AMTAtmospheric Measurement TechniquesAMTAtmos. Meas. Tech.1867-8548Copernicus PublicationsGöttingen, Germany10.5194/amt-10-2129-2017Thin ice clouds in the Arctic: cloud optical depth and particle size retrieved from ground-based thermal infrared radiometryBlanchardYannyann.blanchard@usherbrooke.cahttps://orcid.org/0000-0003-0515-3041RoyerAlainO'NeillNorman T.TurnerDavid D.https://orcid.org/0000-0003-1097-897XElorantaEdwin W.Centre d'Applications et de Recherches en Télédétection,
Université de Sherbrooke, Sherbrooke, Québec, CanadaGlobal Systems Division, Earth System Research Laboratory, National Oceanic and Atmospheric Administration, Boulder, Colorado, USASpace Science and Engineering Center, University of Wisconsin, Madison, Wisconsin, USAYann Blanchard (yann.blanchard@usherbrooke.ca)9June2017106212921478November201614November20166April201728April2017This work is licensed under the Creative Commons Attribution 3.0 Unported License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/3.0/This article is available from https://amt.copernicus.org/articles/10/2129/2017/amt-10-2129-2017.htmlThe full text article is available as a PDF file from https://amt.copernicus.org/articles/10/2129/2017/amt-10-2129-2017.pdf
Multiband downwelling thermal measurements of zenith sky radiance,
along with cloud boundary heights, were used in a retrieval algorithm to
estimate cloud optical depth and effective particle diameter of thin ice
clouds in the Canadian High Arctic. Ground-based thermal infrared (IR)
radiances for 150 semitransparent ice clouds cases were acquired at the
Polar Environment Atmospheric Research Laboratory (PEARL) in Eureka, Nunavut,
Canada (80∘ N, 86∘ W). We analyzed and quantified the
sensitivity of downwelling thermal radiance to several cloud parameters
including optical depth, effective particle diameter and shape, water vapor
content, cloud geometric thickness and cloud base altitude. A lookup table
retrieval method was used to successfully extract, through an optimal
estimation method, cloud optical depth up to a maximum value of 2.6 and to
separate thin ice clouds into two classes: (1) TIC1 clouds characterized by
small crystals (effective particle diameter ≤ 30 µm), and (2) TIC2 clouds characterized by large ice crystals (effective particle diameter
> 30 µm). The retrieval technique was validated using data from
the Arctic High Spectral Resolution Lidar (AHSRL) and Millimeter Wave Cloud
Radar (MMCR). Inversions were performed over three polar winters and
results showed a significant correlation (R2= 0.95) for cloud optical
depth retrievals and an overall accuracy of 83 % for the classification of
TIC1 and TIC2 clouds. A partial validation relative to an algorithm based on
high spectral resolution downwelling IR radiance measurements between 8 and
21 µm was also performed. It confirms the robustness of the
optical depth retrieval and the fact that the broadband thermal radiometer
retrieval was sensitive to small particle (TIC1) sizes.
Introduction
Predictions of future climate change and its regional and global impacts
require that a better understanding of the radiative transfer interactions
between clouds, water vapor and precipitation are incorporated into
appropriate models. Recent CMIP5 model intercomparisons the Coupled
Model Intercomparison Project as described in indicate large
variability in ice cloud parameters (for example ice water content) amongst
high-latitude models. Shortcomings in ice cloud parameterization
impact their representation of radiative effects as well as
water cycles and lead to uncertainties in quantifying cloud feedbacks in the
context of climate change . High-altitude thin ice clouds
consisting of pure ice crystals, which cover between 20 to 40 % of the Earth
, can, for example, have opposing effects on the radiative
properties of the Earth. A surface-cooling effect ensues when scattering by
clouds reduces the solar radiation reaching the Earth's surface (i.e., albedo
effect). By contrast, a reduction in the amount of IR energy escaping the
Earth–atmosphere system occurs when the upwelling IR radiation emitted by the
Earth's surface and lower atmosphere is absorbed by clouds and radiated back
downward (i.e., greenhouse effect; ). The macrophysical
and microphysical properties of thin ice clouds determine which process
dominates and hence determine the net forcing of thin ice clouds on the
climate system . The optical properties of thin ice
clouds can be represented by extensive parameters such as the optical depth
and ice water content as well as intensive parameters such as ice crystal
size and shape. In an Arctic environment, the radiative effects of ice clouds
are unique because their radiative forcing influence on the energy balance
depends on seasonal polar day to polar night variation as well as large-scale
processes like the Arctic Oscillation . The advent of active
sensors on board satellites (for example CALIPSO/CloudSat) has enabled the
application of considerably more resources for polar region ice cloud
studies. This permits the evaluation of climate models and
satellite cloud climatologies . Long-term ground-based
observations, which are also essential for the validation of models and
satellite climatology, are, however, limited in their Arctic coverage
.
Thermal IR radiometry is a well-known technique for investigating the
presence and the emissivity of clouds . Numerous researchers
have exploited the thermal IR behavior of the absorption and scattering
efficiencies of cloud particles as a means of retrieving CODs and particle
effective sizes e.g.,. As cloud altitudes (temperatures)
can lead to large uncertainties in this latter technique,
proposed using lidar backscatter profiles along with IR radiometry to
estimate cloud altitudes and accordingly improve the retrieval accuracy of
cloud emissivity. This active/passive technique (called LIRAD for
lidar/radiometer method by Platt) has evolved over the years with
improvements such as the availability of high-resolution spectrometers
. The LIRAD technique is based on spectral
radiance/brightness temperature comparisons between measurements and
radiative transfer calculations. It performs better in the presence of high
thermal contrast and is thus well suited for cloud retrievals
. In more recent applications, cloud optical depth,
effective radius and ice fraction were retrieved from AERI Atmospheric
Emitted Radiance Interferometer; spectral
downwelling radiance observations in Antarctic, during the surface heat
budget of the Arctic Ocean (SHEBA) campaign and the ARM (Atmospheric Radiation Measurement) Alaska North Slope
site respectively. The
method was also employed at Eureka (Nunavut, Canada) to
retrieve cloud optical depth and cloud microphysical parameters from AERI
spectra acquired between 2006 and 2009 . A proposed
satellite-based instrument, with the goal of characterizing thin
ice clouds in the Arctic using far and thermal infrared channels
, was recently tested during an airborne campaign in the
High Arctic .
In this paper, we examine how multiband thermal measurements of zenith sky
radiance can be used to retrieve what are, as indicated in the early remote
sensing literature see,for example,, the most critical
extensive and intensive parameters influencing the radiative effects of ice
clouds: cloud optical depth (COD) and effective particle diameter
(Deff). We propose an application of this LIRAD technique with a
relatively simple and inexpensive instrument (less than 10 % of the cost of
an AERI) which is well suited to the Arctic environment .
Inasmuch as ice particle size is difficult to retrieve from IR radiometry,
the Deff component of our retrievals will be focused on a simple
discrimination of large and small crystal sizes. This approach was motivated
by previously published research that indicated such a discrimination would
play a key role in characterizing an important aerosol–cloud interaction
process in polar winter, namely precipitative cooling see, for
example,. An important aspect in this paper is
that our COD and Deff retrievals will be validated using independent
lidar and radar retrievals.
Section 2 highlights the importance of studying ice clouds while Sect. 3 is
devoted to the description of the study site and instrumentation. Section 4
examines the sensitivity of thermal IR radiometry to key ice cloud
parameters. In Sect. 5 we describe and verify the proposed methodology for
retrieving COD and Deff using thermal IR radiometry measurements. The
results are presented and discussed in Sect. 6 for the 150 thin ice clouds
cases we observed in the Arctic.
Classification and parameterization of thin ice clouds
Water vapor and clouds are a significant climate modeling challenge since
they represent major radiative forcing influences, while being the least
understood components of the climate system .
Much of the recent research has been focused on aerosol–cloud interactive
processes involving aerosols acting as ice and water cloud nuclei as well as their
subsequent effect on cloud microphysics, precipitation and radiation (see, for
example, on recent ARM campaigns, and
respectively, on the cloud remote sensing mandate of
the A-Train and EarthCARE satellite missions and as part of
the NETCARE project). In particular, understanding aerosols and their
radiative effects, especially their indirect impacts as cloud condensation
nuclei, is of critical importance for climate change models. The indirect
effect of aerosols represents a cooling influence (amplitude is subject
to large uncertainties) on the global radiative budget
Intergovernmental Panel on Climate Change,. The estimated
uncertainty of the indirect forcing component (between -0.1 and -1.3 W m-2) is associated with variations in cloud properties and cloud
lifetime. In the Arctic, the nature of thin ice clouds can effectively induce
an indirect cooling influence given the proper conditions. In terms of the
purpose and motivation for this paper, we note that the presence of
sulfuric-acid-bearing aerosols (viz., Arctic haze) can significantly
increase the size of ice particles (relative to the size of ice particles
formed from more pristine, low-acid aerosols or supercooled droplets). This
process can cause enhanced precipitation and important cooling effects during
the polar winter and could possibly lead to a dehydration greenhouse feedback
(DGF) effect, as proposed by .
