GIGeoscientific Instrumentation, Methods and Data SystemsGIGeosci. Instrum. Method. Data Syst.2193-0864Copernicus PublicationsGöttingen, Germany10.5194/gi-7-39-2018The World Optical Depth Research and Calibration Center
(WORCC) quality assurance and quality control of GAW-PFR AOD measurementsKazadzisSteliosstelios.kazadzis@pmodwrc.chKouremetiNataliaNyekiStephanGröbnerJulianWehrliChristophPhysikalisch-Meteorologisches Observatorium Davos, World Radiation
Center (PMOD/WRC) Dorfstrasse 33, 7260 Davos Dorf, SwitzerlandStelios Kazadzis (stelios.kazadzis@pmodwrc.ch)2February201871395310October201723October201719December20171January2018This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/4.0/This article is available from https://gi.copernicus.org/articles/7/39/2018/gi-7-39-2018.htmlThe full text article is available as a PDF file from https://gi.copernicus.org/articles/7/39/2018/gi-7-39-2018.pdf
The World Optical Depth Research Calibration Center
(WORCC) is a section within the World Radiation Center at
Physikalisches-Meteorologisches Observatorium (PMOD/WRC), Davos, Switzerland,
established after the recommendations of the World Meteorological Organization for calibration of aerosol optical depth (AOD)-related
Sun photometers. WORCC is mandated to develop new methods for instrument
calibration, to initiate homogenization activities among different AOD
networks and to run a network (GAW-PFR) of Sun photometers. In this work we
describe the calibration hierarchy and methods used under WORCC and the
basic procedures, tests and processing techniques in order to ensure the
quality assurance and quality control of the AOD-retrieved data.
Introduction
Aerosols in the atmosphere, through direct and indirect effects, mainly
result in a cooling contribution to the global radiation balance (IPCC,
2013). The parameter that describes their integrated optical attenuation is
the aerosol optical depth (AOD), which can be derived by measurements of the
sunlight transmittance (WMO, 2016b). AOD has been used in case studies and local studies
in order to characterize aerosols and assess atmospheric pollution and the
aerosol-related radiative forcing.
Atmospheric extinction of sunlight has been studied for at least 250 years by
P. Bouger (http://glossary.ametsoc.org/wiki/Bouguer_s_law). Linke (1942) studied turbidity,
Angström (1929) studied
extinction power law and Junge (1952) studied the relationship between particle
volume and aerosol number size distribution. They have mainly set the theoretical
basis for studying aerosol extinction. However, Volz (1959) have developed a
Sun photometer that is able to measure atmospheric turbidity at different wavelengths
using filters, used in the first US (Volz, 1969) and the first European
(Flowers, 1969) networks of turbidity measurements. Since then various sites
have included AOD measurements in their monitoring schedules, constructing long-term
series of AOD (e.g., Barreto et al., 2014; Weller and Gericke, 2005;
Nyeki et al., 2012). Most of these measurements are site specific, with
little relevance to long-term trend analysis on a global scale. However, more
recently, several multiyear spatial studies (Holben, 2001; Che et al., 2015;
Mitchell et al., 2017) have been conducted.
The World Meteorological Organization (WMO) instigated the Global Atmosphere
Watch (GAW) program in 1989 as a successor to the Background Air Pollution
Monitoring Network (BAPMoN). In 1993 (WMO, 1993) it was recommended that AOD
measurements, conducted previously under BAPMoN, should be discontinued until
new instruments, methods and protocols could be established to collect AOD
data of known and assured quality. Based on a recommendation by GAW experts,
the World Optical Depth Research Calibration Center (WORCC) was established
in 1996 at the PMOD/WRC in Switzerland. WORCC has since been advised by the
GAW Scientific Advisory Group for Aerosols. Fifteen existing GAW stations
were chosen for the deployment and operation of 12 N-type precision
filter radiometers (PFRs; manufactured by PMOD/WRC) (Wehrli, 2000), provided
by the Swiss Government. WORCC was assigned the following tasks:
development of a radiometric reference for spectral radiometry
to determine AOD
development of procedures to ensure worldwide homogeneity of AOD
observations
development of new instrumentation and methods for AOD
implementation of a pilot network for AOD at GAW global
observatories including quality control and quality assurance of data
called GAW-PFR
training operators to use and maintain AOD instruments.
There are different global networks measuring AOD, mainly distinguished by
the different instruments used in each of them. The AErosol RObotic NEtwork
(AERONET) (Holben et al., 1998, 2001) (http://aeronet.gsfc.nasa.gov/) is
the major global network with central calibration facilities in the USA,
France and Spain. The sky radiometer network (SKYNET) aerosol network
(Takamura and Nakajima, 2004; Campanelli, 2004) is an observational network dedicated to
aerosol–cloud–radiation interaction research studies. The Australian AOD
network includes 22 stations (Mitchell et al., 2017). The Chinese Aerosol
Remote Sensing network (CARSNET), reporting AOD measurements for 50 sites,
representing remote, rural and urban areas (Che et al., 2015). In addition,
national, regional and global networks such as the French component of
AERONET, PHOTONS (Goloub et al., 2007), the Iberian Network for aerosol
measurements (RIMA) (Toledano et al., 2011) and Aerosol Canada (AEROCAN)
(Bokoye et al., 2001) have contributed to AOD climatology studies.
The Swiss-built PFR (Wehrli in WMO, 2005) has been operating continuously at
15 GAW stations and at another 23 associated ones worldwide. The PFR is
expressly designed to make automated long-term observations at four
wavelengths (368, 412, 500 and 862 nm). Several studies using data from the
GAW-PFR network have been published mainly focusing on long-term changes in
AOD (e.g., Ruckstuhl et al., 2008; Nyeki et al., 2012, 2015). In addition to
these studies, GAW-PFR aims to provide intercomparison information between
networks by overlapping at selected sites
PFR instruments of the GAW-PFR network currently overlap with AERONET,
SKYNET, CARSNET, the Australian Network and other Sun photometers at several
sites. As there is a need for a common strategy to merge the various network
observations into a global data set, the WMO-GAW scientific advisory group
for aerosols recommended that GAW will have to collaborate with existing
major networks to develop this strategy, implementing and developing,
together with satellite agencies, a system for integrating global AOD
observations. Towards the same goal, WORCC organizes a filter radiometer
comparison every 5 years with a number of reference
AOD-measuring instruments from different networks. The last comparison was
held in 2015 with the participation of 30 instruments (WMO, 2016a). In this
work, we present the research activities of WORCC and more specifically the
calibration hierarchy, the quality assurance and quality control of the
GAW-PFR network AOD measurements.
