Compared to the other continents and lands, Antarctica suffers from a severe
shortage of in situ observations
of precipitation. APRES3 (Antarctic Precipitation, Remote Sensing from
Surface and Space) is a program dedicated to improving the observation of
Antarctic precipitation, both from the surface and from space, to assess
climatologies and evaluate and ameliorate meteorological and climate models.
A field measurement campaign was deployed at Dumont d'Urville station at the
coast of Adélie Land in Antarctica, with an intensive observation period
from November 2015 to February 2016 using X-band and K-band radars, a snow
gauge, snowflake cameras and a disdrometer, followed by continuous radar
monitoring through 2016 and beyond. Among other results, the observations
show that a significant fraction of precipitation sublimates in a dry surface
katabatic layer before it reaches and accumulates at the surface, a result
derived from profiling radar measurements. While the bulk of the data
analyses and scientific results are published in specialized journals, this
paper provides a compact description of the dataset now archived in the
PANGAEA data repository (
The Antarctic ice sheet is a huge continental storage of water which, if altered through climate change, has the potential to significantly affect global sea level. While climate models consistently predict an increase in precipitation in the future in Antarctica (e.g. Palerme et al., 2016), most of which falls in the form of snow that will not melt and thus will accumulate further ice, observational data to raise confidence in the current precipitation in the models are still in demand. Antarctica is the poor cousin of global continental precipitation observation and climatology building efforts: citing Schneider et al. (2014) of the Global Precipitation Climatology Center (GPCC), “The GPCC refrains from providing a (precipitation) analysis over Antarctica” because of poor data coverage. The GPCC's global maps of continental precipitation from in situ observations are left blank only over Antarctica. Satellites offer rising prospects to monitor remote, difficult and/or uninhabited regions, but even then Antarctica tends to be excluded from comprehensive and/or global studies (e.g. Funk et al., 2015). Only those studies that specifically focus on the polar regions and Antarctica have presented and discussed aspects of the Antarctic precipitation by satellite (Palerme et al., 2014, 2016, 2017; Behrangi et al., 2016). However, ground-based observations are still lacking to suitably calibrate and validate the satellite products.
The measurement of solid precipitation is notoriously difficult (Goodison et
al., 1998; Nilu, 2013). Difficulties are exacerbated in Antarctica because
access and operations are logistically difficult and environmental conditions
are extreme. Antarctica is the driest continent on Earth in terms of
precipitation: satellite data estimate the mean precipitation at
171 mm yr
However, while radars are customarily used in other regions to monitor liquid precipitation (e.g. Krajewski and Smith, 2002; Fabry, 2015), and many campaigns have also been conducted in high-latitude and high-altitude regions to study snowfall (e.g. Schneebeli et al., 2013; Grazioli et al., 2015; Medina and Houze, 2015; Moisseev et al., 2015; Kneifel et al., 2015), experience is still limited in the Antarctic environment (Gorodetskaya et al., 2015). Because such instruments do not collect and directly measure the mass of falling precipitation, but rather measure the fraction of an emitted radiation which is reflected back by the hydrometeors, quantification in terms of precipitation involves both physically based (electromagnetic laws of diffusion, diffraction and propagation) and hypothesis-based (particle population size and shape, habits) post-processing. The hypothesis-based part requires calibration and validation using various sources of in situ measurements (e.g. Souverijns et al., 2017).
As part of the APRES3 project (Antarctic Precipitation, Remote Sensing from
Surface and Space,
Grazioli et al. (2017a) provide ample information on the observation site, most instruments and methods. A summary and complementary information are provided below.
The main APRES3 (austral) summer field campaign took place at French
Antarctic scientific station Dumont d'Urville (DDU) in Adélie Land
(66.6628
Standard measurements of atmospheric variables (temperature, wind speed, wind direction, relative and specific humidity, atmospheric pressure) are collected regularly all year long by the French meteorological service (Météo France), and a radiosounding is made daily at 00:00 UTC. The routine program does not involve any instrumental measurement of precipitation. There are reports of visual estimation of the occurrence and type in the METAR (METeorological Airport Report) convention, but no quantification. For the APRES3 campaign, several instruments were deployed from the beginning of November 2015 to the end of January 2016 to objectively characterize and quantify the occurrences and amounts of precipitation, as described below.