The small and large ice particles described above are often abbreviated as
TIC1 and TIC2 (thin ice cloud, types 1 and 2). Thin ice cloud classification
was carried out by using the active techniques of lidar
and radar: CALIPSO and CloudSat data were employed to discriminate between
TIC1 and TIC2 ice clouds using the CloudSat small-particle sensitivity
minimum of approximately 30–40 µm. In this study, we seek to
demonstrate that TIC1 and TIC2 discrimination can be determined using
zenith-looking IR radiance measurements acquired at the Eureka observatory in
the Canadian High Arctic. Figure 1 illustrates lidar and radar backscatter
profiles acquired at Eureka for distinct TIC1 and TIC2 cases.
Classification of thin ice clouds using ground-based lidar (up) and
radar backscatter profiles (bottom).
The left-hand lidar profile of Fig. 1 shows a TIC that is largely
transparent to cloud radar while the right-hand lidar profile shows a thin
ice cloud that is readily detected by the radar. The disparity in radar
detectivity enables one to conclude that the former case corresponds to a
small-particle TIC1 event while the latter case corresponds to a
large-particle TIC2 event.
The presence or absence of thin ice clouds in the winter can lead to
significant changes in surface cooling . Given the
important radiative influence of ice crystal size and COD it is necessary
that these parameters are well characterized in order to improve modeling of
their radiative effects and thus their influence within a
context of TIC1 and TIC2 clouds.
The effective diameter of atmospheric ice particles is defined by
:
Deff=3V2A,
where A and V are respectively the total projected area and volume of all ice
particles per unit surface area in a given atmospheric column
. COD is given by
COD=∫πQextD24a(D)dD,
where Qext is the extinction efficiency (extinction cross section per
unit projected-particle area; ), D is the particle diameter
and a(D) is the ice particle number density per unit increment in diameter.
We note that, while the lidar-derived CODs employed in this article are at
0.532 µm, the IR CODs from our retrieval method were referenced,
for convenience, to 0.55 µm (we assume that COD differences
between 0.532 and 0.55 µm are negligible within the context of
other uncertainties encountered in this study).
Study site and instrumentation
The observation site was the Polar Environment Atmospheric Research
Laboratory (PEARL) in Eureka, Nunavut (80∘ N, 86∘ W) which is
one of the high-latitude stations of the Network for the Detection of
Atmospheric Composition Change (NDACC,
http://www.ndsc.ncep.noaa.gov/sites/stat_reps/eureka/). This high-Arctic site is located in the northernmost part of the Canadian Arctic
Archipelago. It was chosen because of our interest in Arctic ice clouds and
to exploit the diverse and complementary inventory of atmospheric instruments
listed in Table 1 (i.e., lidar, radar, IR spectrometer and radiosondes), as
well as the infrastructure and logistics support for field campaigns.
Detailed descriptions of the AHSRL and MMCR data processing and
interpretative techniques can be found in and
. A summary of instrument specifications is given in
. present a
discussion of the AERI performance which is applicable to the present paper.
The AERI instrument is known to have a very small warm bias for low-radiance
measurements, typically for clear-sky events, of the order of 1 % of the
ambient radiance . As the focus of our work
is on clouds with COD greater than 0.1, this warm bias in the AERI has only a
slight to negligible impact for retrievals involving very thin clouds
. The AERI data used in this work have been postprocessed
to reduce the uncorrelated random error in the data using principal component
analysis .
List of the Eureka (PEARL) instruments employed in our analysis.
Ground-based instrumentDuty cycleResearch group or institutionAHSRL (Arctic High Spectral Resolution Lidar)ContinuousSEARCH/NOAA – University ofWisconsin–MadisonMMCR (Millimeter Cloud Radar)ContinuousSEARCH/NOAA – ARMRadiosondeTwice a dayEnvironment CanadaP-AERI (Polar-AERI)ContinuousSEARCH/NOAA – University of Idaho(except during precipitation)
In this paper, we focus on the potential of using data from a ground-based
multiband thermal radiometer, the CIMEL CE-312 developed by CIMEL Inc
seefor descriptions of a similar instrument. The six channels of this radiometer correspond to (full width at
half maximum) limits of 8.2–8.6, 8.5–8.9, 8.9–9.3, 10.2–10.9, 10.9–11.7 and
11.8–13.2 µm and filter response peak values at 8.4, 8.7, 9.2,
10.7, 11.3 and 12.7 µm. The multiband radiometer is also a
robust instrument that, unlike the AERI, does not require a thermally
controlled environment. In actual fact, however, we had to simulate the
response of this radiometer by convolving the spectral transmittance of each
filter with the spectra of the Eureka Polar AERI (P-AERI) instrument
(provided by Von Walden at the U. of Idaho and NOAA). The reason for this was
that the CIMEL radiometer that we hoped to use was not ready for deployment
when we performed the field campaigns.
Sensitivity of thermal infrared radiometry to thin ice clouds
The spectral sensitivity of longwave (thermal) radiation to the microphysical
properties of ice clouds has been investigated for satellite data
, airborne data and ground-based sensors e.g.and articles cited
above. Previous studies have demonstrated that
thermal IR radiometry is relevant in terms of permitting the retrieval of
both COD and, to a degree, Deff. The retrieval of the latter parameter
permits, in turn, a discrimination of TIC1 and TIC2 clouds. The dependence of
thermal IR radiometry on ice particle size is represented by Eq. (2).
The extinction efficiency, a measure of particle attenuation (absorption and
scattering), depends on particle size, composition and shape as well as
wavelength . It is common, in the case of zenith-looking
thermal IR (8–14 µm) radiometry of ice clouds, to neglect the
scattering portion of Qext (where Qext=Qabs+Qsca),
especially for large particles . The result in the presence
of a medium such as cloud is extremely simplified radiative transfer that is
characterized by a strong forward-scattering phase function: in the limit of
a delta-function phase function, all forward-scattered radiation in any given
direction of incidence is returned to the incident beam and the only radiance
loss is due to absorption see, for example, the delta-function
irradiance solution of. and later authors
such as indicated that only a small fraction of the
zenith-looking downwelling radiation emitted by a cloud was due to scattering
(in spite of the fact that Qsca≈Qabs).
(a, b) Absorption efficiency spectra for TIC1 and TIC2
particles across a range of Deff values. (c) Mean absorption efficiency
and standard deviations across the spectra of (a) and (b). The
triangular symbols represent the integration of the absorption efficiency
across the six bands of the CIMEL CE-312. These efficiencies were derived for
the “severely roughened solid column”-type crystals of and
, available at
http://www.ssec.wisc.edu/ice_models/polarization.html.
We accordingly chose to plot absorption efficiency spectra in Fig. 2 in
order to illustrate the spectral sensitivity of this key radiative transfer
parameter. The absorption efficiency of TIC1 particles (Deff from 10 to
30 µm) and TIC2 particles (Deff from 35 to 120 µm) for “severely roughened solid column”-type crystals (Fig. 2a and b)
were obtained from calculations reported by and
. These spectra were then replaced by their mean and standard
deviation over the two (TIC1 and TIC2) Deff regimes in order to better
appreciate the band to band separability of the TIC1 and TIC2 size classes
(Fig. 2c). Prior to computing those means and standard deviations of Fig. 2c, we integrated the individual spectra of Fig. 2a and b across the
pass bands of the six channels employed in this study (triangles in Fig. 2c). The Deff ranges employed to define TIC1/TIC2 particles, for the
averaging carried out in the creation of Fig. 2c, were, as in
, roughly based on their nondetectability to
detectability threshold in radar backscatter returns.
This coarse spectral representation obtained for the six band averages and
standard deviations enables one to better appreciate the more robust nuances
between the two families of spectral curves (especially for the 8.4, 8.7, 9.2
and 12.7 µm channels) and better understand the key
discriminatory elements of the classification into TIC1 and TIC2 clouds. It
is clear from Fig. 2c that the first four bands offer the greatest potential
for discriminating particle size. Figure 2a and b, however, indicate a
decreasing sensitivity to increasing particle size as one approaches
Deff values in the tens of micrometers.
We simulated the influence of COD and Deff variations, on brightness
temperature (Tb) variations using the MODTRAN 4 radiative transfer model
. Figure 3 shows simulated Tb variations for the six
radiometer channels as a function of COD for fixed Deff and as a
function of Deff for fixed COD. The fixed values of Deff and COD
(and other independent parameters of the MODTRAN 4 runs) correspond to a
reference case with parameters that are defined in Table 2. We chose the input
parameters of the reference case as the set of mean parameters obtained by
averaging over the parameters of the 150 cloud cases that we employed to
provide an empirical validation of our retrieval (see Sect. 6.2 and Table 2
for more details). The curves of Fig. 3 represent an illustrative subset of
our inversion lookup table (LUT) that we employed as a means of retrieving
COD and Deff from measured values of Tb. The Deff column is
biased by the fact that the lidar–radar retrieval is insensitive to TIC1
particles: the Deff reference value was accordingly biased downward to
roughly overcome this insensitivity. The value of 50 µm was also
the value employed by .
