The Microwave Radiometer Profiler (MWRP) is similar to the Microwave Radiometer (MWR), except that it creates a profile of the atmosphere in addition to the time series data. The MWRP is capable of acquiring brightness temperatures at several elevation angles, but the current configuration at ENA and AMF1 only uses the zenith value of 90 degrees. At NSA, several elevation angles are used. The MWRP provides vertical profiles of temperature, humidity, and cloud LWC as a function of height or pressure. This occurs at approximately 5 minute intervals, for nearly every weather condition.
The MWRP operates with 12 different Temperature Brightness (TB) channels from 22.23 GHz to 58.8 GHz. It uses the derived temperatures and other measurements to calculate the temperature, humidity, and cloud LWC profiles. This is different than the MWR, which only uses two channels (23.8 & 31.4 GHz) and does not create a profile of the atmosphere.
Question: What are the primary retrieved variables?
Answer: Primary retrieved variables are temperature and humidity profiles as well as PWV and LWP, which are derived by relating the observed radiances to atmospheric water vapor and liquid water. For humidity profiles, the retrieved quantity is water vapor density. Water vapor density is converted to RH by using the retrieved temperature profile. NOTE: The MWRP does not measure RH, it is retrieved with a statistical retrieval. The retrieval does not actually retrieve RH, but absolute humidity, which is then converted to RH using the retrieved temperature profile. Therefore, errors in the absolute humidity and temperature propagates to the RH profile.
Q: How is the retrieval implemented?
A: The profile retrieval is a statistical retrieval and the coefficients are retrieved for each layer. PWV and LWP are retrieved using statistical coefficients derived using the same technique.
Q: Have you separately validated the temperature profiles versus the sonde t_dry?
A: Yes. Temperature retrievals are usually accurate in 1-2 K. Right now there is a bias issue related to modeling of the oxygen absorption. RMS errors in the temperature retrieval vary between 1K to about 3K in the upper layers.
Q: Does the temperature profile show better or worse agreement than the RH profile?
A: Usually temperature profiles are in better agreement. The retrieval of temperature is more linear (therefore a statistical linear retrieval is more appropriate). In addition, the profile of temperature is usually less variable. Apart from inversions you may have, the temperature profile is usually more constrained by adiabatic lapse rate, etc.
Q: Why does the RH occasionally exceed 100% ?
A: One issue with the statistical retrieval is that is difficult to constrain physically because it is essentially a linear regression. The only way to overcome the RH > 100% is to set RH = 100 after the retrieval is performed. This issue is a little similar to the LWP < 0.0. The retrieval has an error margin, so results vary in between + and - that error limit (but this is true of any retrieval). The errors in the retrievals are due to the following factors: Errors in the measurements, errors in the retrieval process, and errors in the forward model. In addition, when we convert using temperature, small errors in the temperature retrieval will propagate to the RH profile. So in the end, we have RH = 100% + or - 10% (it may be up to 20%). If you have, let's say, RH = 105% this is still in the error margin and can be considered 100%. Now if we had something like RH = 130%, that is an indication of a wrong retrieval that can be due to things such as wrong BT measurements, or a profile that is very different from the climatology.
Q: Are there comparisons routinely made between the MWR PWV and the MWRP totalPrecipitableWater values?
A: Yes, all quantities that can be compared between the two instruments are routinely plotted and compared. This includes brightness temperatures, retrieved PWV, LWP, and temperature profiles with sonde profiles.
Q: I expected that the totalPrecipitableWater would have agreed with the integrated waterVaporDensity profile, but while the values are similar, I do not see rigorous agreement. Can you explain?
A: Yes. The PWV is retrieved by directly retrieving PWV and LWP with a linear statistical retrieval of the two quantities. Only the first 6 channels (water vapor channels) are used in this retrieval. The water vapor density and temperature profiles are retrieved using all 12 channels. Even if we were using the same channels for both retrievals, you would still not see exactly the same retrieved amount. This is due to the fact that, when you retrieve the full profile (versus the PWV only), your retrieval has a larger noise because you have uncertainties associated with the retrieval of each layer. What you will see if you plot the PWV from integrated profiles is that the ensemble will have a larger standard deviation. In essence, the process of retrieving the full profile and then integrating it introduces more noise in comparison with the direct retrieval of PWV.