The characterization and calibration of the PFR instruments, together with the
quality assurance and quality control of the measurements, is WORCC's major
task. In this work, we summarize the calibration procedures and hierarchy
since the GAW-PFR network was established 17 years ago and the quality
assurance and control of the data that the instruments provide.
Calibration principles and hierarchy
AOD is a dimensionless quantity that cannot be measured directly. It can be
retrieved from atmospheric transmission measurements and cannot be directly
linked to any SI (International System of Units) reference since the
atmospheric transmission is also a relative factor related to the direct
solar irradiance (I) at a particular wavelength (λ) and
I(λ) at the surface and at the top of the atmosphere
Io(λ, r), where r is the Sun–Earth distance. As a
consequence, transmission can be measured in any unit and in the case of
Sun photometers, the instrument voltage signal V(λ) and the signal
at the top of the atmosphere (extra-terrestrial value Vo(λ, r))
can be written based on the Beer–Lambert law:
V(λ,m,R)=Vo(λ)e-mδ(λ)R-2+ε
where Vo is the exoatmospheric signal at wavelength λ and standard
Sun–Earth distance of 1 astronomical unit, m is the optical air
mass along the line of sight to the Sun, δ is the total optical
depth, R is the Sun–Earth distance in astronomical units, and ε accounts for the circumsolar sky radiance in the field of view of the Sun
photometer. The total optical thickness m×δ includes several terms describing the extinction by
different atmospheric components: molecular scattering, gas absorption and
aerosol extinction. Then in order to calculate the AOD we use the following:
τaer=lnVo-ln(V)m-τrt,
where τaer is the AOD, m is the optical air mass and
τrt is the attenuation due to Rayleigh scattering and other
trace gases for cloudless conditions. Using Eq. (2), we conclude that an
error of 1 % in Vo(λ) results in an AOD of 0.01 for an air
mass equal to 1.
According to WMO (2005), as traceability is not currently possible based on
physical measurement systems, the initial form of traceability will be based
on difference criteria. That is, at an intercomparison or co-location,
traceability will be established if the difference between the AOD of one network
and another is within specific limits. Those limits for finite field-of-view
instruments have been set (WMO, 2005) to optical depths of 0.005 + 0.01 m-1
and the acceptable traceability is when 95 % of the
absolute AODs are within those limits. So requiring 95 % uncertainty (U95)
within optical depths of ±0.005 + 0.01 m-1, where the first term
(0.005) is linked to instrument uncertainties (signal linearity, Sun
pointing, temperature effects, processing, etc.) and the second term to a
calibration uncertainty of 1 %.
(a) Difference between each PFR with the mean of the PFR triad
at the four PFR wavelengths. Grey areas represent the WMO-U95 limits.
(b) Frequency distribution of these differences for the four
measuring wavelengths.
The WORCC standard group of three PFRs (defined as the “PFR triad”) was
established in 2005 by WORCC in order to fulfill the WMO mandate on
“homogenization of global AOD through provision of traceability to the World
Standard Group (WSG) of spectral radiometers for contributing networks at
co-located sites and/or periodic international filter radiometer comparisons,
and further standardization of evaluation algorithms”. Since 2005, five
different well-maintained instruments have been used as part of the PFR
triad. Figure 1 shows the long-term (12 years) comparison of the PFR triad
instruments.
Details of PFR triad Langley calibration measurements.
YearTriad ComparisonCalibrationComparison period referencetype2003N01N26N26MLO-Langley1 Mar 200031 May 20032005N01N26N27N27MLO-Langley1 Sep 200531 Dec 20052009N01N26N27N25IZO-Langley1 Apr 200931 Jun 20092010N01N25N27N24JFJ-Langley1 Jan 201031 Jan 2010N01N25N27N22MLO-Langley1 Jun 201031 Jun 20102011N01N25N27mean ofN01, N25, N272012N01N25N27N21IZO-Langley1 Oct 201231 Dec 20122013N01N25N27N06IZO-Langley1 Aug 201331 Aug 20132014N01N25N27mean ofN01, N25, N272015N01N25N27N06IZO-Langley21 Sep 201528 Sep 2015N01N25N27N21IZO-Langley21 Sep 201528 Sep 2015N01N25N27N24MLO-Langley21 Sep 201528 Sep 20152016N24N25N27N06IZO-Langley01 Oct 201631 Dec 20162017N24N25N27N21IZO-Langley17 Mar 201714 Apr 2017
The long-term relative stability of each of the five PFRs that were part of
the triad is presented in Fig. 1. The left panel shows the 1 min AOD PFR
differences, compared using the WMO-U95 criterion at all four PFR-measuring
wavelengths. It should be noted that all instruments measure at WORCC
in Davos, Switzerland. They are mounted on the same solar tracking system and
their signal is processed using a common processing algorithm. In the
12 years of 1 min measurement data, more than 99 % of retrieved AOD lies
within the U95-WMO criterion, at all wavelengths. The right panel of Fig. 1
shows the individual instrument comparisons with the mean
triad AOD in more detail. As shown, all differences are well within ±0.005 with small
shifts for different PFRs and particular wavelengths.
In order to continuously check and maintain the triad stability we have
defined a calibration protocol including instruments frequently
performing Langley calibrations at high-altitude stations. For this
calibration method (Holben et al., 1998; Michalsky et al., 2001) the main
requirement is the stability of AOD during the measurement Langley periods
(half days). Theoretically, this can be achieved anywhere, but in practice
AOD is variable during the day, so the current practice is to perform such
measurements at high-altitude locations where AOD is very low; thus its
variability is very small on an absolute level. Since 2003, Mauna Loa, Hawaii,
USA (MLO, 19.5∘ N, 155.6∘ W, 3397 m a.s.l.), Izana,
Tenerife, Spain (IZO, 28.3∘ N, 16.5∘ W, 2370 m a.s.l.) and Jungfraujoch, Switzerland (JFJ, 3580 m a.s.l. 46.5∘ N,
7.9∘ E), have mainly been used for such Langley calibrations. PFR
instruments have been permanently deployed at these stations for certain
periods since 2003, and approximately every 6 months, one of these
instruments is returned to WORCC to perform synchronous measurements in
parallel with the triad. Table 1 lists the details of these visits. For each period it
describes the current status of the PFR instruments of the
triad, the transferred instrument performing the Langley plots and the
Langley plot measurement location and period.
Langley plots for Mauna Loa observatory from November 2015 to
April 2016.
