Following are specific comments on the CMDL document at http://www.cmdl.noaa.gov/aero/net/c130/indoex/lidar/. Some of the information presented here is redundant with the email I sent out on 6/15 but the arguments are presented in more detail.

I have analyzed four days of data rather than the 10 days presented by CMDL. My comments on these 10 days are as follows:

RF#1 2/16 : also analyzed here

RF#2 2/18 : 180-backscatter data not available below 1800m ; extinction data will be included in the final analysis

RF#5 2/25 : also analyzed here

RF#6 2/27 : Descent through the haze layer was over KCO, not Male. We did a low altitude leg over KCO then ascended only to 1100m before

RF#8 3/04 : There is over an hour between when the C-130 landed and when the lidar measurements started so I have chosen to exclude this profile.

RF#9 3/07 : also analyzed here

RF#10 3/09 : As for RF#6 we descended over KCO for a low-altitude leg. We only ascended to 1600m (and the top of the haze layer was at about

RF#16 3/21 : There was an open line in the impactor box for this flight so I feel the uncertainty in the in-situ measurement of extinction is too uncertain

RF#17 3/24 : 180-backscatter data not available between 2000m and 200m; extinction data will be included in the final analysis.

RF#18 3/25 : also analyzed here

Comments on the enumerated CMDL items:

This is clearly a source of the discrepancy that needs to be investigated. I have made a chart below of the times bracketing the C-130 descent and the timeframe used to derive the IfT lidar profiles; within the given timeframe the lidar measurements are averaged over 1 hour of cloud-free data. The C-130 and lidar measurements overlap for RF#1 on 2/16 and nearly overlap on RF#18 on 3/25. [For the 2/16 case, IfT will need to provide further time specification/explanation, as the given timeframe covers three hours.]

Profile times:

A specific response to the statement: "The aircraft performed a slant path descent, and aircraft generally attempt to avoid clouds on approach. The Hulule lidar was presumably fixed in it's orientation and intercepted clouds, patchy aerosol layers, or anything else directly overhead."

In some cases the C-130 descents do include cloud data. The descents for RF#5 on 2/25 and RF#9 on 3/07 did include cloud legs at 600 and 660m respectively. Because the lidar data does not go below 1000m this will not affect the comparison of extinction but will affect the comparison of 180-backscatter (Bp).

More importantly, it is possible that both profiles could be "contaminated" with sub-visible clouds. I do not know about how reliably various C-130 instruments can identify very sparse droplet populations, but I would imagine it would be possible not to detect their presence. Within these clouds the optically relevant aerosol concentrations could be depressed because they have been sequestered into droplets. Because droplets do not make it into the CAI (we hope!) this would result in a decrease in measured light extinction. On the other hand, cloud contamination would clearly bias the lidar data high.

I am not familiar with the IfT cloud screening procedure so I will leave explanation of their method to them! However, I can contribute some information. The IfT measurements are bracketed by radiosondes which provide profiles of relative humidity. The Raman signal from the lidar can also be used to derive water vapor profiles. These two techniques provide some capability for cloud detection. Of course the sondes are not perfectly coincident in time/space with the lidar profiles (see chart for times of sonde launches) and they may miss intermittent/broken clouds. If this is the case, the discrepancy should be largest in regions of higher relative humidity where intermittent clouds are more likely to pop up. I have plotted the ratio of the lidar-derived extinction and 180-backscatter to the same as derived on the C-130 and plotted these ratios versus RH. (Figure 14) The discrepancy does not increase at high relative humidities, indicating that cloud contamination is probably not the source of the discrepancy. (In fact the ratio is largest at low relative humidities, but this is because the very low relative humidities correspond to high altitudes where the lidar still shows a signal from aerosol light scattering while the C-130 is measuring almost zero).

Finally, if there are indeed "patchy aerosol layers or anything else" that the lidar was measuring, these are real sources of radiative forcing. If this is the source of the discrepancy and it results in higher values of light extinction then the lidar is doing a better job of measuring what is actually out there than the in-situ instruments on the C-130.

