The HD(CP)
HOPE-Jülich instrumentation included a radio sounding station, 4 Doppler lidars, 4 Raman lidars (3 of them provide temperature, 3 of them water vapour, and all of them particle backscatter data), 1 water vapour differential absorption lidar, 3 cloud radars, 5 microwave radiometers, 3 rain radars, 6 sky imagers, 99 pyranometers, and 5 sun photometers operated at different sites, some of them in synergy. The HOPE-Melpitz campaign combined ground-based remote sensing of aerosols and clouds with helicopter- and balloon-based in situ observations in the atmospheric column and at the surface.
HOPE provided an unprecedented collection of atmospheric dynamical,
thermodynamical, and micro- and macrophysical properties of aerosols, clouds,
and precipitation with high spatial and temporal resolution within a cube of
approximately 10
First applications of HOPE data for model evaluation have shown a general agreement between observed and modelled boundary layer height, turbulence characteristics, and cloud coverage, but they also point to significant differences that deserve further investigations from both the observational and the modelling perspective.
Clouds and precipitation play a central role in the climate system and were repeatedly identified as the largest problem in a realistic modelling of atmospheric processes, forcings, and feedbacks (IPCC, 2013; Jakob, 2010). Uncertainties in the characterization of clouds and precipitation have manifold consequences on virtually all non-atmospheric climate components from ocean mixed-layer stability to vegetation variability, to net mass balance of ice sheets (Wilson and Jetz, 2016).
To achieve progress in the improvement of the representation of clouds and
precipitation in atmospheric models, the German research initiative High
Definition Clouds and Precipitation for advancing Climate Prediction,
HD(CP)
The HD(CP)
Within the M module (modelling) of HD(CP)
The O4 project in the O module of HD(CP)
HOPE complements the larger spatiotemporal full-domain (O2) and supersite
(O1) activities in the O module in HD(CP)
HOPE builds on the experience gained in previous field campaigns like the Convective and Orographically induced Precipitation Study (COPS) (Wulfmeyer et al., 2011), but with a stronger focus on multi-sensor synergy covering a micro- to mesoscale domain. COPS and the associated general observation period (GOP) that was prepared in the context of the Quantitative Precipitation Forecasting priority programme (SPP1167) of the German Science Foundation (DFG) (Crewell et al., 2008) aimed at the observation of orographically driven initiation of convection with supersites several tens of kilometres apart in strongly structured terrain. Complementary to COPS, HOPE covers a smaller domain with higher resolution and is accompanied by long-term supersite observations within the framework of the Terrestrial Environmental Observatories (TERENO) programme (Simmer et al., 2015) around the ground-based remote-sensing supersite Jülich Observatory for Cloud Evolution (JOYCE) (Löhnert et al., 2015), and the TROPOS long-term aerosol observatory in Melpitz (Spindler et al., 2012).
Although phase 1 of HD(CP)
This article mainly serves as a guide through the sites and instrumentation used during the HOPE campaigns and aims to motivate readers to learn about the details and specific conclusions described in the individual publications this overview is built upon. The structure is as follows. Section 2 describes the site set-ups and measurements performed during HOPE including information about the meteorological conditions and data availability. Examples from each of the research topics are presented in Sect. 3. In Sect. 4, first comparisons between models and observations are discussed. A summary and conclusions on the further applications of the HOPE data as well as designs for future observational strategies are presented in Sect. 5. Individual work performed during HOPE is published in this ACP/AMT HOPE special issue or, in part, in other journals and is cited in the present overview correspondingly.
The technological aspect of HOPE was to unite most of the mobile ground-based remote-sensing and surface flux observations available in Germany within a single domain in order to capture the vertical structure and horizontal variability of wind, temperature, humidity, and aerosol and cloud condensate with the best possible temporal and spatial resolution. Thus, we were able to accommodate active remote sensing from lidar and radar and passive remote sensing from microwave radiometer and sun photometer, whenever possible with scanning capabilities. During HOPE, 3-D water vapour, temperature, and wind measurements were possible with unprecedented spatiotemporal resolution in the boundary layer. In order to understand the forcing of and the response to surface properties, distributed surface flux and surface standard meteorological observations were deployed as well. Of course, it is not possible to obtain an instantaneous 3-D picture of the atmosphere from a limited number of directional observations. However, ongoing improvements in sensor detection accuracy and optimized scanning strategies will capture the 4-D boundary layer properties even better in the future.