The determination of Vo with the Langley calibration method using a
6-month period of measurements requires high accuracy and quantification of
the introduced uncertainties. Using a defined calibration method, the Vo
accuracy can be traced back to the variability of the Vo determination
and is related to the instrument precision and the procedures. Practically,
the long-term stability of Vo is mainly related to degradation/changes in
the transmission of the optical interference filters, or hardware-related
failures/changes that are linked to changes in the instrument signal.
WORCC Langley algorithms use half-days to determine Vo values. The main
requirements for accepting a half-day Langley determination of Vo are the
AOD stability, the signal stability and the statistics of the retrieved
signal versus air mass linear regression using specific air mass limitations.
An example of accepted Langley (half-day) measurements for a 6-month period
at MLO is shown in Fig. 2.
Calibration uncertainties derived from 6 months of Langley calibration measurements shown in Fig. 2.
Long-term Vo values for PFR-N06 at Davos and IZO for the
four PFR channels. Bars indicated the standard error.
Figure 2 shows 95 Langley diagrams/days that have been used to analyze
Langley calibration results and related uncertainties. The mean ln(Vo)
calculated for this period at 500 nm was 1.343, the standard deviation was
0.002 and the 5th and 95th percentiles were 1.340 and 1.346, respectively.
The distribution of ln(Vo) values is also shown with the statistics
for mean AOD values (0.010–0.015 at 500 nm). The distribution and the
normal distribution are shown in the upper-left subplot of Fig. 2.
Based on Eq. (1) the AOD absolute uncertainty, δAODVo that is
related only to the Langley calibration factor, using Eq. (2), equals
δln(Vo)m where δln(Vo) is the uncertainty in
ln(Vo). The uncertainty of ln(Vo) can be described by the coefficient of
variation (standard deviation/mean, CV) or in the case of a normal
distribution by the standard error (standard deviation divided by the square
root of the number of measurements, SE). For the particular example in
Fig. 2, the calibration uncertainties are shown in Table 2.
As described above, this uncertainty is directly related to the calculated
δAODVo uncertainty and is equal to δln(Vo) when
m=1. In Fig. 3, Vo values at 500 and 865 nm are shown as a function of
time for measurements of the PFR instrument N06 that measured at Davos
from 2000 to 2005 and at IZO from 2005 to 2016. In addition, the evolution of
the SE is shown.
Each of the data points in Fig. 3 represents average Vo values at the end
of the averaging period which varies between 3 and 6 months. PFR N06 has
exhibited good stability since 2000. All instrument filters have not changed
more than 0.1 in Vo units, which corresponds to maximum changes in AOD of
∼ 0.02 when m=1. However, the 412 nm filter is an exception which
has apparently degraded since 2009. The maximum changes in AOD (at m=1) were ∼ 0.05 from 2009 to 2016. It has to be noted that all the
above-described changes have been taken into account in order to calculate
the corresponding AOD for the individual periods. Results in Fig. 3
illustrate the stability of PFR instruments over time. The very low filter
response changes over long-term periods increase the statistical validity of
each 6-month Langley calibration period.
GAW-PFR station details, location characteristics and AOD time-series information.
Station(abbreviation)LatLongAltitude (m)CountryType of locationMain types ofAir massesPFR AODTime seriesPrevious studiesAlice Springs(ASP)23.80∘ S133.87∘ E547Australiadesertremote continental2002 to presentMitchell et al. (2017)Bratts Lake (BRA)50.28∘ N104.70∘ W576Canadaprairie, agriculturalremote continental2001–2012McArthur et al. (2003)Danum Valley(MAL)4.98∘ N117.84∘ E436Malaysiatropical forestremote continental2007–2016Hohenpeissenberg (HPB)47.80∘ N11.02∘ E995Germanypre-alpine, ruralrural1999 to presentRuckstuhl et al. (2008),Nyeki et al. (2012)Izana (IZO)28.31∘ N16.50∘ E2371Spainislandfree troposphere2001 to presentBarreto et al. (2014)Jungfraujoch (JFJ)46.55∘ N7.99∘ N3580Switzerlandhigh alpinefree troposphere1999 to presentRuckstuhl et al. (2008),Nyeki et al. (2012)Mauna Loa (MLO)19.53∘ N155.58∘ W3397USAislandfree troposphere2000 to presentDutton et al. (1994)Mace Head (MHD)53.33∘ N9.89∘ E20Irelandcoastmarine boundary layer2000–2015Mulcahy et al. (2009)Ny Ålesund (NYA)78.91∘ N11.88∘ N17SvalbardArctic coast, islandArctic/marineboundary layer2002 to presentHerber et al. (2002)Ryori (RYO)39.03∘ N141.83∘ E230Japancoastmarine boundary layer2002 to presentCape Point (CPT)34.35∘ S18.49∘ E230S. Africacoastmarine boundary layer2007 to presentNyeki et al. (2015)Mt. Waliguan(WLG)36.28∘ N100.90∘ E3810Chinahigh mountainfree troposphere2007 to presentChe et al. (2011)Valentia (VAL)51.94∘ N,10.24∘ W24Irelandcoastmarine boundary layer2007 to presentMarambio (MAR)64.24∘ S56.62∘ W205Argentinacoastmarine boundary layer2005 to presentTomasi et al. (2015)Troll (TRO)72.01∘ S,2.54∘ E1309Antarcticapolarfree troposphere2012 to presentTomasi et al. (2015)GAW-PFR network
A primary task of WORCC is the implementation of a global trial network at
selected GAW stations with the objective of demonstrating that PFR
instruments, together with standard calibration techniques and quality
assurance procedures, can be used to determine AOD with a precision adequate
for the fulfilment of the objectives of GAW (WMO, 2001). In addition, it is
intended that long-term high-resolution AOD measurements are conducted and
analyzed at selected GAW locations.
The locations, together with their characteristics in terms of aerosol sources
and their period of measurements, are described in Table 3. Bratts Lake and
Mace Head measurements were unfortunately discontinued in 2012 and 2015,
respectively, due to logistical aspects. However, Valentia (Ireland), Troll
(Antarctica) and Marambio (Argentina) have since been added to the core of
GAW-PFR stations.
In addition to the core GAW-PFR instruments, 30 other locations exist that
perform AOD measurements using PFR instruments belonging to individual
users and institutes. An overview of data flow and availability for every
location is shown in Fig. 4.
Data coverage of GAW-PFR stations. Green indicates periods with data
availability and red indicates missing data.
Calibration of an instrument against the triad: (a)
measurement signals at four wavelengths, (b) comparison of
instrument signals, (c) differences between instrument- and triad-retrieved AOD using the old and the new calibration V0xR.