R.H. and atmospheric variability:

One indicator of atmospheric variability is relative humidity. Variations in R.H. can also lead directly to large changes in extinction and Bp, especially at very high R.H. Therefore I felt it would be informative to compare the R.H. profiles associated with each measurement. The R.H. profiles associated with in-situ data are those measured from the C-130; those associated with the lidar data are from the IfT sondes. I have included in the time chart below the time(s) of the IfT sonde launches that are in closest proximity to their lidar measurements.

Only the data from 2/25 (RF #5) show an overall change in the R.H. profile between the C-130 and lidar measurements. There are smaller variations in the other plots, though, and whenever there are differences in the R.H. and the R.H. is very high (i.e. R.H. > 85-90%) hydroscopic growth may account for large changes in ambient extinction and/or Bp.

The profile where the R.H. is most similar for the two measurements is RF#18. In this case the R.H. profiles are nearly identical, yet there is still greater than a factor of two discrepancy between the lidar and C-130 derived values of extinction and Bp. In this case it is clear that variations in ambient R.H. cannot explain the discrepancy.

Time-variability of Bp and extinction:

Extinction can only be derived from the IfT lidar after sunset, but profiles of 180-backscatter can be retrieved during daylight as well. Detlef and I have agreed to use two approaches to investigate whether atmospheric variability is the source of the discrepancy:

1. IfT has time-height profiles of Bp that span the times between the C-130 descent profile and their nighttime measurements. Inspection of these will show us how much temporal variation there was in Bp over Male for each evening. Translation of invariance in Bp to invariance in extinction of course requires an assumption that the ratio of extinction to Bp (the lidar ratio) does not vary appreciably. However, if the extensive property (Bp) does not vary it is unlikely that there are large enough changes in the lidar ratio to make up for the observed difference. Also, our measurements of the lidar ratio from the C-130 throughout the INDOEX campaign indicate that, especially above 1000m, the lidar ratio was generally very high and rarely was outside of the range of 70-90, which does not allow for a factor of 2-4 discrepancy.
2. We plan to do a direct comparison of Bp at the time of the C-130 descent. Because both Bp and extinction are systematically higher for the lidar than for the C-130

IfT is working on providing the data needed to complete these two test.

I have nothing to add on this subject.

The 180-backscatter nephelometer was drawing aeorosl directly from the aircraft CAI. No impactor was used, so everything that made it into the aircraft shoud have made it to this nephelometer.

In some cases the ratio b180_lidar/b180_C-130 is less than extinc_lidar/extinc_C-130. It seems likely that this is because the 180 neph is measuring the full aerosol, whereas the CMDL system was only measuring aerosol with a dry diameter <1micron. However, b180_lidar/b180_C-130 is almost always quite a bit greater than one, so as John said the 1um cut size probably accounts for some but nowhere near all of the discrepancy.

The 180-backscatter data also show a large discrepancy and there was no heater at the inlet of the 180-backscatter nephelometer, confirming the CMDL assertion that volatilization is not likely to be the cause of the discrepancy.

When volatilization did occur in the CMDL system there was a sharp drop in scattering. For the profiles I have shown, the 180-backscatter data and extinction data have similar shapes so I agree that the data shown have been properly filtered for volatilization.

Again, the 180-backscatter data also show a large discrepancy and the 180-backscatter nephelometer did not have the problem with leaks suffered by the CMDL system. Therefore I agree that it is unlikely that their leaks are the cause of the discrepancy.

All of the extinction data I show are from the final corrected/adjusted CMDL data set. However, I originally received the raw CMDL data set which did not have any corrections applied. I processed all of their data independently, performing corrections for angular truncation, etc. The only correction they applied to their final data set which I did not apply was for leaks. When CMDL published their corrected data set I compared my version of their data to their version. The two data sets agree quite well when I take into account their leak correction. This also confirms the CMDL statement that it is unlikely that there is a fundamental flaw to their data processing that is biasing all of their data low. Finally, there is once again the fact that the Bp data are also biased low relative to the lidar data. These data have been processed independent of the CMDL data.