Average, standard deviation, maximum and minimum values of the
parameters of the cloud cases used in this study. The reference case, defined
in the last row, was employed to produce Fig. 3 (while varying Deff
and the COD) and was used for the sensitivity study of Fig. 4. The means,
standard deviations and extrema of each parameter were derived from our
analysis of the 150 cloud cases. For the reference case, the cloud base
height and thickness values were, for the sake of convenience, rounded to the
nearest incremental step of the MODTRAN vertical layer profiles.
Cloud baseThicknessWater vaporCODDeff (µm)Ice particle shapeheight (km)(km)content (g cm-2)(lidar)(lidar–radar)(three sets of models)Average5.192.300.190.4691.06Solid columnsStandard deviation2.051.620.090.4825.16Aggregate of solid columnsMax9.008.000.852.60158.26A mixture involving a setMin0.800.200.080.1040.07of nine habitsReference case5.202.200.190.5050.00Solid columns
A full description of the models can be found at http://www.ssec.wisc.edu/ice_models/polarization.html.
Figure 3a indicates that, at a fixed Deff value of 50 µm,
there is a strong and monotonic variation in Tb as a function of COD for
all channels. At COD magnitudes greater than 2–3, the Tb values for all
channels converge towards an asymptotic ceiling that is the brightness
temperature of an opaque representation of the cloud. This clearly shows that
the sensitivity of the method decreases progressively as the COD increases
beyond 3.
Differences in Tb behavior over a range of Deff values and a fixed
COD of 0.5 can be observed in Fig. 3b. For the channels of nominal
wavelength less than 10.7 µm, Tb varies monotonically with
Deff up to approximately 30 µm, after which the response
plateaus to variations of ≈ 1K or less. For the 11.3 and
12.7 µm channels, the responses are nonmonotonic (or
considerably less monotonic) for the smaller values of Deff and smoothly
decrease with increasing Deff beyond a peak in the 10–20 µm range. This decrease is associated with the relatively large spectral
changes seen in the refractive index of ice particles at these larger
wavelengths see, for example,.
Variation in brightness temperature (Tb) with (a) cloud optical
depth (COD) and (b) effective diameter (Deff) for the six bands of the
CIMEL CE-312. The color legend of the left graph applies to both graphs.
The MODTRAN input parameters for this reference case are detailed in Table 2.
As discussed above, IR radiance measurements are sensitive to a variety of
cloud parameters as well as to the cloud environment. The simulation results
in Fig. 4 detail the band-dependent effects of six different parameters by
comparing changes in Tb induced by each parameter individually. These were
obtained for 1000 appropriately normalized samples of a random number
generator with a normal probability distribution, the mean and standard
deviation of which was controlled by the six parameter values of Table 2 (COD,
Deff, cloud base height, cloud thickness, column-integrated water vapor
of the atmosphere, WVC, and particle shape). The particle shape parameter is
based on three particle characterizations as defined by :
severely roughened solid columns, a general habit mixture involving a set of
nine habits and severely roughened aggregate of solid columns. The standard
deviations in Tb that result from the variation of the six parameters are
computed relative to the reference case defined above (Table 2). We note that
there was little sensitivity to the choice of a 50 µm effective
diameter for the reference case: changing this typical TIC2 value to a value
more representative of TIC1 particles produced differences of less than 1 K in Fig. 4 .
Sensitivity of Tb as a function of six key radiative transfer
parameters. The standard deviations (in units of K) are obtained by
stochastically varying, with a sample size of 1000, the parameters of
interest within the limits given in the Table 2.
Figure 4 shows that the chosen COD variation had the strongest Tb
influence of all of the parameters, especially for the bands at 10.7 and 11.3 µm (as one could infer by referring to Fig. 3a). Changes in
Deff (in red) lead to a standard deviation around 2 K for the
four first bands as can be qualitatively appreciated by referring to Fig. 3b. Changes in the altitude of the cloud induce a standard
deviation up to 5 K. Indeed, because measurements of thermal IR radiometry are
sensitive to temperature, a change in altitude causes a Tb difference that
is sensitive to the range of temperatures within which the cloud is located.
Cloud thickness and WVC are marginally important parameters in terms of the
magnitude of the changes induced in Tb. If the altitude and the thickness
of the clouds are known from vertical lidar (and radar) profiles, then the
magnitude of the altitude and cloud thickness uncertainties of Fig. 4 will
fall to levels commensurate with the standard deviations ascribed to the
uncertainty in the ice particle shape (≈ 1 K as per the dark
blue curve). This latter uncertainty (determined from the habit
parameterizations listed in Table 2) is largely inflexible inasmuch as our
ability to distinguish ice particle shape from lidar depolarization data is
extremely limited. Water vapor content (WVC) in the atmosphere, which remains
relatively low during the polar winter at Eureka, has a weak absorption
influence on the radiance measurements acquired in the CE-312 bands. Its
associated uncertainty is commensurate with the uncertainties due to particle
shape and cloud thickness: however integrated WVC is estimated from
radiosonde profiles at Eureka and thus its uncertainty can be reduced to
levels significantly below the particle shape and cloud thickness
uncertainties. These reductions in the uncertainty of nominally known input
parameters will be such that the variability of the parameters to be inverted
(COD and Deff) is significantly larger than the uncertainty of the known
input parameters (for all bands in the case of the COD and at least in the
case of the first four bands for Deff).
Methodology
Our LIRAD objective was performed using LUTs and MODTRAN 4 radiative transfer
simulations to parameterize the behavior of the downwelling zenith sky
radiance as a function of key input parameters, including COD and Deff.
The methodology is represented in the flow chart of Fig. 5. The core of this
method, inspired by ground-based retrievals e.g., is to
compare thermal IR radiance measurements with LUTs derived from MODTRAN 4
simulations. This inverse problem is solved using the optimal estimation
method (OEM; ). The method seeks the state of maximal
probability, conditional on the value of the measurements, associated errors
and a priori knowledge. This OEM is an efficient inversion method that has
already been employed for ice cloud retrievals see, for
example,.
Flow chart of the retrieval method.
The steps of the retrieval method are as follows:
First, knowledge of the cloud environment at the time of a given radiometer measurement is required. Specific input
auxiliary data includes pressure, temperature and water vapor profiles from radiosonde data and the effective cloud layer
height from lidar backscatter data. The radiosonde parameters are interpolated to the radiometer times while the time of
the selected lidar profile is the nearest to the radiometer time. To avoid the issue of interpolating radiosondes over
extensively long periods of time, the cases were selected to be as close as possible to radiosonde launch times. Radiosonde
humidity sensors are known to be subject to dry bias, especially in dry conditions and could yield relative humidity underestimates
of 10 % . The six channels are, however, far less sensitive to WVC than to COD (see Fig. 4) and therefore the bias is
expected to be lower in cloudy conditions. Cloud heights are estimated for sustained cloud features where clouds are defined by lidar
backscatter coefficients greater than 1.10-6 m-1 sr-1 and a lidar depolarization ratio greater than 20 %
(thresholds were inspired by but adapted to a different vertical resolution of our lidar). Upper and lower cloud boundaries
are then obtained where four continuous vertical samples of the lidar profile (4×30=120m for an AHSRL resolution of
30 m) comply with that requirement (preceded by a series of lower, noncloud samples).
Using MODTRAN 4, we simulated surface-based zenith-looking brightness temperatures of a cloudy atmosphere as a function of the
environmental data. A LUT is then constructed for 23 values of Deff between 10 and 120 µm and 31 values of COD
(from 0 to 3 with an increment of 0.1). Because Tb is so strongly dependent on COD, the LUT is linearly interpolated between
MODTRAN 4 calculations with a COD increment of 0.01.
Brightness temperatures are then derived from radiance measurements in the six CIMEL CE-312 radiometer channels extracted from band-integrated P-AERI spectra.
The OEM was used to compare the LUT spectra with the measured Tb spectra. This method requires precise quantification of
errors attributed to each variable of the state and measurement vectors as detailed inand in Appendix A. We
retrieve the best estimates of COD and Deff from the most optimal fit to the measured Tb.
Specific validation elements for our retrieval algorithm included profiles of
the effective ice particle diameter prime (Deff′) that were extracted
from the combination of AHSRL and MMCR backscatter coefficients. This was
carried out as per the technique developed by and applied
to the instruments at our study site . Deff′ is
given by
Deff′=βradarβlidar4,
where βradar and βlidar are the extinction cross sections
of the radar and lidar respectively. Deff′ can be related to Deff
assuming an analytical form for the size distribution which in our case was
taken as a modified gamma distribution of hexagonal columns
. In order to compare Deff with our retrievals, we
averaged this parameter over the vertical extent of a given cloud . We chose,
for simplicity's sake, to assume a specific particle shape (i.e.,the
hexagonal column shape) when retrieving Deff from the lidar/radar
profiles to enable consistent comparisons with our passive retrievals. As a
general quality assurance step for the radar data, those cases for which the
radar backscatter coefficient was less than 10-15 m-1 sr-1 (an empirically determined value of minimum
detectability) were eliminated from any retrieval processing (this generally
meant the elimination of TIC1 points). It is also important to state that
radar signal is proportional to the sixth power of the hydrometeor diameter,
whereas IR instruments are sensitive to the ratio of the third to the second
moment. This means that the equivalent Deff is not strictly the same and
their comparison can generate biases in some conditions see discussion
in.