At high-latitude locations such as Troll and Ny Ålesund, AOD
measurements can only be performed during part of the year due to the luck
of direct Sun. Big gaps (red colors) are linked to instrument damage (e.g., MHD, DMV and CPT) due to various factors (corrosion, lightning etc.).
Smaller (red) gaps are due to instrument recalibrations through transfer and
measurements in parallel with the PFR triad at Davos, Switzerland.
Instrument calibrations
Instruments are regularly calibrated (every 1 to 2 years, depending on
instrument-related and logistical aspects). The calibration of the filter
radiometer has to be assured with an uncertainty of ±1 % in order to
achieve the required AOD uncertainty within the U95-WMO limits. Quality-assured AOD data can only be obtained when pre- and post-deployment
calibration constants are available. That means that AOD data for a certain
location and period can only be considered as final after the recalibration
of the instrument, which is performed at the end of the specific period.
Graphs of V0U95 calculated over an extended calibration
period of 35 days for a PFR instrument measuring against each of the three
triad instruments for the four PFR wavelengths.
(a) Daily mean pressure during 2016 at Davos, Switzerland.
(b) Brewer and OMI ozone values.
Postcalibrations can be obtained by different methods:
WORCC calibration certificate
The WORCC calibration certificate is obtained by instruments/stations that have their PFR calibrated at Davos,
against the WORCC triad. Polar and high-altitude stations are often set for
calibration annually during polar night/longest night periods. For other
stations, a re-calibration should be performed every 12 to 24 months. This
method implies that preparation of the final version of the AOD data might
be postponed by 1 to 2 years. Comparison of each of the instruments with the
triad is performed based on the WMO criteria for AOD intercomparisons. The
intercomparison lasts for at least 5 cloudless days.
The extraterrestrial calibration constants of an instrument x, V0xR are
determined by using 1 min measurements Six of the instrument to be
calibrated and synchronous measurement SiR of the reference (mean of
the three triad PFRs) instrument.
V0xR=1NV0R∑n=1NSixSiR
The daily calibration constants V0xR, according to (Eq. 4), are determined
for days on which a number (N≥120) of solar measurements unobstructed by
clouds were collected. Comparing the instrument measurement signal to that of
the triad, a new V0xR is calculated and compared with the last one used.
Details of such an analysis for a single day are shown in Fig. 5.
Instrument signals are shown in Fig. 5a, then the percentage differences of
1 min data are calculated for all four wavelengths in Fig. 5b. An
example of the calculated AOD differences from the triad before and after the
calibration is finally shown in Fig. 5c. For the specific instrument, the
V0xR differences for the particular day were up to 0.5 % and
depended on the wavelength. The impact of this difference in AOD calculation
is an air-mass-dependent difference of 0.008 to 0.003. In practice, when an
instrument is calibrated against the triad, the only limitation on using the
synchronous signals is the cloud presence. So, no air mass or AOD limits are
included.
The average over the number of measurements over a day, values
V‾0xR and standard deviations σ0x of the
daily mean calibration constants from each reference instrument are averaged
to give the final calibration constants V0x with an expanded
uncertainty U95:
U95=1.96∑i=1NrefV‾0xR-V0x2Nref2+∑i=1Nrefσ0xRNdays2.
The two terms under the square root in Eq. (4) describe the combined
statistical (comparison) and triad uncertainty during the calibration
period.
For a normal distribution, V0U95 corresponds to a
coverage probability of approximately 95 %.
The calibration is considered successful when the coefficient of variation CV
=V0U95/V0x becomes smaller than
±0.5 % for all four channels of the instrument to be calibrated. This
limit is typically reached after 3 to 5 days of comparison. In Fig. 6, we
show V0U95 calculated over an extended calibration
period of 35 days for a PFR instrument measuring against the triad. For each
of the four PFR wavelengths, 2σ within the period is on the order of
0.4 to 0.9 % of the mean Vo. In addition, we have calculated
the V0U95 in percent using Eq. (3) for the
instrument under calibration with each of the three PFRs that are part of the
WORCC triad. For this particular case, all wavelengths are within the limit
CV<0.4 % after 3 days and after 10 days within CV<0.2 %.
Langley sites
Calibrations can be obtained by statistical analysis of objective Langley
plots collected in situ over an extended period of time at high-altitude
(IZO, JFJ, MLO) or remote background (ASP, BRA, TRO) sites. Such an
evaluation of Langley plots is routinely performed every 6 months using
Langley results from 6 months before and after the anchor dates of 1
January and 1 July for each year. This method implies that annual quality-assured data can be obtained in July or August of the following year.
A smaller (< 1 %) calibration uncertainty can be expected and is
required for Langley sites where AOD is lower than elsewhere, and inconsistent
calibration may lead to erroneous conditions such as an inverted
Ångström relation (channels crossing over) or negative AOD.
Calibration Vo values calculated for the four AOD channels are used in
order to retrieve the AOD. For the data obtained between two calibrations a
calibration slope of Vo values is applied. If the differences between two
calibrations are larger than 2 % then an in situ estimate of the
instrument stability is investigated from a number of in situ Langley plots
or cross-calibrations between different PFR channels. This is conducted in
order to determine nonlinear (over time) changes (steps) of instrumental
Vo values.
Example of good and bad pointing. (a) Instrument pointing,
(b) instrument signal at 500 nm. The two lower-right panels include
some cloud-related signal changes after 15:00 h
UT.
Quality control
After finalizing the calibration constants to be used for the AOD retrieval,
a series of QA/QC procedures are used before finalizing the AOD data.
Check for ancillary data
Ancillary data of atmospheric pressure and total ozone are needed to retrieve
the Rayleigh and ozone optical thicknesses, respectively, according to
Eq. (1). Atmospheric pressure measurements, required for Rayleigh scattering,
should be accurate to about 3 hPa. This accuracy is readily achieved by
meteorological grade barometers built into new PFR loggers. Accurate pressure
data are requested from each station and compared to the daily PFR logger
values. If the mean differences are larger than ∼ 3 hPa, then the
atmospheric pressure is corrected and all data are reprocessed. The use of
average atmospheric pressure data over a day or longer periods can lead to
wavelength-dependent AOD retrieval errors and to large Ångström
exponent errors.
