Reading notes-11

Tropomi will feature a wide swath of ~ 2600 km with a ground pixel area of 7×7 km2 （for the sub-satellite point）mapping into daily global coverage with ~7×10E6 exposures per day. The spectrometer will cover various spectral ranges.

Here, we suggest to use a combination of the SWIR and the NIR channels for the simultaneous retrieval of CH4 total column concentrations and atmospheric scattering properties to minimise scattering induced retrieval errors on CH4. The SWIR channel provides information on the CH4 concentrations while the combination of the SWIR and NIR channels constrains the atmospheric scattering properties. The scattering properties to be retrieved are parameterisations of particle amount, height distribution, and particle microphysical properties largely following the ‘physics based’ approach we have developed for CH4 and/or CO2 retrievals from GOSAT and the Orbiting Carbon Observatory (OCO). We aim at showing that TROPOMI will be able to carry forward highly accurate monitoring of atmospheric CH4 concentrations currently undertaken by SCIAMACHY and GOSAT.

We first introduce our methodology(Section 2), in particular covering the instrument design (Section 2.1), the setup of the retrieval method(Section 2.2) and the generation of the trial ensemble(Section 2.3). We then discuss retrieval performance for the trial ensemble(Section 3) and discuss implications(Section 4).

TROPOMI will measure radiance spectra of sunlight backscattered by the Earth’s surface and atmosphere in various spectral channels ranging from the UV to the SWIR. Here, we focus on the NIR(710-775nm) and SWIR(2305-2385nm) channels exhibiting a spectral resolution of 0.5nm and 0.25nm, respectively. The instrument model assumes a Gaussian instrument function with the respective full-width-at-half-maximum (FWHM) sampled by 3(NIR) and 2.5(SWIR) detector pixels. Following the planning for instrument design, simulated measurement noise is pure shot-noise in the NIR and shot-noise plus a signal independent term in the SWIR. The corresponding continuum signal-to-noise ratio is ~500 in the NIR channel and ~100 in the SWIR channel for a reference scene with Lambertian albedo 0.05 bserved at nadir and SZA =70°. Figure 1 illustrates typical spectra and contributions from the absorbers O2, CH4 and H2O.

The proposed method is derived from our algorithm development dedicated to CO2 and CH4 retrievals from OCO and GOSAT observations. Generally, we aim at inferring the state vector x (with xj the j-th retrieval parameter) from simulated measurements y (with yi the i-th spectral element) of backscattered sunlight in TROPOMI’s NIR and SWIR channels such as shown in Fig1. To this end, we set up a forward model F(x) that models the measurements y in both channels as accurately as possible.

Eq.(10) calculates the new state vector by adding an update derived from the measurements to the state vector of the previous iteration step. Thereby, the averaging kernel matrix filters out the null-space components that the measurements are not sensitive to. Since the retrieval is performed iteratively and we need to feed the forward model with a physically meaningful state vector, we add the null space components through a prior estimate xa.

Generally, Phillips-Tikhonov regulatization aims at reducing contributions from measurement noise to the retrieved state vector while retaining valuable information. The side-constraint in Eq.(5) and the choice of its weight via r play a crucial role. We choose the norm of the state vector as side constrain since the norm is particularly sensitive to noise contributions. The weighting matrix W renders the side constraint dimensionless and ensures that only the CH4 parameters and the scattering parameters contribute to its norm.Thus, we simultaneously regularise the CH4 vertical profile and the scattering parameters but retrieve the other parameters in a least-squares sense.

Performance of the proposed retrieval method is tested for an ensemble of simulated spectra. Our goal is to cover a realistic range of observations for aerosol and cirrus loaded scenes that TROPOMI will potentially encounter.

引自文章《TROPOMI aboard Sentinel-5 Precursor: Prospective performance of CH4 retrievals for aerosol and cirrus loaded atmospheres》