The CODs from the passive algorithm was validated by comparison with
estimates of COD derived from AHSRL observations using Eq. (4) and
averaged over the cloud geometric thickness. By its design, the AHSRL
measures two signals which can be processed to yield separate lidar returns
for aerosol and molecular scattering, and then to make reliable measurements
of the extinction profile. The optical depth τ across a range interval
(r, r0) is computed as
τ(r)-τ(r0)=12lnρ(r)ρ(r0)-12lnSm(r)Sm(r0),
where ρ is the molecular density, and Sm is the range-squared,
background-corrected, molecular lidar return of the AHSRL. COD was
calculated, using Eq. (4), across the cloud layer where the vertical
cloud boundaries were determined according to the backscatter coefficient and
depolarization criteria described above. Due to the very small angular
field of view of the AHSRL receiver (45 µrad), it is common to
assume that the backscatter return is negligibly affected by multiple-scattered photons . However, to better quantify the
effect of multiple scattering, we applied Eloranta's (1998) approximate model
to calculate the multiply scattered lidar returns for the 150 cloud cases.
The inputs included molecular and particulate backscatter coefficients,
effective diameters (inferred from the AHSRL+MMCR technique discussed below)
and cloud height boundaries. The approximate model was employed to compute
multiple scattering returns up to the fourth order. The impact of multiple
scattering (MS) can then be evaluated in terms of COD inasmuch as multiple
scattering lowers the apparent COD. From Eq. (4), we define ΔCODms as
ΔCODms=-0.5log(1.0+Pt/P1),
where Pt is an output of the MS model, representing all orders of multiple
scattering while P1 is the single-scattering return. This term can be used
to correct the retrieved optical depth by inserting a multiplicative factor
η, as per to correct for the reduction of the
extinction coefficient. The parameter η is not constant
and Platt argued that it should vary between 0.5 and
1. Our best estimate of η over the ensemble of test cases was 0.95. This
factor was used to correct the lidar-derived CODs that we employed for
validation purposes in this paper.
Although intervals of P-AERI spectra (wavelength range of 3–20 µm) were used to simulate the response of the CIMEL radiometer bands, the
P-AERI instrument has more extensive capabilities and is sensitive to a
larger range of Deff, according to the absorption efficiency spectra
see Fig. 2 in this article or Fig. 5 in. The
mixed-phase cloud retrieval algorithm (MIXCRA; ) is designed
to estimate microphysical properties of both the ice and liquid components of
a cloud using spectral IR radiances supplemented with data from various
instruments. By using the spectral behavior of several “microwindows” between
gaseous absorption lines in the thermal and far IR, MIXCRA can determine
cloud phase and retrieve COD and Deff (a detailed description of the
algorithm can be found in ). have
demonstrated good agreement between the MIXCRA retrievals and HSRL optical
depth measurements during an experiment at the ARM NSA site.
describe the specifics of the Eureka implementation, including the auxiliary
measurements that were employed (notably the AHSRL, MMCR, radiosonde data and
a microwave radiometer). Within the scope of this current study, the
cloud-phase determination from MIXCRA is used to ensure the comparison of
ice-only cloud properties (i.e.,those MIXCRA retrievals that yielded
negligible liquid water path, LWP < 0.2 g m-2, were taken as
being pure ice-cloud cases). MIXCRA results are used here as an alternative
point of reference for our retrievals.
Results and discussionPhysical coherence of a specific case study
In this section we seek to illustrate the temporal variation of particle size
and COD in a precipitating cloud and demonstrate that our retrieval gives
physically coherent results for a specific case that was chosen to exercise
both the COD and Deff retrievals. Figure 6 shows the selected 2009
winter campaign case where we compare AHSRL backscatter coefficient, the MMCR
backscatter coefficient profile, the Deff profile (which, as pointed out
above, is related to Deff′) and the results of our inversion
(Fig. 6a, b, c and d respectively).
Radar reflectivity is commonly used to describe the reflection, scattering
and diffraction effects of a target on the incident signal. Radar
reflectivity, expressed in dBZ, is logarithmically proportional to
the backscatter coefficient and is proportional to the sixth power of the
hydrometeors diameter . However, to ensure a consistent
approach within the context of Deff retrieval, we chose to display
βradar in Fig. 6b.
Evolution of validation and retrieval cloud parameters during a
particular precipitating cloud event (13 January 2009) at Eureka. The
error bars of the bottommost graph represent the retrieval errors. The error
for the lidar COD retrievals is sufficiently small to be obscured by the size
of the symbols representing this component. The red crosses in the uppermost
plot are the cloud boundary limits used as input to our method. The points
upon which a star has been superimposed satisfied the criteria defined at the
beginning of Sect. 6.2 and are thus an example of points accepted in the
validation part of this article.
One can see (Fig. 6d) that Deff, for the lidar–radar technique (in
blue), increases from 44 to 103 µm. This increase
appears, in turn, to be correlated with cloud precipitation as evidenced by
the accompanying decrease in altitude of the cloud structure seen in the
lidar and radar profiles as well as increasing values of the radar Doppler
fall velocity profiles (not shown). The passive Deff retrievals (the
green colored curve of Fig. 6d) show a roughly similar trend from 20:00 to
23:00 (largely characterized, however, by significantly smaller Deff
values). The insensitivity of the latter retrieval to larger size particles
during the period from about 19:00 to 22:30 and the sudden jump in retrieved
Deff value after that time is the result of the type of asymptotic
ceiling that one sees in Fig. 3b and the choices made in the LUT algorithm
retrieval: as one approaches the asymptotic ceiling from smaller Deff
values, there is clearly a progressive increase in the range of acceptable
Deff values for a given ΔTb (a decrease in the robustness of
the retrieved value). This example also illustrates an important issue
related to our TIC1/TIC2 classification goal, for which some points, around 20:00
appear to be classified as TIC1 particles by our algorithm while the
lidar–radar values between 60 and 80 µm would be classified as
TIC2 particles. One possible explanation is that the cloud vertical
inhomogeneity, as evident in the cloud structure observable in Fig. 6a is the
source of the misclassification. The regions of the effective diameter that
are not detected by the lidar–radar retrieval (see the profiles of Fig. 6c)
are indeed optically thick regions with more impact on the radiometric
retrieval of Fig. 6d and likely contain smaller particles.
The COD retrievals are, as one would expect, visually coherent with the
general strength and extent of the lidar backscatter coefficient. After
23:00 UTC our COD retrievals approach the limit of retrieval sensitivity suggested
in Fig. 3. This is manifested by an artificial nonmonotonic increase in
the variability of the retrieved COD (not very obvious in Fig. 6 but
obvious from our inversions in general) and is coherent with the asymptotic
invariance of Tb with increasing COD in Fig. 3.
This example suggests that the dynamic evolution of cloud particle properties
provided by continuous temporal analysis can lend support in helping to
understand cloud dynamics and more specifically in discriminating TIC1 and
TIC2 particles. In the latter case the passively retrieved evidence for
progressively increasing values of COD and Deff, supported by the lidar
and radar data, would lend more confidence to a classification result which
indicated the presence of TIC2-type particles during the latter part of the
day.
Validation of our retrieval algorithm
Figure 7 shows COD and Deff comparisons between the radiometric
retrievals and the combined AHSRL and MMCR retrievals for over 150 ice clouds
observed between September 2006 and March 2009. The selection of the 150
cases was driven by different criteria: a requirement for monolayer clouds; a
cloud thickness greater than 200 m (to equal or exceed the MODTRAN
vertical layer thickness of 200 m); the time difference between two
samples was more than 30 min; the clouds were nonprecipitating; a
subjective criterion of cloud homogeneity; a constraint whereby the evidence
for cases of TIC1 only, TIC2 only or a combination of the two was determined
by whether the cloud was detected by the lidar and the radar; a requirement
that the cloud is semitransparent (AHSRL optical depth < 3); a constraint
that the IR signal in any band is not saturated; the visible optical
depth should be greater than 0.1 and the visible optical depth across
the first two kilometers (where diamond dust particles are very often present
in winter) should be less than 0.1. As an illustration of the influence of
these criteria, only 8 of the 41 points seen in Fig. 6 were selected to be
part of the 150 cases.
Those clouds are not meant to be representative of the Eureka cloud
climatology. Their mean base altitude (5.2 km) and vertical extent
(2.3 km) is substantially higher than a cloud climatology that was
generated across 4 years of data (; 1.8 and 2 km respectively) as well as the CALIPSO/CloudSat and ground-based
climatologies on the vertical distribution of ice-only cloud
.