Total column ozone values are needed to correct optical depth at 500 nm for
ozone absorption. As the absorption coefficient at 500 nm is low, total
ozone needs to be known to ±30 Dobson units or 10 % of typical
values, for an uncertainty of ±0.001 optical depths at 500 nm. GAW-PFR
uses (AURA) satellite overpass observations with the Ozone Monitoring
Instrument (OMI) for daily operations (McPeters et al., 2015). OMI values are
validated to in situ observations for stations operating a Dobson or Brewer
instrument. Where available, total column ozone may be found at the World
Ozone and Ultraviolet Radiation Data Centre database
(http://woudc.org/). Figure 7
shows the evolution of daily mean pressure used for the Rayleigh calculations
and ozone values at Davos, Switzerland as measured in situ with a Brewer
spectrophotometer and from the OMI ozone retrieval.
Corrections for temperature, dark signal
The PFR sensor temperature is checked for deviations from its active
stabilized set point, indicating potential problems during extremely hot or
cold ambient conditions. The PFR dark signal is checked for values
> 0.25 mV, and if found on approximately 5 % of days, a
correction is applied. The dark signal is the mean signal when the solar
elevation is less than -6∘, i.e., below the horizon. The temperature
dependence of its PFR is based on characterization measurements in a climate
chamber. Corrections are applied only in cases in which the dependence on Vo
is more than ±2 % for the range from -20 to 40 ∘C.
Sun pointing
In order to ensure that the full solar disk is included in the field of view
(FOV) of the instrument an accurate Sun-tracking system is required. While a
PFR instrument can be readily aligned to the Sun with the required accuracy,
a solar-pointing monitor of the PFR is included in order to control the
Sun-pointing accuracy. This monitor consists of a four-quadrant silicon
detector that is illuminated through a pinhole of 1 mm diameter at a
distance of 70 mm. When the light is centered, all four quadrants produce
equal signals. By subtracting signals from paired pixels, the sunspot can be
localized (Wehrli, 2008).
Figure 8 shows four examples of perfect to bad instruments pointing at four
cloudless, low aerosol concentration (AOD < 0.1) days. The
instrument pointing is shown in the upper plots. The first (from the left)
case shows a perfect pointing accuracy where 328 measurements during the day
are almost identical in terms of pointing direction. The second case shows
when all measurements can be found inside the 15 arcmin limit. Finally, the
last two cases show instruments with pointing issues where only 62.9
and 59.5 % cases respectively are inside the 15 arcmin limit, and in
the last case a number of measurements are outside the 25 and 30 arcmin
limit.
The result of the measured signal (which has a direct impact on calculated
AOD) is shown in the lower panel, where the first two days show a very
smooth daily pattern, while artificial signal features can be seen in the
last two cases. The use of these data will end up in an artificially
overestimated AOD retrieval.
Check for crossing of wavelengths
An additional quality control check detects instances when AOD at one of the
four PFR wavelengths is less than that at a higher wavelength. This quality
check is mainly performed in order to detect the erroneous performance of
one of the four channels. The test also becomes very important when low AOD
values are measured, which is the case at a number of GAW-PFR stations.
Small errors related to the calibration of one channel can be easily
identified, as this results in a wavelength crossing of AOD. During
evaluation of the data, the processing software tools give the opportunity
to define the limits of the accepted offset for the wavelength crossing.
An example of the graphic representation of wavelength crossing is shown in
Fig. 9. Based on the fact that AOD at lower wavelengths has to be at least
equal to or higher than that at higher wavelengths, colored points show the
correlation of pairs of AOD values at different wavelengths and black points
represent the cases in which wavelength crossing occurs. The figure is composed
of 70 350 1 min, cloudless measurements from the year 2015 as recorded at
Cape Point, South Africa. For example, as many as 14.3 % of data in the
412–500 nm panel (red-blue section) do not pass the wavelength crossing
test, so they are discarded form the postprocessing analysis.
Other issues
The spectral bandpasses of all PFR instruments have been characterized for their
effective spectral wavelength and bandwidth. This is determined as the
average wavelength weighted by the spectral response and equivalent width of
a rectangular bandpass with equal throughput as the filter. Minimum and
maximum central wavelengths (and bandwidths) that have been calculated were
367.2–367.7 (3.5–3.7), 411.8–412.6 (4.3–4.4), 500.6–501.5 (5–5.1) and
861.3–863 (5.5–5.6). These measurements were performed using illumination
by a grating monochromator (Jobin Yvon HR640) with 0.6 nm spectral
resolution. Lately, a pulsed tuneable laser system for the characterization
of spectrometers and filter radiometers has been available at PMOD/WRC. Test
measurements with PFR instruments did not show significant differences from the
older characterization measurements.
Correlation curves of the four PFR-measuring wavelengths. Colored
points represent the data that have passed the wavelength crossing test and
black ones are those that have not.
Example of a day with variable cloudiness, (a) instrument
signal at 500 nm and minute-by-minute application of the three cloud-flagging
methods. The two inset pictures show a 360∘ view of the sky using a
cloud camera. (b) Calculation of AOD at four wavelengths.
Monthly means and 1σ standard deviations for Davos,
Izaña and Mauna Loa using 15 years of 1 min quality-controlled PFR
measurements at four wavelengths.
Cloud flagging
As AOD measurements cannot be performed under cloudy conditions, a cloud
detection algorithm is used for the PFR measurements. Three different
criteria are used (Wehrli, 2008):
The instrument signal derivative with respect to air mass is always
negative. The method has been developed and described in detail by Harrison
and Michalsky (1994). For cases in which air mass values < 2 and the
influence of clouds on the noon-side of perturbations cannot be easily
detected, we compare the derivative with the estimate of the clear Rayleigh
atmosphere and flag it as cloudy if the rate of change is twice as much
(objective method).
The use of a test for optically thick clouds with AOD500nm
> 2.
The use of the Smirnov triplet measurement (Smirnov et al., 2000) by
calculating AOD and looking at the signal variability for 3 consecutive
minutes (triplet method).
An example of the use of these three criteria can be seen in Fig. 10, where a
day with variable cloudiness at Davos is presented.
For this particular day, all three criteria are applied. In the early morning
and evening, the thick-cloud criteria are applied. Then both the triplet and
the objective method are applied due to variable cloudiness in front of the
Sun. However, there are times during the day when only the objective method
is applied (thin clouds in front of the Sun as seen in the first picture that
is superimposed in Fig. 10). During the last part of the day (second
picture), clouds completely disappear and cloud flagging is set to zero, which
means that all three criteria are passed. It has to be noted that cloud
flagging is always kept as a constant number describing which one of the
three criteria or combination of criteria is valid at a certain minute.
The lower panel shows the calculation of AOD for the whole day, with obvious
deviations due to cloud occurrence for the parts of the day when both
criteria are fulfilled. It is interesting to see the 10:50 to 12:00 period
which is a difficult period when defining the presence of clouds only with
direct Sun measurements. For this particular period, even if the AOD is low,
the objective method shows the presence of thin clouds in front of the Sun.