The measurement activities during HOPE mainly consisted of a major field experiment in Jülich, Germany, denoted as HOPE-Jülich, conducted from 3 April to 30 May 2013 followed by a smaller campaign that was performed in Melpitz, denoted as HOPE-Melpitz, Germany, which was conducted from 9 to 29 September 2013. Figures 1 and 2 give an overview of the broad spectrum of instruments installed during the two campaigns and their overall set-up. A detailed introduction is given below.
Set-up of the HOPE-Jülich campaign showing the location of the three supersites Jülich (JUE), Hambach (HAM), and Krauthausen (KRA) as well as the outpost Wasserwerk (WAS) with their main instrumentation. The cones and arrows illustrate the field of view and scanning capabilities of the specific remote-sensing instruments.
In order to derive the atmospheric state of water vapour, temperature, wind,
and cloud and precipitation properties with 100 m resolution for an area of
about 10
Illustration of the set-up of the HOPE-Melpitz campaign showing the deployed main instrumentation. The cones illustrate the field of view of the specific remote-sensing instruments.
Sites and networks deployed during HOPE-Jülich. Information on the individual instruments are given in Table 2. For details on the affiliations see Sect. 2.1.1. as well as the title page of this article.
Map of the spatial distribution of the measurement sites and
networks deployed according to Table 1
As can be seen from Table 1, most instruments were deployed at the three supersites Jülich (JUE), Krauthausen (KRA), and Hambach (HAM) with its outpost close to a pump station “Wasserwerk” (WAS). At each supersite one or several main remote-sensing facilities were deployed. At JUE this was the instrumentation of the permanently installed JOYCE, at HAM the Karlsruhe Institute for Technology mobile facility KITcube and the lidar systems of the Institute for Physics and Meteorology (IPM) of the University of Hohenheim (UHOH) were deployed, and at KRA the Leipzig Aerosol and Cloud Remote Observations System (LACROS) was operated. In some publications that are based on HOPE-Jülich observations, the supersite names are also referring to the main facility deployed at each site, e.g. LAC for LACROS at the supersite KRA, JOY for JOYCE at the supersite JUE, and KIT for KITcube at the supersite HAM. The instrumentation that was present at each site is listed in Table 2. In total, the HOPE-Jülich set of instruments included a radio sounding station, 5 Doppler lidars, 4 Raman lidars, 1 differential absorption lidar (DIAL), 3 cloud radars, 5 microwave radiometers, 3 precipitation radars, 6 sky imagers, 99 pyranometers, and 5 sun photometers. Below, the operating institutions and available measurement devices at all three supersites are briefly outlined. Concerning technical details of the individual instruments, such as instrument calibration and stability, restrictions in the instrument resolution, or the assessment of uncertainties, we refer the reader to the literature cited in Table 2. In addition, results shown in Sect. 3 and 4 of this article are based on already published articles which are cited at the respective positions in text and contain detailed information on the applied instrumentation and methodologies.
Details of instruments deployed during HOPE-Jülich and (in part; see Sect. 2.1.2) during HOPE-Melpitz. For details on the sites, network, and affiliations see Sect. 2.1.1 and Table 1.