As mentioned above I have calculated light extinction at 532nm and Bp is measured at 532nm. The wavelength difference does not compensate for the discrepancy.

E) Comparison with aerosol optical depth profiles measured on the C-130

In addition to the radiometer on board the C-130 there were also sun photometers measuring total optical depth from the ground at Male. Detlef has provided me with the a.o.d.’s for the four dates I’ve presented. Looking at the chart below you can see that these a.o.d.’s were measured at times closer to when the C-130 landed than when the lidar profiles were taken.

While we do not have full profiles of light extinction from either the C-130 or the lidar for the cases presented here I have done some calculations to try and compare the three measurements of aerosol optical depth for three of the profiles. This at least gives some indication of which method is in better agreement with the sun photometer data.

RF#1 on 2/16

AOD altitude range:

C-130: 0.15-0.18 320m - 3820m

lidar: 0.27 1020m - 5000m

sun photom.: 0.37 total column

Notes: C-130 AOD range is due to assuming ssa of between 0.75 and 0.9

RF#5 on 2/25

AOD altitude range:

C-130: 0.21 420m - 2870m

lidar: 0.18 970m - 3720m

sun photom.: 0.36 total column

Notes: C-130 AOD assumes a linear increase in extinction between 570 and 870m.

RF#18 on 3/25

AOD altitude range:

C-130: 0.26 350m - 3550m

lidar: 0.45 1020m - 5000m

sun photom.: 0.63 total column

Notes:

C-130: Assumes light extinction drops of linearly between 2500 and 3500 meters (i.e. assuming the "shape" of the lidar profile is representative).

lidar: Using time-height plots of 180-backscatter IfT was able to see that there was an increase in extinction of about 20% between the 1700 and 1900m altitude that would account for the difference between the lidar and sun photometer measurements.

The sun photometer-derived AODs are quite a bit higher than the C-130-derived AODs. While the full aerosol layer is not encompassed by the C-130 measurements there would have to be unrealistically high values of light extinction at the missing altitudes to make up for the difference. On the other hand, the lidar data consistently gives higher values of light extinction than the C-130 data and the lower 1000m of light extinction could account for the discrepancy between the lidar and sun photometer AODs.

Finally, I have some further comments on humidification issues. CMDL measures light scattering by drying the incoming aerosol, measuring light scattering at low (40-45%) R.H., then humidifying the aerosol and measuring light scattering at high (80-85%) R.H. From these two values we can interpolate scattering at the ambient R.H. However, it is possible that there is a fundamental error in the humidification approach to determining f(RH). If the aerosol is not capable of fully rehydrating in the short residence time through the humidification system there will be a systematic underestimation of light scattering. This "kinetic effect" could be responsible for some of the discrepancy, especially at higher relative humidities. The argument against this, of course, is the statement I made above that the discrepancy does not appear to be correlated with R.H.! Another argument against it is that the 180 nephelometer does not have a humidification system and so should not be subject to the same errors. Where the measurement R.H. and ambient R.H. are the same the measured scattering should be the same. However, it is very possible that the aerosol goes through heating between the inlet to the aircraft and the inlet to the 180-backscatter nephelometer. The piping leading from the aircraft wall to the instrument inlet will change temperature more rapidly than the instrument itself and so may have warmed up while the nephelometer did not. Ram heating, as the CMDL group pointed out, is also a possibility.

Light absorption is measured by CMDL at low R.H. Some people have argued that light absorption could be magnified considerably in humidified aerosol where the light absorbing component is in the core. This magnification effect is due to multiple scattering within the aerosol. Kind of out there, I know, but it could mean some underestimation of light extinction. Again, however, this should be most pronounced in regions of high R.H.

It is very possible the the discrepancy is due to a combination of factors, and both systems may have errors that happen to be biasing them in opposite directions. The focus of the CMDL and my analyses have been largely on our systems, not the lidar system. I have asked Detlef for comments regarding possible sources of error in the lidar measurements and I believe he is working on that.