As reported in the introduction, traditional collecting precipitation gauges are unreliable in Antarctica in general, and in particular in the coastal regions strongly affected by katabatic winds. Radars are the core instruments of the APRES3 campaign. Radars remotely sense the hydrometeors, estimate quantities and speed, and from this derive precipitation rates. Radars can scan and profile through atmospheric and hydrometeor layers and look beyond blowing snow near the surface. Two radars were deployed: a K-band frequency-modulated continuous-wave vertically staring profiler and an X-band pulsed dual-polarization scanning Doppler radar. The first instrument, a Metek micro-rain radar (MRR), is designed to measure rainfall rather than snowfall using the backscattering and vertical velocity information. However, the raw Doppler spectra can be reprocessed using Maahn and Kollias (2012)'s improved and innovative processing chain for data collected in snow to retrieve Doppler radar moments such as reflectivity Z and Doppler velocity. Most Z–S relations for radars have been derived for 10, 35 or 94 GHz and therefore the measured equivalent radar reflectivity at 24 GHz is first converted to X-band. Once mapped to X-band reflectivity this can be converted to snowfall rate S by means of a Z–S power law fitted to the local conditions using the weighing gauge information or parameterizations from the existing literature (for more details, see Grazioli et al., 2017a). The MRR was used with a 100 m vertical resolution. The second instrument, a mobile X-band polarimetric radar (MXPol), for which extensive experience with the measurement of snow is available (Schneebeli et al., 2013; Scipión et al., 2013; Grazioli et al., 2015), provided more detailed information and served as a control and reference for the calibration of the method to use the MRR data. It was used with a 75 m radial resolution, maximum radial distance 30 km, and different types of scans within a repeating scanning sequence of 5 min (plan position indicator (PPI), range height indicator (RHI), vertical profiles). While the X-band radar could only be deployed during the summer campaign and had to be shipped back after completion in February 2016, the K-band radar sheltered by a radome could remain on site after the summer campaign. The radome significantly attenuates the signal (6.14 dB, Fig. 4 and Eq. 1 of Grazioli et al., 2017a), but it is necessary to protect the radar against the fierce winter winds in Adélie Land.
Summary of data from the APRES3 observation campaigns available by download from the PANGAEA repository (Berne et al., 2017) or by request to the authors. MASC data are provided for each picture taken, the taking of which varies with the occurrence of particle detection.
The Biral VPF 730 disdrometer is also a non-capture instrument, which
estimates the size and speed of airborne particles from the diffusion and
diffraction of an infrared light beam within a 400 cm
A MASC was deployed next to the disdrometer. This instrument collects
high-resolution stereoscopic photographs of snowflakes in free fall while
they cross the sampling area (Garrett et al., 2012), thus providing
information about snowfall microphysics and particle fall velocity. The MASC
uses three identical
What fraction of snowfall a traditional precipitation gauge captures is
unwarranted. On the other hand, unlike remote sensing instruments, the mass
quantification of any captured snow is direct and straightforward. An OTT
Pluvio2 precipitation gauge was deployed for the duration of the summer
campaign. Snow falling in the instrument is definitely captured and weighted.
The instrument used here was equipped with a manufacturer-design wind shield
meant to limit wind impacts on capture efficiency. Further, the instrument
was relatively shielded from the strongest wind due to its location, on the
roof of a container but on the side of a building. The MASC and disdrometer
were deployed at the same partially sheltered site (Fig. 1), the local
meteorology of which was sampled locally by a Vaisala WXT520 weather
transmitter, the principles, instrumental accuracy and performance of which
can be found in the manufacturer's User's Guide
(
Setting of the in situ sensors (weather station WXT520, disdrometer Biral, snow gauge Pluvio 2 and MASC) on the roof of a small shelter close to other buildings.
Table 1 summarizes the data streams from the APRES3 measurement campaign.
Grazioli et al. (2017a) extensively process and discuss the data from the
different instruments. Further analyses and presentation are beyond the scope
of this data paper, and only a few snapshots are provided to illustrate the
content of the database. Figure 2 shows the cumulative precipitation during
the intensive summer campaign, as yielded by the Pluvio2 snow gauge and the
processed MRR at the lowest useful level and at 741 m above sea level. Only
28 out of 31 MRR levels are provided in the database. This is because several
simplifications necessary for a tractable quantitative interpretation of
radar signal power do not apply in the two lowest levels. Data processing is
based on assumptions that are not valid as it may lead to overestimation of
reflectivity (Peters et al., 2005). In the uppermost level, the data become
noisy, as according to Kneifel et al. (2011), the detectability is highest
close to the ground, at
Cumulative precipitation during the APRES3 summer campaign, from the Pluvio2 and MRR instruments. Thin black vertical lines bracket the largest precipitation event in the period, from 12 to 17 December 2015. Precipitation from the MRR is reported for two levels above sea level, 341 and 741 m.