Radiometer-based retrieval results for (a) CODs compared with lidar
derived CODs and (b)Deff compared with the lidar–radar retrieval
product. The same comparisons, with MIXCRA retrievals being the reference,
are shown in graphs (c) and (d). The TIC1 results, because of the lidar–radar
insensitivity of these small particles, are excluded from the scattergram but
their Deff frequency distribution is shown in the inlaid histogram. The
Deff comparisons with the MIXCRA retrievals show fewer points than the
comparisons with the lidar–radar retrieval because the MIXCRA retrievals of
Deff have relatively large uncertainties for cases where COD ≤ 0.2
(and thus are not shown).
Confusion matrix of the TIC1/TIC2 classification compared to
lidar–radar retrievals. The term “err. comm.” stands for the error of
commission.
The COD results (Fig. 7a) show a significant correlation with lidar (R2= 0.95) over a large optical depth range (from 0.1 to 2.6). This level of
agreement confirms the relatively strong sensitivity of the ensemble of the
radiometer bands to the COD (c.f. Fig. 3a). As seen in that figure, the
Tb sensitivity decreases with increasing COD such that the asymptotic
behavior of the COD variation tends towards an upper limit of COD
detectability of 2–3 (where the spread of Tb values ≈ the
measured uncertainty in those Tb values).
The quality of the particle size retrieval was difficult to quantify because
the thermal IR channels become increasingly less sensitive to particles
larger than ≈ 100 µm as suggested in Fig. 2b. This
insensitivity to large particle sizes likely contributes to the large
dispersion (and hence the marginal correlation) of retrieved TIC2 values seen
in Fig. 7b. The separation between TIC1 and TIC2 particles was affected
based on the Deff values from the lidar–radar retrieval: the lower bound
TIC2 value/upper bound TIC1 value was set at 30 µm. We show
below that while the radiometer Deff retrievals are of significantly
lower amplitude than the lidar–radar retrievals, the 30 µm
crossover criterion from TIC1 to TIC2 is sufficiently well delineated to
achieve acceptable classification accuracy. One factor that complicated the
Deff comparison over all particles sizes was the MMCR sensitivity limit
at smaller particle sizes (≈ 30–40 µm) and the
constraints this imposed on the lidar–radar retrieval. For that reason, the
TIC1 results were not considered in the R2 statistics. However, the TIC1
frequency distribution derived for our retrieval algorithm (inlaid histogram
in Fig. 7b) confirms the robustness of the retrievals inasmuch as 78 % of
the retrieved TIC1 population have a Deff less than 30 µm
(and 96 % less or equal to 50 µm).
The lack of TIC1 sensitivity of the lidar–radar combination means that our
radiometric retrieval algorithm cannot be verified with the lidar–radar
retrieval in the TIC1 particle-size region. Nonetheless, the lidar–radar
classification scheme of Sect. 3 can at least separate out the TIC1 and TIC2
cases. Table 3 presents the retrieved results in terms of the TIC1/TIC2
classification compared with the validation data. A threshold value of 30 µm was used to discriminate between the two classes in the case
of the radiometer retrieval. Those cases for which βradar was less
than 10-15 m-1 sr-1 were classified as TIC1 cloud (this
cutoff is illustrated by the dark regions of the βradar plot seen in
the case study of Fig. 6). The classification yielded satisfactory results
with an overall accuracy of 83 %. The TIC1 retrievals were associated with a
22 % detection (omission) error (11/50×100) while the TIC2 omission
error was 15 %. We should note that the classification results are moderately
sensitive to the threshold Deff value assumed between the TIC1 and TIC2
classes: for threshold values of 35 and 40 µm, the overall
accuracies were respectively 82 and 82 % with moderately smaller TIC1
omission errors (18 and 12 % respectively).
Confusion matrix of the TIC1/TIC2 classification compared to MIXCRA
retrievals. The term “err. comm.” stands for the error of commission.
A comparison with the MIXCRA retrievals (Fig. 7c and d and Table 4)
amounts to a coherency check between the two passive inversion techniques.
While limited in terms of absolute validation, this comparison effectively
reduces the array of confounding influences that can affect the retrieval
quality of both approaches (and in so doing, permits a more direct evaluation
of the strengths and weaknesses of either technique). In Fig. 7c, the good
correlation between MIXCRA's and our COD retrievals and a slope near 1
confirm the robustness of the COD retrieval. The good COD correlation is
expected inasmuch as a similar degree of correlation between MIXCRA and AHSRL
results for ice-only clouds was previously observed . The
comparison of the Deff values from our radiometer retrieval and the
MIXCRA retrieval (Fig. 7d) shows a somewhat better absolute agreement
relative to the comparisons of our radiometer retrieval with the lidar–radar
retrieval (point scatter closer to y=x but a value of R2 which is also
at the margins of significance).
Histogram of active and passive classifications of TIC (TIC1 in red
and pink; TIC2 in blue and cyan) as a function of COD.
Figure 8 indicates that all of cases with COD > 1 were classified as TIC2
for the passive and active retrievals. Clouds classified as TIC1 are, in
contrast, preferentially associated with COD less than 0.3. The degree of
TIC1/TIC2 classification coherence between the lidar–radar and the radiometer
retrievals, in the histogram of Fig. 8, illustrates the value of our
semiqualitative (binary) classification approach for an application
the DGF effect of that specifically requires such
binary information.
Downwelling longwave cloud radiative forcing of the 150 cases
decomposed in TIC1 and TIC2.
In order to better understand the physical implications of the retrievals, we
plotted, in Fig. 9, passive TIC1/TIC2 discrimination results along with
downwelling longwave cloud radiative forcing (DLCRF). These data represent
the 150 cloud cases that we employed above for the retrieval validation. The
general distribution of the DLCRF is similar to the Fig. 4 results of
, who applied the MIXCRA algorithm continuously (without the
separation into specific events) during the same period. As DLCRF is closely
linked to the COD, one can note that the TIC1 generally have a small DLCRF of
less than 10 W m-2. The DGF impact of the TIC2 particles (meaning
their progressive removal by precipitation) would accordingly be that their
radiative forcing influence would progressively decrease with an attendant
cooling due to a reduction in thermal interaction with the remaining TIC1
particles (and the unprecipitated TIC2 particles). A long-term analysis would
help to support modeling conclusions on the impact of acid-coated ice nuclei
on Arctic cloud as reported by . These authors reported a
mean downward longwave (negative) radiation anomaly at the surface of -3–5 W m-2 close to Eureka.
Additional sensitivity studies
The effect of particle shape on the retrievals was analyzed by reapplying
the retrieval algorithm to the 150 thin ice clouds using particle shapes
other than our solid column crystal standard (not shown). Retrievals results
showed that the shape employed in the LUT generation has only a small
influence on the performance of our retrievals (as one could have inferred
from Fig. 4). This confirms the previous conclusion from
. The validation of COD retrievals, expressed in terms of
RMSE, varied from 0.10 to 0.11 as a function of shape while overall
classification accuracy varied from 82 to 83 % (versus the results of 0.10
and 83 % respectively for the solid column crystal shape).
We exploited the extinction coefficient profile retrieval obtainable from
Eq. (4) in order to evaluate the impact of the extinction profile as an
added dimension of input to the radiative transfer model. Rather than
assuming a constant extinction coefficient profile across the width of the
cloud, we broke the cloud into vertical segments and obtained a coarse
vertical profile. Retrieval results showed a moderate impact of the
extinction profile since the RMSE is 0.09 instead of 0.10 while the overall
classification accuracy remained at 83 %. A similar comparison from MIXCRA
results (not shown here) confirms the moderate impact of the use of real
extinction profiles.
Conclusions
We developed a simple inversion technique to retrieve the optical depth and
effective particulate diameter of Arctic thin ice clouds from a
multibroadband thermal radiometer using lidar and radiosonde measurements as
auxiliary inputs to the inversion routine. Specific validation elements were
extracted from the combination of lidar and radar data, as well as AERI data.
Our retrieval technique was applied to 150 thin ice clouds measured at the
PEARL observatory (Nunavut, Canada).
The results of this study demonstrate the potential for retrieving key ice
cloud parameters from thermal IR radiometry (with bands between 8 and 13 µm). The COD retrieval algorithm showed good agreement with the
validation COD obtained from integrated lidar profiles. The retrievals of
Deff showed a marginal correlation for particle sizes restricted to the
TIC2 category (a constraint that was driven by the insensitivity of the
lidar–radar retrievals to small cloud particles). This is likely due to the
weak sensitivity of thermal IR measurements for particles larger than
≈ 100 µm. However, a classification of thin ice clouds in
terms of TIC1 and TIC2 classes, using a threshold discrimination of 30 µm results in a significant classification accuracy of 83 % for
our passive retrieval algorithm. Further analysis showed that the extinction
profile and particle shape had a relatively weak impact on the retrieval
results. Comparisons with the MIXCRA algorithm confirm the robustness of the
optical depth retrieval.