It has to be noted that final AOD data include all available
measurements that have passed the quality control procedures, except the
cloud-flagging ones. So, all reported AODs are available, accompanied by a
flag showing which cloud-flagging
criteria have been assigned to the particular 1 min measurement.
Final AOD data
During the calibration and quality control procedure, three levels of data
are defined.
Level 1: These are the raw signal data as measured by the PFR instrument at
the four different channels.
Level 2: These are AOD values. The data are produced at each measuring
station using standardized software including QC tests, cloud screening,
Sun-tracking details and signal-to-AOD conversion using an existing
calibration file. Each of the mentioned test results is characterized by a
specific flag. In addition, the true solar elevation is calculated and
included. None of the level 1 data are discarded.
Level 3: AOD data are re-evaluated at WORCC which includes AOD results and
Ångström coefficients. Additional checks are included, such as the
detection of wavelength crossing
AOD(λ1) > AOD(λ2), where
λ1 > λ2. In addition, a day-to-day visual
inspection is performed in order to identify other technical issues or the
possible presence of undetected clouds. For the latter, additional cloud
flags are included in the final data files. Data control of level 3 data
includes overviews of the instrument's tracking performance, wavelength
crossing and ancillary data.
Hourly data records are prepared from quality-assured level 3 data, which are
then submitted to the World Data Center for Aerosols (WDCA) hosted by the
Norwegian Norsk Institutt for Luftforskning (NILU; ebas.nilu.no). Final data
files include the mean, median, standard deviation and the number of 1 min
samples used to calculate the hourly value at all four wavelengths.
In order to calculate hourly, daily and monthly statistics, we apply the
following criteria:
A minimum of 50 cloudless 1 min measurements per day are required to
calculate daily statistics. In this case, we eliminate days with less than
1 h of sunshine.
A minimum of six 1 min cloud free measurements are required to calculate
the hourly mean.
A minimum of 30 hourly values and 10 days per month are required to
calculate the monthly mean.
Measurements that lie beyond 2 standard deviations for an hourly mean are
considered outliers, as they are considered to be affected by cloud
contamination.
Monthly statistics can be presented with different approaches. In most
studies, AOD is usually reported as the arithmetic mean and associated
standard deviations over a selected period. This is based on the hypothesis
of an underlying normal distribution. However, AOD is often better
characterized by a lognormal distribution and described by geometric mean and
standard deviation. Based on a statistical analysis of skewness and kurtosis
in a multiyear and multistation AOD data set, O'Neill and co-authors
(O'Neill et al., 2000) have shown that a lognormal distribution
systematically provides a more robust base for reporting AOD statistics than
the normal distribution. Using long-term series of the final-selected 1 min
AOD data, users can then try to draw conclusions on the AOD climatology of
each station, the aerosol changes, if any, or the daily monthly and annual
patterns. As an example of the three Langley calibration-related stations
IZO, DAV and MLO, monthly means calculated from 15 years of measurements are
presented in Fig. 11. IZO and MLO are the Langley calibration sites and DAV
the triad host site.
For the particular sites that are all considered to have low AOD, we can
clearly see that Davos shows an increase in AOD during the summer months,
while the other two sites show much lower AOD with the exception of Saharan
dust intrusions at Izaña for the July–September months.
When comparing MLO and IZO statistics, we calculate long-term AOD550
means of 0.050 and 0.020, respectively, and geometrical means of 0.033 and
0.017. Respective geometrical standard deviations are about 1.7 for MLO and
2.6 for IZO, meaning that AOD varies from 60 to 170 % of 0.017 for MLO and
from 38 to 260 % of 0.033 for IZO. This is linked with the Saharan dust
events (AOD outliers for a normal distribution) that affect IZO. An overview
of the GAW-PFR AOD time series at all stations will be reported in a future
study.
Summary and conclusions
“AOD is the single most comprehensive variable to assess the total aerosol
load of the atmosphere and represents the least common denominator by which
ground-based remote sensing, satellite retrievals and global modeling of
aerosol properties are compared” (WMO, 2016a). According to the WMO,
multiwavelength AOD is one of the essential variables that critically
contributes to the characterization of Earth's climate. In addition, the
Global Climate Observing System (GCOS) includes atmospheric aerosols
including AOD as an essential climate variable. Finally, the European Space
Agency has included aerosols and AOD as one of the 10 climate change
initiative (CCI) variables to be investigated with a view towards building
space-based databases.
In order to monitor AOD over the long-term and provide data of traceable
quality, the World Optical depth Research and Calibration Centre (WORCC),
Davos, was established by the WMO Global Atmosphere Watch (GAW) program.
Fifteen existing GAW baseline stations were chosen for the deployment of PFRs
(precision filter radiometer; in-house manufacture). Quality-controlled and
-assured AOD data from this GAW-PFR network (www.pmodwrc.ch/worcc) are
being submitted by WORCC to the World Data Centre for Aerosols
(ebas.nilu.no).
Under conditions of low aerosol loading, e.g., AOD < 0.1 at 500 nm, a
calibration error of 1 % results in an error of ∼ 12 % in the
mean daily AOD. WMO has recommended (WMO, 1993) an absolute limit to the
estimated uncertainty of 0.02 optical depths for acceptable data and
< 0.01 as a goal to be achieved in the near future. These
specifications require a calibration uncertainty better than 2 % to be
achieved for spectral radiometers. In addition, measurement quality control
and quality assurance in different processing levels of the actual measured
direct Sun signals or retrieved AOD have to be included.
The calibration hierarchy of any network of Sun photometers is linked with
the instrument performance and stability over time. Instruments which do not
exhibit good stability (e.g., filter degradation) over time tend to utilize
short periods for Langley calibrations where the instrument response can be
considered constant. This can impact the calibration constant uncertainty
through the limited number of measurements and the statistical analysis that
is used. The PFR development and construction has been based on the use of
specific hardware and manufacturing techniques that make them reliable for
long-term measurements without rapid interference filter changes (e.g., Fig. 2). This provides the opportunity of using longer periods for collecting
Langley calibration results and thus results in better statistics for the
determination of the calibration constants.
Quality control of routine WORCC/PFR measurements includes a number of
measurement-related checks, including the optical window cleanliness and
the accuracy of the Sun pointing. In addition, a number of parameters such
as pressure, ozone and NO2 concentrations have to be
measured, assumed and/or modeled. Further QC procedures involve data
evaluation,
especially rejecting measurements with wavelength-related drifts (crossing)
and suspected cloud contamination in the line of sight. Cloud screening
becomes a difficult task, especially in the case of optically thin clouds
that cannot be easily distinguished from AOD associated with coarse-mode
aerosols. Finally, quality assurance of AOD data mainly include the
determination of a proper calibration (extraterrestrial signals) within the
required uncertainty.
