The observed summertime drying over Northern Central India (NCI) during the latter half of the twentieth century is not reproduced by the Coupled Model Intercomparison Project Phase 5 (CMIP5) model ensemble average. At the same time, the spread between precipitation trends from individual model realizations is large, indicating that internal variability potentially plays an important role in explaining the observed trend. Here we show that the drying is indeed related to the observed 1950–1999 positive trend of the Pacific Decadal Oscillation (PDO) index and that the relationship is even stronger for a simpler index (S1). Adjusting the CMIP5-simulated precipitation trends to account for the difference between the observed and simulated S1 trend increases the original multimodel average NCI drying trend from −0.09 ± 0.31 mm d−1 (50 years)−1 to −0.54 ± 0.40 mm d−1 (50 years)−1. Thus, our estimate of the 1950–1999 NCI drying associated with Pacific decadal variability is of similar magnitude as our previous CMIP5-based estimate of the drying due to anthropogenic aerosol. The drying (moistening) associated with increasing (decreasing) S1 can partially be attributed to a southeastward (northwestward) shift of the boundary between ascent and descent affecting NCI. This shift of the ascent region strongly affects NCI but not Southeast Asia and south China. The average spread between individual model realizations is only slightly reduced when adjusting for S1 as smaller-scale variability also plays an important role.

The observed summertime drying over Northern Central India (NCI) during the latter half of the twentieth century is not reproduced by the Coupled Model Intercomparison Project Phase 5 (CMIP5) model ensemble average. At the same time, the spread between precipitation trends from individual model realizations is large, indicating that internal variability potentially plays an important role in explaining the observed trend. Here we show that the drying is indeed related to the observed 1950–1999 positive trend of the Pacific Decadal Oscillation (PDO) index and that the relationship is even stronger for a simpler index (S1). Adjusting the CMIP5-simulated precipitation trends to account for the difference between the observed and simulated S1 trend increases the original multimodel average NCI drying trend from −0.09 ± 0.31 mm d−1 (50 years)−1 to −0.54 ± 0.40 mm d−1 (50 years)−1. Thus, our estimate of the 1950–1999 NCI drying associated with Pacific decadal variability is of similar magnitude as our previous CMIP5-based estimate of the drying due to anthropogenic aerosol. The drying (moistening) associated with increasing (decreasing) S1 can partially be attributed to a southeastward (northwestward) shift of the boundary between ascent and descent affecting NCI. This shift of the ascent region strongly affects NCI but not Southeast Asia and south China. The average spread between individual model realizations is only slightly reduced when adjusting for S1 as smaller-scale variability also plays an important role.