We report on the ground-based FTIR (Fourier transform infrared) tropospheric
water vapour isotopologue remote sensing data that have been recently made
available via the database of NDACC (Network for the Detection of Atmospheric
Composition Change;
Simultaneous observations of different tropospheric water
isotopologues can provide valuable information on moisture source, transport,
cloud processes, and precipitation
In recent years, there has been significant progress in measuring the
tropospheric water vapour isotopologues; remote sensing observations are
particularly interesting since they can provide data for the free troposphere
and they can be performed continuously (for cloud-free conditions). During
MUSICA (MUlti-platform remote
Sensing of Isotopologues for investigating the Cycle of Atmospheric water), a
method has been developed to obtain tropospheric water vapour profiles as
well as
In this paper, we present the MUSICA NDACC/FTIR data as provided recently via
the
Atmospheric remote sensing retrievals characterise an atmospheric state from
a measured spectrum. However, such an inversion problem is often ill-posed (a
lot of different atmospheric states can explain the measured spectrum).
Consequently, for solving this problem, some kind of regularisation is
required. This can be introduced by means of a cost function:
Water vapour isotopologues with reasonably strong and well-discernible
spectral infrared signatures are
Tropospheric water vapour shows a strong variation (in space and time) and it
can be better described by a log-normal than by a normal distribution
the humidity-proxy state: the the deuterium-excess proxy state (
The water vapour isotopologue state can be expressed on the basis of
The variation in humidity,
The covariance matrix
Matrix
In
For the previous retrieval version, we used 11 spectral windows with lines of
water vapour isotopologues
The spectral windows used for the MUSICA ground-based NDACC/FTIR
retrievals. Shown is an example of a typical measurement (Karlsruhe,
15 September 2011, 12:03 UT; solar elevation: 43.1
For v2015, we perform an optimal estimation of the
For the previous version, we used HITRAN 2008 line parameters
The empirical assessment study of
MUSICA NDACC/FTIR water vapour isotopologue data are available via the NDACC
database and can be accessed via
For isotopologue data, a new metadata template, GEOMS-TE-FTIR-ISO-001, has
been set up. It is almost identical to the GEOMS-TE-FTIR template (used for
all other FTIR data provided via NDACC in HDF4 format) but has the additional
variable “CROSSCORRELATE.N”. For MUSICA data, this new variable has three
entries: “
All the isotopologue data have been normalised with respect to their natural
isotopologue abundances. These natural abundances are 0.997317 for
List of current MUSICA NDACC/FTIR sites (ordered from north to
south) and available MUSICA data record. DOFS (type 1) reports the typical
trace of
Two different data types are provided. The first type is stored in HDF files
called “ftir.iso.h2o”. These files report the best estimate of the
The number of stations contributing to the MUSICA NDACC/FTIR data set is
gradually increasing and Table
An averaging kernel describes how a retrieved state vector responds to
variations in the real atmospheric state vector. The MUSICA NDACC/FTIR
isotopologue state vector consists of three trace gases and the corresponding
full averaging kernel matrix
Column entries of the nine blocks of the full averaging kernel
matrix
For error calculations, the same uncertainty sources are assumed for all MUSICA
NDACC/FTIR stations and are grouped into statistical and systematic errors.
An overview of the uncertainty assumptions is given in
Table
Uncertainty sources used for the error estimation. The second column gives the assumed uncertainty value and the third column the assumed partitioning between statistical and systematic sources.
Partial and total column densities are also provided as well as column
sensitivity averaging kernels and column error covariances. Partial columns
are calculated for the layers between the
All column data are calculated from mixing ratio data (mixing ratio state
vector, error covariances, and averaging kernels). For a conversion of mixing
ratios (ppmv) to number densities (
Please note: since GEOMS requires the same number of elements of column data
and the respective mixing ratio data, value 0 has been added for all column
data values at level
In order to be compliant with GEOMS, data are provided for the full state
vector consisting of the
For operations with averaging kernels (e.g. when adjusting model data to the
sensitivity of a remote sensing system), it has to be considered that the
retrieval works on a logarithmic scale, because only on this scale are
linearity assumptions valid. Therefore, it is strongly recommended to
transfer the averaging kernels to a logarithmic scale; in doing so, it has to
be considered that the derivatives are calculated for the state as given by
the retrieval state vector
Same as Fig.
For the purpose of error analyses, it is also very useful to transfer the
error covariances on a logarithmic scale. On this scale, the
For many purposes, remote sensing data may be compared to other data. For
instance, they may be compared to vertically resolved observational data in
order to empirically assess the quality of the different data sets, or they
are compared to model data in order to investigate model performances. For
such comparisons, it is important to consider the sensitivity of the remote
sensing system. While highly resolved profile data or model data generally
capture atmospheric signals well even on rather small scales, remote sensing
data report atmospheric signals according to the averaging kernel. In order
to make different data sets comparable, we have to convolve the data with the
averaging kernels. A full water vapour isotopologue state vector obtained
from highly resolved profile measurements or model calculations
(
About 99.7 % of all water vapour is present in form of the isotopologue
In order to simulate how a remote sensing system observes a
Equation
The MUSICA NDACC/FTIR retrieval performs an optimal estimation of the
humidity,
A transformation of the error covariance matrices onto the
Logarithmic-scale kernel matrix
During MUSICA, an a posteriori processing method for obtaining a quasi-optimal
estimation product of
Figure
Column entries of the nine blocks of the logarithmic-scale kernel
matrix in the
Same as Fig.
Figure
Same as Fig.
Figure
Summary of recommendations and comments for the two principal types of data users.