An important application of our work would be to deploy this technique as
part of a network of low-cost, robust instruments to monitor Arctic
clouds. Because their occurrence, type and altitude are spatially
inhomogeneous according to, we believe that
additional ground-based stations would be helpful to broaden our knowledge of
arctic ice clouds. Aside from being a ground-based retrieval approach in its
own right (in tandem with a lidar system), this method can also be used for
comparison with CALIPSO's level 2 products. The CALIPSO remote sensing suite
technique employs an on-board imaging IR radiometer and the CALIOP lidar to
enable the retrieval of particle size and optical depth across a narrow swath
image . Our retrieval, viewed as a CALIPSO validation
technique is rendered all the more interesting because of the geographic
position of the PEARL site; it is a high-Arctic site that sees frequent
thin-cloud events and its position near the maximum latitude of the CALIOP
polar orbit ensures that there are frequent overpasses of that sensor package
(within a radius of hundreds of kilometers).
The data we used are accessible from the CANDAC server using SSH protocol.
Registration to access CANDAC data can be done through the form (http://candac.ca/candacweb/contact-us) or by contacting
CANDAC data manager Yan Tsehtik (yan.tsehtik@candac.ca).
Optimal estimation method
The optimal estimation method OEM, is an efficient
solution to inverse problems, especially in atmospheric science. A good
understanding of the technique and its associated errors is a prerequisite
for the proper use of this method. Inasmuch as our application of OEM is very
similar to , we used the same formalism to
define the OEM components. As set out in Sect. 5, the OEM goal is to
retrieve state variables with the maximum probability of occurrence by
minimizing a cost function ϕ:
ϕ=(y-F(x))TSe-1(y-F(x))+(x-xa)TSa-1(x-xa),
where F is the forward model, i.e.,radiative transfer computation in our
case. The state (x), a priori (xa) and measurement (y) vectors are
defined as
x=DeffCOD;xa=Deff_aCODa;y=Tb_8.4Tb_8.7Tb_9.2Tb_10.7Tb_11.3Tb_12.7.
The a priori vector is the prior knowledge of the state vector, and typically
corresponds to climatological values of the state vector components. In our
case, the reference case of Table 2 was used to define the a priori vector
and its covariance matrix. Even if any particular a priori vector values have
an impact on the retrievals, it is common to attribute large uncertainties to
them in the covariance matrix Sa in order to let the measurement vector be
the dominant driver of the retrieval. This covariance matrix is given by
Sa=σDeff_a200σCODa2.
The total error covariance matrix Se is the quadratic sum of the
measurement error covariance matrix Sy and the forward model parameter
uncertainty covariance matrix Sf. We assumed that the components of the
measurement or state vectors are independent (i.e., that the covariance
matrix is diagonal). The measurement errors depend on the accuracy of the
radiometer which is assumed to be 0.1 K for each
band,. We presumed the measurement errors are
wavelength independent. The forward model errors represent the quadrature sum
of the uncertainties of each input parameter (cloud base height, thickness,
water vapor content and particle size) of the MODTRAN calculation. We then
use the sensitivity study, Fig. 4, to define the standard error (σ/1000) of each parameter from the stochastic analysis. The components
of Se are close to 0.30 K (between 0.28 and 0.34 K) and of the same order
of magnitude as Sy.
The authors declare that they have no conflict of
interest.
Acknowledgements
We would like to thank the Canadian Network for the Detection of Atmospheric
Change (CANDAC), Study of Environmental Arctic Change (SEARCH) and
Environment and Climate Change Canada for their operational support. We are
grateful to P. Von Walden, of the Washington State University, for the quality
checked P-AERI data. This research was supported by the Natural Sciences and
Engineering Research Council of Canada (NSERC), Fonds Québécois de la
Recherche sur la Nature et les Technologies (FQRNT) and the Canadian Space
Agency (CSA). This work was also partially supported by the US Department
of Energy Atmospheric System Research Program DE-SC0008830.
Edited by: H. Maring
Reviewed by: three anonymous referees
ReferencesAllen, J. R.: Measurements of Cloud Emissivity in the 8–13 μ Waveband, J.
Appl. Meteor., 10, 260–265,
10.1175/1520-0450(1971)010<0260:MOCEIT>2.0.CO;2,
1971.Baran, A. J.: From the single-scattering properties of ice crystals to climate
prediction: A way forward, Atmos. Res., 112, 45–69,
10.1016/j.atmosres.2012.04.010,
2012.Battan, L. J.: Radar observation of the atmosphere, Q. J. Roy. Meteorol. Soc., 99, 793–793, 10.1002/qj.49709942229,
1973.Baum, B. A., Yang, P., Heymsfield, A. J., Bansemer, A., Cole, B. H., Merrelli,
A., Schmitt, C., and Wang, C.: Ice cloud single-scattering property models
with the full phase matrix at wavelengths from 0.2 to 100 μ m, J. Quant. Spectrosc. Ra., 146, 123–139,
10.1016/j.jqsrt.2014.02.029,
2014.
Berk, A., Anderson, G. P., Acharya, P. K., Chetwynd, J. H., Bernstein, L. S.,
Shettle, E. P., Matthew, M. W., and Adler-Golden, S. M.: Modtran 4 user's
manual, Air Force Research Laboratory, 1999.
Bissonnette, L. R.: Lidar: Range-Resolved Optical Remote Sensing of the
Atmosphere, chap. Lidar and Multiple Scattering, Klaus Weitkamp editor,
Springer Series in Optical Sciences, Springer-Verlag, New York, 2005.Blanchard, Y., Pelon, J., Eloranta, E. W., Moran, K. P., Delanoë, J., and
Sèze, G.: A Synergistic Analysis of Cloud Cover and Vertical Distribution
from A-Train and Ground-Based Sensors over the High Arctic Station Eureka
from 2006 to 2010, J. Appl. Meteorol. Climatol., 53,
2553–2570, 10.1175/JAMC-D-14-0021.1,
2014.Blanchet, J.-P. and Girard, E.: Arctic “greenhouse effect”, Nature, 371,
383–383, 10.1038/371383a0, 1994.Blanchet, J.-P., Royer, A., Châteauneuf, F., Bouzid, Y., Blanchard, Y.,
Hamel, J.-F., de Lafontaine, J., Gauthier, P., O'Neill, N. T., Pancrati, O.,
and Garand, L.: TICFIRE: a far infrared payload to monitor the evolution of
thin ice clouds, Proc. SPIE 8176, Sensors, Systems, and Next-Generation Satellites XV, 81761K, 10.1117/12.898577,
2011.Bourdages, L., Duck, T. J., Lesins, G., Drummond, J. R., and Eloranta, E. W.: Physical properties of High Arctic tropospheric particles
during winter, Atmos. Chem. Phys., 9, 6881–6897, 10.5194/acp-9-6881-2009, 2009.Brogniez, G., Pietras, C., Legrand, M., Dubuisson, P., and Haeffelin, M.: A
High-Accuracy Multiwavelength Radiometer for In Situ Measurements in the
Thermal Infrared. Part II: Behavior in Field Experiments, J. Atmos. Ocean. Technol., 20, 1023–1033,
10.1175/1520-0426(2003)20<1023:AHMRFI>2.0.CO;2,
2003.Brogniez, G., Parol, F., Becu, L., Pelon, J., Jourdan, O., Gayet, J.-F.,
Auriol, F., Verwaerde, C., Balois, J.-Y., and Damiri, B.: Determination of
cirrus radiative parameters from combination between active and passive
remote sensing measurements during FRENCH/DIRAC 2001, Atmos. Res.,
72, 425–452, 10.1016/j.atmosres.2004.03.026,
2004.Chiriaco, M., Chepfer, H., Noel, V., Delaval, A., Haeffelin, M., Dubuisson, P.,
and Yang, P.: Improving Retrievals of Cirrus Cloud Particle Size Coupling
Lidar and Three-Channel Radiometric Techniques, Mon. Weather Rev., 132,
1684–1700, 10.1175/1520-0493(2004)132<1684:IROCCP>2.0.CO;2,
2004.Comstock, J. M. and Sassen, K.: Retrieval of Cirrus Cloud Radiative and
Backscattering Properties Using Combined Lidar and Infrared Radiometer
(LIRAD) Measurements, J. Atmos. Ocean. Technol., 18,
1658–1673, 10.1175/1520-0426(2001)018<1658:ROCCRA>2.0.CO;2,
2001.Cox, C. J., Turner, D. D., Rowe, P. M., Shupe, M. D., and Walden, V. P.: Cloud
Microphysical Properties Retrieved from Downwelling Infrared Radiance
Measurements Made at Eureka, Nunavut, Canada (2006–09), J. Appl. Meteor.
