How is the monsoon this year?
Some statistical analysis seems to be possible. Averaged rainfall over India (above graph) seems to allow easily 20-30% daily variation over the mean. In other words, if you expected 10 mm of rain and got 12 mm instead, or even 8 mm, it would still be considered normal.
Also, spates of higher than average or lower than average rainfall seem to come in periods of 5-7 days together: this is called an auto-correlation time in statistics. The monsoon season length is then 20-24 auto-correlation times.
Hence the normal seasonal fluctuation should be (20-30%)/square root of (20-24), ie, about 3-6%. In other words, 3 years out of 10 you would expect the rainfall to be only about 95% of normal. In another 2 years out of 10 you would expect the rainfall to be about 90% of normal. Similarly, about 2 years out of 10 the rainfall should be about 110% of normal.
Those who deal with these things regularly will notice that I have used “normal statistics” (also called the bell-curve or Gaussian) for the seasonal rainfall. This may be completely wrong for the statistics of rainfall in one day in one place, but is very likely to be right for the seasonal rainfall averaged over the whole country. This assertion is called the central limit theorem.
