If Rain Falls in India and No One Reports It, Are Historical Trends in Monsoon Extremes Biased?

With two thirds of the total Indian population employed by the agriculture sector, changes in Indian monsoon precipitation have widespread implications for human welfare. Increased extreme precipitation since 1950 has been widely reported for central India. Major studies have relied upon the gridded daily precipitation observations provided by the India Meteorological Department (IMD), which assimilate observations from a variable network of weather stations. Replicating the IMD's interpolation method on satellite‐based precipitation observations, however, indicates that temporal changes in the observing weather station network introduce a jump in the record toward more extreme rainfall after 1975. Trends evaluated across this jump are suspect, and trends evaluated subsequent to it are insignificant (p > 0.1). This positive bias may also mask declines in average monsoon rainfall. Greater accuracy in these trends may resolve distinctions between climate model simulations of future changes. Access to the underlying data from IMD rain‐gauges would facilitate improved rainfall reconstruction. Plain Language Summary Previously reported trends in daily monsoon rainfall since 1950 have been estimated using interpolated weather station observations released by the India Meteorological Department. The number of reporting weather stations changes over time, and poor coverage by weather stations can overlook extreme rainfall events. By applying the interpolation of this changing network to satellite‐based rainfall data, we show that the changing coverage of weather stations in the Indian rainfall data leads to spurious increases in extreme rainfall. This suggests that previously reported trends of extreme rainfall are biased positive. Access to the raw weather station data would improve our ability to track changes in the Indian monsoon and assess modeled predictions given climate change. Key Points A method is introduced to diagnose how changes in gauge locations bias rainfall estimates Central Indian rainfall estimates from 1951 to 2016 are biased toward increasing extreme rainfall