Summary of 7 satellite-derived products providing a global snapshot of water isotope distribution in 3D, with good information on zonal gradients and seasonality.
Remote-sensing observations of the water vapor isotopic composition can be used as a complementary tool to evaluate the hydrological cycle in atmospheric models.
Whatever the dataset used to evaluate a model, to take into account the spatio-temporal sampling biases of the data, it is preferable to collocate the model output with each of the measurements at a scale daily or shorter. To ensure that the large-scale circulation in the model is consistent with that of the data on a day-to-day basis, it is preferable to use simulations whose winds are nudged by reanalyses.
In addition to the collocation, to take into account the sensitivity of the instruments, the model outputs need to be convolved with averaging kernels available for some of the datasets.
Clear-sky sampling bias:
All datasets sample preferentially clear-sky scenes. Even when collocating model outputs from a nudged simulation, the model might simulate cloudy conditions when the data sees clear-sky and vice versa. Therefore, in the model the clear-sky sampling bias can be under-estimated compared to the data. This effect is difficult to take into account since the definition of a cloud varies between instruments and between models.
We select only retrievals for which the degree of freedom for signal is
higher than 0.5
When using the TES data, a correction must be applied which decreases the deltaD by about 4 permil. This correction depends on the averaging kernels of individual measurements, as described in Lee et al 2011 and Risi et al submitted.
The sensitivity of the TES deltaD retrievals decreases as surface temperature decreases, as humidity decreases and as cloud cover increases. The sensitivity becomes very small poleward of 45°S and 45°N. When the sensitivity is small, the retrieved deltaD tends toward a constant a-priori profile which is relatively enriched. Therefore, the TES retrievals underestimate the deltaD latitudinal gradient (Risi et al submitted).
Model outputs need to be convolved with averaging kernels, as described in Worden et al 2006, Risi et al submitted, Yoshimura et al in press, kurita et al submitted. This is crucial for a fair model-data comparison.
To avoid potential isotopic biases related to the presence of clouds or sampling of an incomplete atmospheric column, we discard all retrievals associated with a cloud fraction higher than 10% or with a retrieved precipitable water differing from ECMWF reanalyses by more than 10% (Risi et al 2010, Risi et al submitted).
No averaging kernel are available. We just need to calculate the total column average deltaD from the model outputs.
We discarded measurements with errors in H2O and HDO higher than the retrieved values. However, this leads to a slight bias towards measurements when H2O. In addition, we apply a 3 median average deviation filter to remove outliers (Risi et al submitted).
ACE does not use optimal estimation, and averaging kernels are not computed. To take into account the vertical resolution of the data, we convolved the model outputs with a triangular kernel of base 3 km (Dupuy et al 2008).
The sampling is very sparse. In the upper troposphere, the spatial coverage is insufficient to plot maps. Only zonal averages at best can be analyzed (Risi et al submitted).
We discard data with the visibility flag equal to zero and with diagonal elements of the averaging kernels lower than 0.03 (Risi et al submitted).
Model outputs need to be convolved with averaging kernels and a-priori profiles for each measurement the model is being collocated with.
Alternatively, since averaging kernels depend mainly on the tropopause height, we can use pre-computed representative averaging kernels for different tropopause height (Risi et al submitted).
Risi, Camille & National Center for Atmospheric Research Staff (Eds). Last modified 20 Aug 2013. "The Climate Data Guide: Water Isotopes from Satellites." Retrieved from https://climatedataguide.ucar.edu/climate-data/water-isotopes-satellites.