Climatol., 53, 772–791, 10.1175/JAMC-D-13-0113.1,
2014.Delamere, J. S., Clough, S. A., Payne, V. H., Mlawer, E. J., Turner, D. D., and
Gamache, R. R.: A far-infrared radiative closure study in the Arctic:
Application to water vapor, J. Geophys. Res.-Atmos.,
115, D17106, 10.1029/2009JD012968,
2010.Donovan, D. P. and van Lammeren, A. C. A. P.: Cloud effective particle size and
water content profile retrievals using combined lidar and radar observations
1. Theory and examples, J. Geophys. Res., 106, 27425–27448,
10.1029/2001JD900243, 2001.Dubuisson, P., Giraud, V., Pelon, J., Cadet, B., and Yang, P.: Sensitivity of
Thermal Infrared Radiation at the Top of the Atmosphere and the Surface to
Ice Cloud Microphysics, J. Appl. Meteor. Climatol., 47, 2545–2560,
10.1175/2008JAMC1805.1,
2008.Eastman, R. and Warren, S.: Arctic Cloud Changes from Surface and Satellite
Observations, . Climate, 23, 4233–4242,
10.1175/2010JCLI3544.1,
2010.Ebert, E. E. and Curry, J. A.: A Parameterization of Ice Cloud Optical
Properties for Climate Models, J. Geophys. Res., 97, 3831–3836,
10.1029/91JD02472, 1992.
Eloranta, E. W.: Lidar: Range-Resolved Optical Remote Sensing of the
Atmosphere, chap. High Spectral Resolution Lidar, Klaus Weitkamp editor,
Springer Series in Optical Sciences, Springer-Verlag, New York, 2005.Eloranta, E. W., Uttal, T., and Shupe, M.: Cloud particle size measurements in
Arctic clouds using lidar and radar data, in: 2007 IEEE International
Geoscience and Remote Sensing Symposium, 2265–2267,
10.1109/IGARSS.2007.4423292, 2007.Feingold, G. and McComiskey, A.: ARM's Aerosol-Cloud-Precipitation Research
(Aerosol Indirect Effects), Meteorol. Monogr., 57, 22.1–22.15,
10.1175/AMSMONOGRAPHS-D-15-0022.1,
2016.Garnier, A., Pelon, J., Dubuisson, P., Faivre, M., Chomette, O., Pascal, N.,
and Kratz, D. P.: Retrieval of Cloud Properties Using CALIPSO Imaging
Infrared Radiometer. Part I: Effective Emissivity and Optical Depth, J. Appl. Meteorol. Climatol., 51, 1407–1425,
10.1175/JAMC-D-11-0220.1,
2012.Girard, E., Dueymes, G., Du, P., and Bertram, A. K.: Assessment of the effects
of acid-coated ice nuclei on the Arctic cloud microstructure, atmospheric
dehydration, radiation and temperature during winter, Int. J.
Climatol., 33, 599–614, 10.1002/joc.3454,
2013.Grenier, P., Blanchet, J.-P., and Muñoz-Alpizar, R.: Study of polar thin ice
clouds and aerosols seen by CloudSat and CALIPSO during midwinter 2007,
J. Geophys. Res.-Atmos., 114, D09201,
10.1029/2008JD010927,
2009.Hansen, J. E. and Travis, L. D.: Light scattering in planetary atmospheres,
Space Sci. Rev., 16, 527–610,
10.1007/BF00168069,
1974.Heymsfield, A. J., Krämer, M., Luebke, A., Brown, P., Cziczo, D. J.,
Franklin, C., Lawson, P., Lohmann, U., McFarquhar, G., Ulanowski, Z., and
Tricht, K. V.: Cirrus Clouds, Meteorol. Monogr., 58, 2.1–2.26,
10.1175/AMSMONOGRAPHS-D-16-0010.1,
2017.Illingworth, A. J., Barker, H. W., Beljaars, A., Ceccaldi, M., Chepfer, H.,
Clerbaux, N., Cole, J., Delanoë, J., Domenech, C., Donovan, D. P.,
Fukuda, S., Hirakata, M., Hogan, R. J., Huenerbein, A., Kollias, P., Kubota,
T., Nakajima, T., Nakajima, T. Y., Nishizawa, T., Ohno, Y., Okamoto, H., Oki,
R., Sato, K., Satoh, M., Shephard, M. W., Velazquez-Blazquez, A., Wandinger,
U., Wehr, T., and van Zadelhoff, G.-J.: The EarthCARE Satellite: The Next
Step Forward in Global Measurements of Clouds, Aerosols, Precipitation, and
Radiation, B. Am. Meteorol. Soc., 96, 1311–1332,
10.1175/BAMS-D-12-00227.1,
2015.Inoue, T.: On the temperature and effective emissivity determination of
semi-transparent cirrus clouds by bi-spectral measurements in the 10 μ
window region, J. Meteorol. Soc. Jpn., 63, 88–99,
1985.IPCC: Climate Change 2013: The Physical Science Basis. Contribution of Working
Group I to the Fifth Assessment Report of the Intergovernmental Panel on
Climate Change, Cambridge University Press, Cambridge, United Kingdom and New
York, NY, USA, 10.1017/CBO9781107415324,
2013.Jiang, J. H., Su, H., Zhai, C., Perun, V. S., Del Genio, A., Nazarenko, L. S.,
Donner, L. J., Horowitz, L., Seman, C., Cole, J., Gettelman, A., Ringer,
M. A., Rotstayn, L., Jeffrey, S., Wu, T., Brient, F., Dufresne, J.-L., Kawai,
H., Koshiro, T., Watanabe, M., LÉcuyer, T. S., Volodin, E. M., Iversen, T.,
Drange, H., Mesquita, M. D. S., Read, W. G., Waters, J. W., Tian, B.,
Teixeira, J., and Stephens, G. L.: Evaluation of cloud and water vapor
simulations in CMIP5 climate models using NASA “A-Train” satellite
observations, J. Geophys. Res.-Atmos., 117, D14105,
10.1029/2011JD017237, 2012.Jouan, C., Pelon, J., Girard, E., Ancellet, G., Blanchet, J. P., and Delanoë, J.: On the relationship between Arctic ice clouds and
polluted air masses over the North Slope of Alaska in April 2008, Atmos. Chem. Phys., 14, 1205–1224, 10.5194/acp-14-1205-2014, 2014.Knuteson, R. O., Revercomb, H. E., Best, F. A., Ciganovich, N. C., Dedecker,
R. G., Dirkx, T. P., Ellington, S. C., Feltz, W. F., Garcia, R. K., Howell,
H. B., Smith, W. L., Short, J. F., and Tobin, D. C.: Atmospheric Emitted
Radiance Interferometer. Part I: Instrument Design, J. Atmos. Ocean. Technol., 21, 1763–1776, 10.1175/JTECH-1662.1,
2004a.Knuteson, R. O., Revercomb, H. E., Best, F. A., Ciganovich, N. C., Dedecker,
R. G., Dirkx, T. P., Ellington, S. C., Feltz, W. F., Garcia, R. K., Howell,
H. B., Smith, W. L., Short, J. F., and Tobin, D. C.: Atmospheric Emitted
Radiance Interferometer. Part II: Instrument Performance, J. Atmos. Ocean. Technol., 21, 1777–1789,
10.1175/JTECH-1663.1,
2004b.Legrand, M., Pietras, C., Brogniez, G., Haeffelin, M., Abuhassan, N. K., and
Sicard, M.: A High-Accuracy Multiwavelength Radiometer for In Situ
Measurements in the Thermal Infrared. Part I: Characterization of the
Instrument, J. Atmos. Ocean. Technol., 17, 1203–1214,
10.1175/1520-0426(2000)017<1203:AHAMRF>2.0.CO;2,
2000.Libois, Q., Proulx, C., Ivanescu, L., Coursol, L., Pelletier, L. S., Bouzid, Y., Barbero, F., Girard, É., and Blanchet, J.-P.:
A microbolometer-based far infrared radiometer to study thin ice clouds in the Arctic, Atmos. Meas. Tech., 9, 1817–1832, 10.5194/amt-9-1817-2016, 2016.Lubin, D.: Infrared Radiative Properties Of the Maritime Antarctic Atmosphere,
J. Climate, 7, 121–140,
10.1175/1520-0442(1994)007<0121:IRPOTM>2.0.CO;2,
1994.Mahesh, A., Walden, V. P., and Warren, S. G.: Ground-Based Infrared Remote
Sensing of Cloud Properties over the Antarctic Plateau. Part II: Cloud
Optical Depths and Particle Sizes, J. Appl. Meteorol., 40, 1279–1294,
10.1175/1520-0450(2001)040<1279:GBIRSO>2.0.CO;2,
2001.Meador, W. E. and Weaver, W. R.: Two-Stream Approximations to Radiative
Transfer in Planetary Atmospheres: A Unified Description of Existing Methods
and a New Improvement, J. Atmos. Sci., 37, 630–643,
10.1175/1520-0469(1980)037<0630:TSATRT>2.0.CO;2,
1980.Moran, K. P., Martner, B. E., Post, M. J., Kropfli, R. A., Welsh, D. C., and
Widener, K. B.: An Unattended Cloud-Profiling Radar for Use in Climate
Research, B. Am. Meteorol. Soc., 79, 443–455,
10.1175/1520-0477(1998)079<0443:AUCPRF>2.0.CO;2,
1998.Nakajima, T. and King, M. D.: Determination of the Optical Thickness and
Effective Particle Radius of Clouds from Reflected Solar Radiation
Measurements. Part I: Theory, J. Atmos. Sci., 47,
1878–1893, 10.1175/1520-0469(1990)047<1878:DOTOTA>2.0.CO;2,
1990.Platt, C. M. R.: Lidar and Radiometric Observations of Cirrus Clouds, J. Atmos.
