This paper describes a new algorithm to derive the PIA for EarthCARE which improves on the method that has been used operationally for CloudSat. The results appear to convincingly show that the EarthCARE algorithm is superior because of its ability to use an interpolation-based method over much longer distances. This is a really nice advance. I have only minor comments described below. Specifically, I can’t understand some of the discussion of Figure 1 and I think they need to consider that error in PIA should be a function of wind speed at both the calibration and the target point.

-Matt Lebsock-

Specific comments:

Line 42: Add Lebsock and Suzuki 2016 DOI:10.1175/JTECH-D-16-0023.1.

Equation 6: I agree that the fact that there are differences in the equation is an advantage IF the water vapor profile and the surface wind speed are approximately equal at points x and x1. However, in the presence of cloud at point x and clear sky ‘calibration point’ at x1 we don’t expect this to necessarily be true, especially for small scale unresolved by the model fields from which water vapor is derived. For example, Lebsock and Suzuki 2016 show using an LES that water vapor attenuation is larger in the cloudy targets than the clear targets, which makes physical sense.

Line 112: ‘chose’ -> chosen’

Figure 1: I need help with this figure. First, I think you should show another panel with both the ‘model-method’ and ‘interpolation-method’ error plotted as a function of wind speed. Second, I think you should label on the existing panel which region is best for each method for clarity. Third, I can’t quite understand why the interpolation method is better for a much greater distance between x and x1 when the wind speed at x1 drops below about 3 m/s. I actually would expect the opposite – that the interpolation would work better over greater distances for higher wind speeds . Fourth, the residuals should be a function of both the wind speed at x and the wind speed at x1 since they each influence one of the sigma_0_e terms. Can you comment on points 3 and 4?

Line 159: Related to point above about low wind speeds here you say you exclude the low wind speeds from interpolation which is what I would expect. ‘In contrast, the method used here allows interpolation even when the calibration points are ≈200 km to ≈100 km from the cloudy pixel in wind speed conditions between 4 and 15 m/s’.

Line 164: I think you will get an even better uncertainty estimate if you bin by wind speed at both x and x1. ‘Each calibration point used in the PIA estimation is weighted based not only on it’s distance from the point of interest 165 but also on the potential uncertainty associated with wind speed at that location’

Equation 14: Several terms are not defined: lambda, c, tau_p

Lines 288-300: The ‘model’ used in precip-column is actually an empirical look-up-table derived from clear sky observations not the li model. ‘The first approach, referred to as the Wind/SST method, estimates the NRCS at cloudy region in absence of hydrometeor and presence of gaseous attenuation (σ gas 0 ) as a function of surface wind speed and SST using geophysical models (Li et al., 2005)’

I've detailed my comments in the attached pdf.  Although I have only one substantial concern, its resolution might impact the findings and conclusions, so I have classifed this work as requiring major revisions.