criteria, focussing 32 on Coupled Modelling, which is responsible for CMIP, and we thank the climate modelling 7 groups (listed in Table 1 Table 1). By

19
The BC ratio was calculated using the discounted annual benefits or present value (PV  longer than two weeks after which maximum damages occur.

28
-Not only the gaps in the current embankments need to be closed. In fact several river 29 branches run through the area which either have to be embanked as well, or closed off 30 from the main river. But up to 200 km extra embankments the BC ratio remains above -The assumption that there will be no future changes in land use is not realistic because 1 improving the protection level would attract more investments (leading to higher 2 agricultural yields for instance), which would increase the benefits of protection.
3 -The embankment costs should also take into account land acquisition, opportunity 4 costs, sluices for drainage, reduced preparedness of the people for larger floods, the 5 loss in soil fertility and aquifer recharge, etc.
6 Therefore, a more detailed cost-benefit analysis should take into account the whole range of 7 benefits and dis-benefits of embankment construction and should also compare it with other 8 options for reducing the flood damage in the area. Furthermore, attention should be given to 9 the distribution of costs and benefits among the different stakeholders, which would broaden 10 the current macro-economic scope of the analysis.

11
The advantage of the used method it that it can easily simulate future conditions, for instance 12 based on downscaled projections of climate change models for the river catchment. This