WORCC has defined a protocol for calibrating the PFR instruments by
maintaining a triad of reference PFRs that exhibit differences well within
(more than 99 % of 1 min data over a 12-year period) the U95 WMO
criterion. The procedure includes systematic checks including comparisons
with instruments that perform measurements (Langley calibrations) at high-altitude stations.
One of the aims of WORCC is the provision of instrumentation and protocols
for uniform global measurement and records of AOD and the maintenance of the
radiometric reference for such measurements. So in addition to the hosting
and maintenance of the AOD triad, WORCC hosts the filter radiometer
comparison every 5 years (e.g., WMO, 2016b) and maintains long-term AOD
measurements at the main calibration sites of other aerosol networks such as
AERONET (Mauna Loa, USA; Izaña, Spain) and SKYNET (Chiba, Japan;
Valencia, Spain).
The data that has been used in this work is available upon request.
The authors declare that they have no conflict of
interest.
Acknowledgements
The authors would like acknowledge the Mauna Loa and Izaña Observatory staff for the inspection
and in situ problem solving of the PFR instrumentation presented in this work. They would also like to thank the Scientific
Advisory Group for Aerosols of the Global Atmospheric Watch, World Meteorological Organization Program, for the scientific support for WORCC.
Edited by: Mehrez Zribi
Reviewed by: Lionel Doppler and one anonymous referee
References
Angström, A.: On the atmospheric transmission of sun radiation and on dust in
the air, Geografiska Annaler, 11, 156–166,
1929.Barreto, A., Cuevas, E., Pallé, P., Romero, P. M., Guirado, C., Wehrli,
C. J., and Almansa, F.: Recovering long-term aerosol optical depth series
(1976–2012) from an astronomical potassium-based resonance scattering
spectrometer, Atmos. Meas. Tech., 7, 4103–4116, 10.5194/amt-7-4103-2014,
2014.
Bokoye, A. I., Royer, A., O'neill, N. T., Cliche, P., Fedosejevs, G.,
Teillet, P. M., and McArthur, L. J. B.: Characterization of atmospheric
aerosols across Canada from a ground-based sunphotometer network: Aerocan,
Atmos. Ocean., 39, 429–456, 2001.
Campanelli, M., Nakajima, T., and Olivieri, B.: Determination of the solar
calibration constant a sun-sky radiometer: Proposal of an in situ procedure,
Appl. Opt., 43, 651–659, 2004.
Che, H., Wang, Y., and Sun, J.: Aerosol optical properties at Mt. Waliguan
Observatory, Chinga, Atmos. Environ., 45, 6004–6009, 2011.Che, H., Zhang, X.-Y., Xia, X., Goloub, P., Holben, B., Zhao, H., Wang, Y.,
Zhang, X.-C., Wang, H., Blarel, L., Damiri, B., Zhang, R., Deng, X., Ma, Y.,
Wang, T., Geng, F., Qi, B., Zhu, J., Yu, J., Chen, Q., and Shi, G.:
Ground-based aerosol climatology of China: aerosol optical depths from the
China Aerosol Remote Sensing Network (CARSNET) 2002–2013, Atmos. Chem.
Phys., 15, 7619–7652, 10.5194/acp-15-7619-2015, 2015.
Dutton, E. G., Reddy, P., Ryan, S., and DeLuisi J. J.: Features and effects
of aerosol optical depth observed ant Mauna Loa, Hawaii, 1982–1992, J.
Geophys. Res., 99, 8295–8306, 1994.
Flowers, E. C., McCormick, R. A., and Kurfis, K. R.: Atmospheric Turbidity
over United States, 1961–1966, J. Appl. Meteorol., 8, 955–962, 1969.Goloub, P., Li, Z., Dubovik, O., Blarel, L., Podvin, T., Jankowiak, I.,
Lecoq, R., Deroo, C., Chatenet, B., Morel, J. P., Cuevas, E., and Ramos, R.:
PHOTONS/AERONET sunphotometer network overview: description, activities,
results, Proc. SPIE, 6936, 69360V, 10.1117/12.783171, 2007.
Harrison L., Michalsky, J., and Berndt, J.: Automated multifilter rotating
shadow-band radiometer: an instrument for optical depth and radiation
measurements, Appl. Opt., 33, 5118–5125, 1994.Herber, A., Thomason, L. W., Gernandt, H., Leiterer, U., Nagel, D., Schulz,
K. H., Kaptur, J., Albrecht, T., and Notholt, J.: Continuous day and night
aerosol optical depth observations in the Arctic between 1991 and 1999, J.
Geophys. Res., 107, 4097,
10.1029/2001JD000536, 2002.
Holben, B. N., Eck, T. F., Slutsker, I., Tanre, D., Buis, J. P., Setzer, A.,
Vermote, E., Reagan, J. A., Kaufman, Y. J., Nakajima, T., Lavenu, F.,
Jankowiak, I., and Smirnov, S.: AERONET – A Federated Instrument Network and
Data Archive for Aerosol Characterization, Remote Sens. Environ., 66, 1–16,
1998.
Holben, B. N., Tanré, D., Smirnov, A., Eck, T. F., Slutsker, I.,
Abuhassan, N., Newcomb, W. W., Schafer, J. S., Chatenet, B., Lavenu, F.,
Kaufman, Y. J., Vande Castle, J., Setzer, A., Markham, B., Clark, D., Frouin,
R., Halthore, R., Karneli, A., O'Neill, N. T., Pietras, C., Pinker, R. T.,
Voss, K., and Zibordi, G.: An emerging ground-based aerosol climatology:
Aerosol optical depth from AERONET, J. Geophys. Res., 106, 12067–12097,
2001.
IPCC: climate change 2013: the physical science basis, edited by: Stocker, T.
F., Qin, D., Plattner, G. K., Tignor, M., Allen, S. K., Boschung, J., Nauels,
A., Xia, Y., Bex, B., and Midgley, B. M.,
Contribution of working
group I to the fifth assessment report of the intergovernmental panel on
climate change, 2013.
Junge, C.: Gesetzmässigkeiten in der Grössenverteilung
atmosphärischer Aerosole über dem Kontinent, Berichte des Deutschen
Wetterdienstes in der US-Zone, Nr. 35, 261–275, 1952.