An example of the
PDF of snow particle riming from the MASC data over the observation period.
One year (November 2015–November 2016) of cumulative precipitation from MRR backscattering at 341 and 741 m above the surface.
The MRR precipitation at the lowest useful level (341 m a.s.l.) is
significantly less than that at 741 m a.s.l., showing that a significant
fraction of the precipitation formed above sublimates in the dry katabatic
air layer near the surface. Further observations show that this frequently
occurs in all seasons of the year (see below). Meteorological and climate
models suggest that at the full scale of the Antarctic ice sheet up to
17 % of the precipitation evaporates in a dry surface layer before
reaching the ground, and thus does not contribute to feeding the ice sheet
(Grazioli et al., 2017b). Altogether, the 2015–2016 summer was relatively
dry and few strong precipitation events occurred. One such event happened
from 12 to 17 December 2015 (delineated by thin vertical black lines in
Fig. 2), during which the largest part of the total cumulative precipitation
this summer was recorded. Figure 3 shows an example of the Biral disdrometer
size–speed matrix during this event. The local wind was relatively strong
(5.4 m s
Figure 4 shows the probability distribution function of the degree of riming
of the snowfall particles as obtained by processing the MASC photographs. No
less than 426 229 photographs of falling snow particles were collected
during the season. Each picture is processed as described in Praz et
al. (2017). The database offers the processed results in the form of a
classification, rather than the photographs themselves. Figure 4 cumulates
all single estimates of the degree of riming in the database. The degree of
riming is defined in this context as a continuous index between 0 (no riming
on the particle detected) and 1 (fully rimed, graupel-like particle). Almost
half of the particles are close to fully rimed, indicating that cloud liquid
water is very frequent in summer. Finally, Fig. 5 shows precipitation from
the MRR dataset over the full record in the database, for more than a year
from 21 November 2015 to 11 December 2016. Again, reports from two
elevations, 341 and 741 m a.s.l., are displayed. This shows that at DDU,
cumulated over a full year,
The APRES3 field campaign database is available in open
access on PANGAEA,
In conclusion, observations at DDU carried out as part of the APRS3 project
provide an unprecedented dataset of precipitation at the coast of Antarctica,
complementing existing documentation efforts (Gorodetskaya et al., 2015) in a
region which otherwise suffers from a severe shortage of such data. Our
analysis of the data yields new insights into the characteristics and
particularities of Antarctic snowfall, in particular that a large fraction of
the precipitation formed in the atmosphere sublimates before reaching the
surface. This information could only be obtained with instruments that can
profile through the atmospheric layers, like radars here. However, the
dataset goes beyond radar data and provides extensive complementary
characterization of snow particle geometry and cumulative quantities of
snowfall at the surface. Except for the dataset from the MXPol
dual-polarization scanning radar during the summer campaign, the size of
which (about 4 TB) is too large to be shared online but can be obtained by
direct request to the authors, all data are now distributed (Berne et al.,
2017) and can be freely accessed from the PANGAEA repository
(
CG, AB and JG organized and ran the field campaigns. JG, AB, CDA and BB developed radar data processing methods and produced data series. CG processed Biral disdrometer data and CP processed MASC data. All authors contributed data analysis, discussion and conclusions.
The authors declare that they have no conflict of interest.
We thank the French Polar Institute, which logistically supports the APRES3 measurement campaigns. We particularly acknowledge the support of the French National Research Agency (ANR) to the APRES3 project. The Expecting Earth-Care, Learning from the ATrain (EECLAT) project funded by the Centre National d'Etudes Spatiales (CNES) also contributed support to this work. The Swiss National Science foundation (SNF) is acknowledged for grant 200021_163287, financing the Swiss participation in the project. PANGAEA is gratefully acknowledged for hosting and distributing the APRES3 data. Steve Colwell, of the British Antarctic Survey, and two anonymous reviewers provided thoughtful comments and suggestions that helped correct and clarify a number of issues in the preliminary manuscript. Edited by: David Carlson Reviewed by: Steve Colwell and two anonymous referees