The MUSICA NDACC/FTIR data are publicly available via the database of NDACC
(
For a correct optimal estimation retrieval of the full water vapour
isotopologue state, we have to consider that atmospheric variations in
different isotopologues are strongly correlated. This strong correlation is
then also present in the retrieved state vectors and it has to be considered
when interpreting averaging kernels and error covariances. As a consequence,
it makes little sense to provide different isotopologues in the form of
individual states and via individual data sets. Instead, it is essential that
water vapour isotopologues are made available as single full state vectors
together with their full averaging kernels and error covariances. The
standard GEOMS metadata template for FTIR data on the NDACC database (called
GEOMS-TE-FTIR) does not allow data to be provided in such a format and a
slight extension of the standard FTIR template has been made (the modified
template is called GEOMS-TE-FTIR-ISO-001). The MUSICA NDACC/FTIR data are now
available in this new data format on the
In order to be compliant with GEOMS, the data are provided on a linear scale
and as volume mixing ratio (ppmv). However, since the retrieval is performed
on a logarithmic scale, it is recommended to transfer states, kernels, and
averaging kernels onto a logarithmic scale. On this scale, linearity in the
context of averaging kernel operations can be assumed and we can furthermore
make transformations between the
The MUSICA NDACC/FTIR data are made available in the form of two different
data types. The first type (“ftir.iso.h2o”) is the direct retrieval output.
It reports the optimal estimations of the states
Atmospheric remote sensing means that the atmospheric state is retrieved from
the radiation measured after having interacted with the atmosphere. For a
mathematical treatment, vectors are used for describing the atmospheric state
and the measured radiation. Hence, vector and matrix algebra is the tool used
in the context of remote sensing retrievals and remote sensing product
characterisation. In the following, we briefly explain the connections
between vector and matrix algebra and atmospheric remote sensing that are
relevant for our paper. For a more detailed introduction, please refer to
We are interested in the vertical distribution of
The interaction of radiation with the atmosphere is modelled by a radiative
transport model (also called a forward model,
The averaging kernel relates the real atmospheric state to the atmospheric
state as provided by the remote sensing retrieval (see, for instance,
Eq.
A coordinate system for a vector space is called a basis and consists of
linearly independent unit vectors. Vector spaces can be equivalently
described by different bases. One possibility for describing the atmospheric
water vapour isotopologue state is to use three independent basis systems for
the
Both basis systems are equivalent, but using both of them helps to correctly
constrain the inversion problem and to adequately describe the
characteristics of the remote sensing product. The retrieval processor works
in the
For the ground-based FTIR retrieval, consideration of a non-Voigt line shape
parameterisation becomes important because of the very high-resolution
spectra (full width at half maximum of the instrumental line shape of about
0.005
We allow for a speed-dependent Voigt line shape, and in doing so we assume a
Table
Modifications in the line parameters (line intensity and pressure broadening) made with respect to HITRAN 2012.
We would like to thank the many different technicians, PhD students, postdocs, and scientists from the different research groups that have been involved in the NDACC-FTIR activities during the last two decades. Thanks to their excellent work (maintenance, calibration, observation activities, etc.), high-quality, long-term data sets can be generated.
The Eureka measurements were made at the Polar Environment Atmospheric Research Laboratory (PEARL) by the Canadian Network for the Detection of Atmospheric Change (CANDAC), led by James R. Drummond, and in part by the Canadian Arctic ACE Validation Campaigns, led by Kaley A. Walker. They were supported by the AIF/NSRIT, CFI, CFCAS, CSA, EC, GOC-IPY, NSERC, NSTP, OIT, PCSP, and ORF. The authors wish to thank PEARL site manager Pierre F. Fogal, the CANDAC operators, and the staff at Environment Canada's Eureka weather station for their contributions to data acquisition, and logistical and on-site support.
We thank the Alfred Wegener Institute Bremerhaven for support in using the AWIPEV research base, Spitsbergen, Norway. The work has been supported by the EU project NORS.
We gratefully acknowledge the support by the SFB/TR 172 “ArctiC Amplification: Climate Relevant Atmospheric and SurfaCe Processes, and Feedback Mechanisms (AC) 3” in Projects B06 and E02 funded by the DFG.
We would like to thank Uwe Raffalski and Peter Völger for technical support at IRF Kiruna.
The University of Liège contribution to the present work has primarily been supported by the A3C PRODEX programme, funded by the Belgian Federal Science Policy Office (BELSPO, Brussels), and by the Swiss GAW-CH programme of MeteoSwiss (Zurich). Laboratory developments and mission expenses were funded by FRS-FNRS and the Fédération Wallonie-Bruxelles, respectively. We thank the International Foundation High Altitude Research Stations Jungfraujoch and Gornergrat (HFSJG, Bern) for supporting the facilities needed to perform the observations.
Eliezer Sepúlveda is supported by the Ministerio de Economía y Competitividad from Spain under the project CGL2012-37505 (NOVIA project).
The measurements in Mexico (Altzomoni) are supported by UNAM-DGAPA grants
(IN109914, IN112216) and Conacyt (239618, 249374). Start-up of the
measurements in Altzomoni was supported by the International Bureau of BMBF
under contract no. 01DN12064. Special thanks to A. Bezanilla for data
management and the RUOA programme (
Measurements at Wollongong are supported by the Australian Research Council, grant DP110103118.
We would like to thank Antarctica New Zealand and the Scott Base staff for providing logistical support for the NDACC-FTIR measurement programme at Arrival Heights.
This study has been conducted in the framework of the project MUSICA, which was funded by the European Research Council under the European Community's Seventh Framework Programme (FP7/2007-2013)/ERC grant agreement number 256961. Edited by: H. Maring Reviewed by: two anonymous referees