Correcting negatively biased refractivity below ducts in GNSS radio occultation: an optimal estimation approach towards improving planetary boundary layer (PBL) characterization

Abstract

Abstract. Global Navigation Satellite System (GNSS) radio occultation (RO) measurements are promising in sensing the vertical structure of the Earth's planetary boundary layer (PBL). However, large refractivity changes near the top of PBL can cause ducting and lead to a negative bias in the retrieved refractivity within the PBL (below ∼ 2 km). To remove the bias, a reconstruction method with assumption of linear structure inside the ducting layer models has been proposed by Xie et al. (2006). While the negative bias can be reduced drastically as demonstrated in the simulation, the lack of high-quality surface refractivity constraint makes its application to real RO data difficult. In this paper, we use the widely available precipitable water (PW) satellite observation as the external constraint for the bias correction. A new framework is proposed to incorporate optimization into the RO reconstruction retrievals in the presence of ducting conditions. The new method uses optimal estimation to select the best refractivity solution whose PW and PBL height best match the externally retrieved PW and the known a priori states, respectively. The near-coincident PW retrievals from AMSR-E microwave radiometer instruments are used as an external observational constraint. This new reconstruction method is tested on both the simulated GNSS-RO profiles and the actual GNSS-RO data. Our results show that the proposed method can greatly reduce the negative refractivity bias when compared to the traditional Abel inversion.

GNSS-RO Refractivity Bias Correction Under Ducting Layer Using Surface-Reflection Signal

“…Given an Abel-inverse refractivity profile, one can analytically derive the "family" of the candidate profiles N f (h, dx) [16,18], as shown in Appendix A, where dx = x m − x b . While, in Wang et al [23], two independent variables were used to account for the uncertainty of x b , here, we assume the x b a priori is accurate enough and only one independent variable x m is necessary to simplify the process. Since the measurement error of x b is usually small and unbiased relative to the a priori profiles [23], the results are not sensitive to this simplification.…”

“…While, in Wang et al [23], two independent variables were used to account for the uncertainty of x b , here, we assume the x b a priori is accurate enough and only one independent variable x m is necessary to simplify the process. Since the measurement error of x b is usually small and unbiased relative to the a priori profiles [23], the results are not sensitive to this simplification. The family of refractivity profiles derived from the Abel-inversion results of Equation 4, with respect to different x m , is shown in Figure 6a.…”

“…Unlike the approach proposed by Wang et al [23], which used optimal estimation to estimate both dx and x b simultaneously, here, dx is the only variable to be estimated which does not have an a priori; hence, a simple non-linear least squares can fit the needs here. The reflected bending angle of the candidate profile α C R for each iteration is calculated by the forward model described in Section 3.1.…”

“…In practice, however, the lowest point of GNSS-RO bending angle profiles do not often reach the surface due to measurement noise and tracking error [10], which makes the application of this constraint difficult, in practice. As an alternative constraint, Wang et al [23] proposed distinguishing different profiles in the continuum with the integrated precipitable water (PW) and using the collocated external PW retrievals from other instruments as the constraint. This approach requires collocated observations, which are not always available and could induce the collocation error in the N-bias correction process.…”

Assessment of global navigation satellite system (GNSS) radio occultation refractivity under heavy precipitation

“…This bias is not assessed here, since it has already been discussed previously in other studies (e.g., Ao et al, 2003;Sokolovskiy, 2003;Xie et al, 2006Xie et al, , 2012Wang et al, 2017). Similarly, other potential sources of bias have been checked, for example, the angle of incidence of the occultation ray to the receiver, with respect to the transmitter position.…”