Linke, F.: Die Sonnenstrahlung und ihre Schwächung in der Atmosphäre,
Handbuch der Geophysik, 8, 239–338, 1942.McArthur, L. J. B., Halliwell, D. H., Niebergall, O. J., O'Neill, N. T.,
Slusser, J. R., and Wehrli, C.: Field comparison of network photometers, J.
Geophys. Res., 108, 4596, 10.1029/2002JD002964, 2003.McPeters, R. D., Frith, S., and Labow, G. J.: OMI total column ozone:
extending the long-term data record, Atmos. Meas. Tech., 8, 4845–4850,
10.5194/amt-8-4845-2015, 2015.Mitchell, R. M., Forgan, B. W., and Campbell, S. K.: The Climatology of
Australian Aerosol, Atmos. Chem. Phys., 17, 5131–5154,
10.5194/acp-17-5131-2017, 2017.
Michalsky, J. J., Schlemmer, J. A., Berkheiser, W. E., Berndt, J. L.,
Harrison, L. C., Laulainen, N. S., Larson, N. R., and Barnard, J. C.:
Multiyear measurements of aerosol optical depth in the Atmospheric Radiation
Measurement and Quantitative Links programs, J. Geophys. Res.-Atmos., 106,
12099–12107, 2001Mulcahy, J. P., O'Dowd, C. D., and Jennings, S. G.: Aerosol optical depth in
clean marine and continental northeast Atlantic air, J. Geophys. Res., 114,
D20204, 10.1029/2009JD011992, 2009.Nyeki, S., Halios, C. H., Baum, W., Eleftheriadis, K., Flentje, H.,
Gröbner, J., Vuilleumier, L., and Wehrli, C.: Ground-based aerosol
optical depth trends at three high-altitude sites in Switzerland and southern
Germany from 1995 to 2010, J. Geophys. Res., 117, D18202,
10.1029/2012JD017493, 2012.Nyeki, S., Wehrli, C., Gröbner, J., Kouremeti, N., Wacker, S.,
Labuschagne, C., Mbatha, N., and Brunke, E.-G.: The GAW-PFR aerosol optical
depth network: The 2008–2013 time series at Cape Point Station, South
Africa, J. Geophys. Res.-Atmos., 120, 5070–5084, 10.1002/2014JD022954,
2015.
O'Neill, N. T., Ignatov, A., Holben, B. N., and Eck, T. F.: The lognormal
distribution as a reference for reporting aerosol optical depth statistics,
Geophys. Res. Lett., 27, 3333–3336, 2000.Ruckstuhl, C., Philipona, R., Behrens, K., Collaud Coen, M., Dürr, B.,
Heimo, A., Mätzler, C., Nyeki, S., Ohmura, A., Vuilleumier, L., Weller,
M., Wehrli, C., and Zelenka, A.: Aerosol and cloud effects on solar
brightening and the recent rapid warming, Geophys. Res. Lett., 35, L12708,
10.1029/2008GL034228, 2008.
Smirnov, A., Holben, B. N., Eck, T. F., Dubovik, O., and Slutsker, I.:
Cloud-screening and quality control algorithms for the AERONET database,
Remote Sens. Environ., 30, 337–349, 2000.
Takamura, T. and Nakajima, T.: Overview of SKYNET and its activities, Opt.
Pura Apl., 37, 3303–3308, 2004.
Toledano, C., Cachorro, V. E., Berjon, A., de Frutos, A. M., Fuertes,
D., Gonzalez, R., Torres, B., Rodrigo, R., Bennouna, Y., Martin, L., and Guirado, C.: RIMA-AERONET network: long-term
monitoring of aerosol properties, Opt. Pura Apl., 44, 629–633,
2011 .Tomasi, C., Kokhanovsky, A. A., Lupi, A., Ritter, C., Smirnov, A., O'Neill,
N. T., Stone, R. S., Holben, B. N., Nyeki, S., Wehrli, C., Stohl, A.,
Mazzola, M., Lanconelli, C., Vitale, V., Stebel, K., Aaltonen, V., de Leeuw,
G., Rodriguez, E., Herber, A. B., Radionov, Zielinski, T., Petelski, T.,
Sakerin, S. M., Kabanov, D. M., Xue, Y., Mei, L., Istomina, L., Wagener, R.,
McArthur, B., Sobolewski, P. S., Kivi, R., Courcoux, Y., Larouche, P.,
Broccardo, S., and Piketh, S. J.: Aerosol remote sensing in polar regions,
Earth Sci. Rev., 140, 108–157, 10.1016/j.earscirev.2014.11.001, 2015.
Volz, F.: Photometer mit Selen-Photoelement zur spektralen Messung der
Sonnenstrahlung und zur Bestimmung der Wellenlängenabhängigkeit der
Dunsttrübung, Arch. Met. Geoph. Biokl. B, 10, 100–131, 1959.
Volz, F.: Some results of turbidity networks, Tellus XXI, 5, 625–630, 1969.
Wehrli, C.: Calibrations of filter radiometers for determination of
atmospheric optical depths, Metrologia, 37, 419–422, 2000.Wehrli, C.: Remote sensing of Aerosol Optical Depth in a Global surface
network, available at:
https://www.research-collection.ethz.ch/handle/20.500.11850/150574, 2008.Weller, M. and Gericke, K.: Long-term observations of aerosol optical depths
at the Meteorological Observatory Lindenberg, Meteorol. Z., 14, 651–662,
10.1127/0941-2948/2005/0070, 2005.
WMO/GAW: report No. 101, Report of the WMO workshop on the measurement of
atmospheric optical depth and turbidity, (WMO TD No. 659), Chapter 4: Working
Group Discussions – Sunphotometry, 4–5, 1993.
WMO/GAW: report No. 143, Global Atmosphere Watch measurements guide, WMO/TD-
No. 1073, Chapter 3: Aerosol and Optical Depth, 33–49, 2001.
WMO/GAW: report No. 162, Experts Workshop on a Global Surface-based Network
for Long Term Observations of Column Aerosol Optical Properties (WMO TD No.
1287), Chapter: GAWPFR: A Network of Aerosol Optical Depth Observations with
Precision Filter Radiometers (from Christoph Wehrli, 36–39), 153 pp., 2005.
WMO/GAW: report No. 227, WMO/GAW Aerosol Measurement Procedures, Guidelines
and Recommendations, 2nd Edn., WMO- No. 1177, ISBN 978-92-63-11177-7,
(WMO/TD- No. 1177), ISBN 978-92-63-11177-7, Chapter 7: Aerosol Optical Depth
60–67, 2016a.
WMO/GAW: Report No. 231, The Fourth WMO Filter Radiometer Comparison
(FRC-IV), 2016b.