Sci., 30, 1191–1204, 10.1175/1520-0469(1973)030<1191:LAROOC>2.0.CO;2,
1973.
Rodgers, C. D.: Inverse Methods for Atmospheric Sounding – Theory and Practice., Series: Series on Atmospheric Oceanic and Planetary Physics,
ISBN: 9789812813718, World Scientific Publishing Co. Pte. Ltd., edited by: Rodgers, C. D., vol.
2, 2000.Rowe, P. M., Miloshevich, L. M., Turner, D. D., and Walden, V. P.: Dry Bias in
Vaisala RS90 Radiosonde Humidity Profiles over Antarctica, J. Atmos. Ocean. Technol., 25, 1529–1541,
10.1175/2008JTECHA1009.1,
2008.
Royer, A., Cliche, P., Marchand, N., Roy, A., Blanchard, Y., O'Neill, N. T.,
Damiri, B., Leclercq, J., Meunier, S., and Crozel, D.: An Automatic Sky
Spectral Thermal Infrared Radiometer (ASSTIR) system, in: 4th Int. Symposium
on Recent Advancesin Quantitative Remote Sensing: RAQRS'IV,22-26th September
2014, Torrent (Valencia), Spain, 2014.Sassen, K., Wang, Z., and Liu, D.: Global distribution of cirrus clouds from
CloudSat/Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations
(CALIPSO) measurements, J. Geophys. Res.-Atmos., 113,
D00A12, 10.1029/2008JD009972,
2008.Shupe, M. D.: A ground-based multisensor cloud phase classifier, Geophys. Res.
Lett., 34, L22809,
10.1029/2007GL031008, 2007.Shupe, M. D., Walden, V. P., Eloranta, E., Uttal, T., Campbell, J. R.,
Starkweather, S. M., and Shiobara, M.: Clouds at Arctic Atmospheric
Observatories. Part I: Occurrence and Macrophysical Properties, J. Appl. Meteorol. Climatol., 50, 626–644,
10.1175/2010JAMC2467.1,
2011.Shupe, M. D., Turner, D. D., Zwink, A., Thieman, M. M., Mlawer, E. J., and
Shippert, T.: Deriving Arctic Cloud Microphysics at Barrow, Alaska:
Algorithms, Results, and Radiative Closure, J. Appl. Meteorol. Climatol., 54, 1675–1689, 10.1175/JAMC-D-15-0054.1,
2015.Smith, W. L., Ma, X. L., Ackerman, S. A., Revercomb, H. E., and Knuteson,
R. O.: Remote Sensing Cloud Properties from High Spectral Resolution Infrared
Observations, J. Atmos. Sci., 50, 1708–1720,
10.1175/1520-0469(1993)050<1708:RSCPFH>2.0.CO;2,
1993.Sourdeval, O., Labonnote, L. C., Brogniez, G., Jourdan, O., Pelon, J., and Garnier, A.: A variational approach for
retrieving ice cloud properties from infrared measurements: application in the context of two IIR validation campaigns,
Atmos. Chem. Phys., 13, 8229–8244, 10.5194/acp-13-8229-2013, 2013.Stephens, G. L.: Cloud Feedbacks in the Climate System: A Critical Review,
J. Climate, 18, 237–273, 10.1175/JCLI-3243.1,
2005.Stephens, G. L. and Webster, P. J.: Clouds and Climate: Sensitivity of Simple
Systems, J. Atmos. Sci., 38, 235–247,
10.1175/1520-0469(1981)038<0235:CACSOS>2.0.CO;2,
1981.Stephens, G. L., Tsay, S.-C., Stackhouse, P. W., and Flatau, P. J.: The
Relevance of the Microphysical and Radiative Properties of Cirrus Clouds to
Climate and Climatic Feedback, J. Atmos. Sci., 47, 1742–1754,
10.1175/1520-0469(1990)047<1742:TROTMA>2.0.CO;2,
1990.Turner, D. D.: Microphysical properties of single and mixed-phase Arctic clouds
derived from ground-based AERI observations, Ph.d. dissertation, University
of Wisconsin – Madison, Madison, Wisconsin,
available at: http://www.ssec.wisc.edu/library/turnerdissertation.pdf (last access: 6 June 2017),
2003.Turner, D. D.: Arctic Mixed-Phase Cloud Properties from AERI Lidar
Observations: Algorithm and Results from SHEBA, J. Appl. Meteorol., 44,
427–444, 10.1175/JAM2208.1,
2005.Turner, D. D. and Eloranta, E.: Validating Mixed-Phase Cloud Optical Depth
Retrieved From Infrared Observations With High Spectral Resolution Lidar,
IEEE Geosci. Remote Sensing, 5, 285–288,
10.1109/LGRS.2008.915940, 2008.Turner, D. D. and Löhnert, U.: Information Content and Uncertainties in
Thermodynamic Profiles and Liquid Cloud Properties Retrieved from the
Ground-Based Atmospheric Emitted Radiance Interferometer (AERI), J. Appl. Meteorol. Climatol., 53, 752–771,
10.1175/JAMC-D-13-0126.1,
2014.Turner, D. D., Ackerman, S. A., Baum, B. A., Revercomb, H. E., and Yang, P.:
Cloud Phase Determination Using Ground-Based AERI Observations at SHEBA, J.
Appl. Meteorol., 42, 701–715,
10.1175/1520-0450(2003)042<0701:CPDUGA>2.0.CO;2,
2003.Turner, D. D., Knuteson, R. O., Revercomb, H. E., Lo, C., and Dedecker, R. G.:
Noise Reduction of Atmospheric Emitted Radiance Interferometer (AERI)
Observations Using Principal Component Analysis, J. Atmos. Ocean. Technol.,
23, 1223–1238, 10.1175/JTECH1906.1,
2006.Waliser, D. E., Li, J.-L. F., Woods, C. P., Austin, R. T., Bacmeister, J.,
Chern, J., Del Genio, A., Jiang, J. H., Kuang, Z., Meng, H., Minnis, P.,
Platnick, S., Rossow, W. B., Stephens, G. L., Sun-Mack, S., Tao, W.-K.,
Tompkins, A. M., Vane, D. G., Walker, C., and Wu, D.: Cloud ice: A climate
model challenge with signs and expectations of progress, J. Geophys. Res.-Atmos., 114, D00A21, 10.1029/2008JD010015,
10.1029/2008JD010015, 2009.
Wang, X. and Key, J. R.: Recent Trends in Arctic Surface, Cloud, and Radiation
Properties from Space, Science, 299, 1725–1728,
2003.Warren, S. G. and Brandt, R. E.: Optical constants of ice from the ultraviolet
to the microwave: A revised compilation, J. Geophys. Res., 113, D14220,
10.1029/2007JD009744, 2008.Wendisch, M., Yang, P., and Pilewskie, P.: Effects of ice crystal habit on
thermal infrared radiative properties and forcing of cirrus,
J. Geophys. Res.-Atmos., 112, D08201, 10.1029/2006JD007899,
2007.Winker, D. M., Pelon, J., Jr., J. A. C., Ackerman, S. A., Charlson, R. J.,
Colarco, P. R., Flamant, P., Fu, Q., Hoff, R. M., Kittaka, C., Kubar, T. L.,
Treut, H. L., McCormick, M. P., Mégie, G., Poole, L., Powell, K., Trepte,
C., Vaughan, M. A., and Wielicki, B. A.: The CALIPSO Mission: A Global 3D
View of Aerosols and Clouds, B. Am. Meteorol. Soc.,
91, 1211–1229, 10.1175/2010BAMS3009.1,
2010.Wylie, D. P. and Menzel, W. P.: Eight Years of High Cloud Statistics Using
HIRS, J. Climate, 12, 170–184,
10.1175/1520-0442(1999)012<0170:EYOHCS>2.0.CO;2,
1999.Yang, P., Mlynczak, M. G., Wei, H., Kratz, D. P., Baum, B. A., Hu, Y. X.,
Wiscombe, W. J., Heidinger, A., and Mishchenko, M. I.: Spectral signature of
ice clouds in the far-infrared region: Single-scattering calculations and
radiative sensitivity study, J. Geophys. Res.-Atmos.,
108, 4569, 10.1029/2002JD003291,
2003.Yang, P., Tsay, S.-C., Wei, H., Guo, G., and Ji, Q.: Remote sensing of cirrus
optical and microphysical properties from ground-based infrared radiometric
Measurements-part I: a new retrieval method based on microwindow spectral
signature, IEEE Geosci. Remote Sensing, 2, 128–131,
10.1109/LGRS.2005.844733, 2005.
Yang, P., Bi, L., Baum, B. A., Liou, K.-N., Kattawar, G. W., Mishchenko, M. I.,
and Cole, B.: Spectrally Consistent Scattering, Absorption, and Polarization
Properties of Atmospheric Ice Crystals at Wavelengths from 0.2 to 100 μm,
J. Atmos. Sci., 70, 330–347,
10.1175/JAS-D-12-039.1,
2013.