Issue 5:  Impact of assumptions on both refractivity and retrieved profiles

How large is the uncertainty, in both the raw refractivity and the retrieved profiles, of assumptions related to: ionspheric structure, first-guess temperature or humidity profile, and other issues we are not aware of?  Is there any reason to believe that these sources of uncertainty will ever be reduced or eliminated?

Response from Kevin Trenberth:
The ionospheric structure issue is commented on above as an inherent limitation.  Given refractivity, it is not possible to get temperature or humidity alone without assumptions about the other.  In some parts of the atmosphere (8 to 22 km) an assumed water vapor produces very tiny errors in T, but elsewhere these are more important. 

 Back to GPSRO Home Page

  • No labels

4 Comments

  1. Tom Yunck

    A comprehensive update of the seminal Kursinski 1996 error analysis is certainly in order. Nice little thesis topic.

  2. Tony Mannucci

    These sources of uncertainty need to be addressed. I believe that the nature of the retrieval method and the measurement physics will permit us to address these sources of uncertainty sufficient to establish a useful benchmark. Our primary task right now is to understand in greater detail how the assumptions affect accuracy. A great deal of work has already been done in addressing the sources of uncertainty. More work is needed because the measurement requirements for climate are so stringent.

  3. Seth Gutman

    Consolidated Comments from Gutman, Yoe and Reale

    Seth Gutman's Response
    I don't think there's any reason to think that the accuracy and precision of a GPS-RO measurement in the ionosphere should be substantially different than the estimates SWPC and others (e.g. JPL) currently make from ground-based observations, but the "point of view" is substantially different. When it comes to estimating TEC, no first guess is required in the traditional sense since ionospheric delay is proportional to TEC/frequency**2. In practice, the range measurement provides the "absolute" magnitude of the delay and the carrier phase observation provides the precision. That would put the accuracy and precision of an ionospheric sounding at about 2 TEC +/- ~0.1 TEC units under most circumstances. The neutral atmosphere is a totally different matter because temperature, pressure and humidity retrievals require foreknowledge or constraint on the other two-of-three refractivity parameters. My understanding is that refractivity and bending angle are usually calculated using a forward 1-D variational model, but I'm not sure about the sensitivity of the model to errors in the first guess. Lidia Cucurull at JCSDA would have more information about that. With regard to reducing uncertainty, assuming that GPS-RO has small systematic errors I would think that as more GPS and other relevant observations are assimilated into global variational models, the errors in the model backgrounds will gradually decrease, and the accuracy of the first-guess should improve accordingly, but it's an iterative process.

    Tony Reale's Response
    Collecting respective "retrieval" data streams (assuming tropospheric retrieval is relevant to climate) covering the various assumptions (refractivity calculation, retrieval (1dvar), apriori, etc) in an NPIVS environment could respectively quantify their impacts on retrievals.

  4. Jens Wickert

    Response from Jens Wickert

    Influence of assumptions on temperature should be larger compared to refractivity.

    Influence of ionosphere is already in the refractivity at higher altitudes. This influence can be further reduced in future by better receivers and antennas (less noise), use of e.g. 3rd carrier frequency of the new navigation satellites or by using information on the ionospheric state. Currently we use a quite straightforward technique (linear combination of bending angles at 2 frequencies) without information on the ionosphere, but this works already quite effective (but definitely can be improved).