Does wind speed correction factor depend on unorm or resid?

[jedokaplan]

According to Sommer and Kaplan (2017), eqns 21-23 and text on page 3784 the wind speed correction factor depends on the uniform random number used for windspeed. In the code, however, instead of using unorm(4), resid(4) is used.

gwgen_f90/weathergenmod.f90

Lines 555 to 572 in 9b11a3f

	! ---- wind bias correction
	if (wind_slope_bias_L > 0.0) then
	slopecorr = wind_slope_bias_L / ( 1 + exp( - wind_slope_bias_k * ( &
	resid(4) - wind_slope_bias_x0)))
	else
	slopecorr = sum(wind_bias_coeffs(:) * ( &
	max(wind_bias_min, min(wind_bias_max, resid(4))) ** (/ 0, 1, 2, 3, 4, 5 /)))
	end if
	if (abs(wind_intercept_bias_a + 9999.) > 1e-7) then
	intercept_corr = exp(wind_intercept_bias_b + &
	wind_intercept_bias_a * max(wind_bias_min, min(wind_bias_max, resid(4))))
	else
	intercept_corr = sum(wind_intercept_bias_coeffs(:) * ( &
	max(wind_bias_min, min(wind_bias_max, resid(4))) ** (/ 0, 1, 2, 3, 4, 5 /)))
	end if
	wind = (wind - intercept_corr) / max(slopecorr, 9e-4)

Which one of these is correct? Using resid(4) seems to generate unrealistically high values for corrected windspeed.

[Chilipp]

Yes, you are definitely right!