You can't. See:
x1 = [1 2 4; 1 1 1; 3 2 2; 0 0 0]
[y1,ps] = removerows(x1,'ind',[2 4])
% Reverse the processing of y1 to get x1 again.
x1_again = removerows('reverse',y1,ps)
Even the help says:
In the reverse calculation, the unknown values of replaced rows are represented with NaN values.
If you want rain, then just keep it in scope. If, for some reason, you want to multiply high rain values by some factor, you can use masking:
rain = 5 * rand(10, 1)
mask = (rain>=1.0) %Index of rainfall more than 1mm/h
rain(mask) = rain(mask) * 2.5