All measurements during HOPE-Jülich were built around the central
supersite Jülich where JOYCE (Löhnert et al., 2015) is operated
continuously at FZJ. JOYCE (
With the newly designed observing system KITcube (Kalthoff et al., 2013), the Institute of Meteorology and Climate Research (IMK) of the Karlsruhe Institute of Technology (KIT) provides meteorological and convection-related parameters and contributed to measurements of the development of clouds with high temporal and spatial resolution in the HOPE area. KITcube was the main facility at the supersite HAM and consists of a surface-based network with meteorological stations and a 30-metre tower measuring the standard parameters of temperature, humidity, air pressure, wind speed and direction, sensible heat fluxes, the energy balance components at the Earth's surface (Kalthoff et al., 2006), and soil moisture and soil temperature profiles (Krauss et al., 2010). These stations in general are distributed over the whole area of KITcube to account for surface inhomogeneity. For instance, KIT operated two eddy-covariance stations – one at the main site HAM, and a second one at the outpost WAS, approximately 2.5 km to the west. KITcube also includes scanning Doppler wind lidars to measure wind speed, wind direction, and turbulence characteristics in the CBL. One Lockheed WindTracer was installed at supersite HAM, with a second WindTracer at the outpost WAS (see Fig. 3b) to allow dual-Doppler applications. Both were installed together with a Leosphere Windcube. Additionally, a Doppler lidar of KIT IMK-IFU (Halo Photonics Streamline) was operated at the TERENO site Selhausen. These instruments were complemented by a microwave radiometer, a scanning 35 GHz cloud radar monitoring the development of clouds, a vertically pointing micro rain radar and disdrometers providing information about precipitation, and a ceilometer for cloud base height detection. At a second KITcube outpost denoted KiXPol, approximately 7.5 km southwest of HAM, a polarimetric X-band rain radar was operated, providing volume scans of polarimetric moments, vertical cross sections (RHI scans) on demand, as well as the horizontal precipitation field for the HOPE-Jülich area every 5 min and with 250 m radial resolution. In situ vertical profiles of temperature, humidity, and wind profiles as well as convective indices were gathered by radiosondes launched regularly every sixth full hour at the KITcube main site. Land and full-sky images were taken by S14 camera systems at HAM and WAS.
Also at supersite HAM, two lidar systems from IPM of UHOH observed 3-D thermodynamic fields of temperature and moisture including their turbulent fluctuations. A temperature rotational Raman lidar (TRRL) measured temperature profiles (Behrendt et al., 2015; Hammann et al., 2015; Radlach et al., 2008) and a water vapour DIAL measured absolute humidity profiles (Muppa et al., 2016; Späth et al., 2016; Wagner et al., 2013). In contrast to the Raman lidar technique, the DIAL technique, which is based on the alternating emission of laser pulses at frequencies strongly and weakly absorbed by water vapour, does not require calibration. By sending out the laser beam vertically into the atmosphere, high-resolution observations of the convective boundary layer and the lower free troposphere can be made with the instrument (Muppa et al., 2016; Wagner et al., 2013). But the same system also allows for observations in any direction of interest and thus to map the structure of the water vapour field and its development (Milovac et al., 2016). Like the DIAL, the TRRL of IPM also has scanning capabilities and an intrinsic high spatial and temporal resolution of 1–10 s and 15–100 m up to a range of about 5 km. Consequently, both systems are capable of resolving turbulent fluctuations in the convective boundary layer from the surface to the entrainment zone. Derived products include statistical moments of moisture and temperature turbulent fluctuations (Behrendt et al., 2015; Muppa et al., 2016; Wulfmeyer et al., 2015), profiles of stability variables such as buoyancy (Behrendt et al., 2011), and the boundary layer depth, aerosol backscatter fields, and cloud boundaries. The self-calibrating DIAL technique has excellent absolute accuracy (Bhawar et al., 2011) and has been acknowledged as water vapour reference standard of WMO.
Continuous observations with the TROPOS mobile facility LACROS (Bühl et al., 2013) were
performed at the supersite KRA. LACROS employs a 35 GHz cloud radar, a
multi-wavelength Raman polarization lidar, a ceilometer, a Doppler lidar, a
microwave radiometer, an optical disdrometer, and an all-sky imager.
The Raman polarization lidar Polly
Beside the supersite observations at JUE, KRA, and HAM, different instrument networks were also distributed in the vicinity of the three supersites. The pyranometer network (PYR) of 99 autonomous meteorological stations including pyranometers developed by TROPOS (Madhavan et al., 2016) was deployed within a radius of about 5 km around the supersite JUE to capture the broadband downwelling solar irradiance with high spatial and temporal resolution.
The Meteorological Institute of the University of Bonn (MIUB) coordinated the operation of six sky imagers within the SKY network that were provided by several partner institutes to obtain imagery for cloud classification and the determination of cloud morphology (Beekmans et al., 2016).
Three scanning polarimetric X-band rain radars jointly operated within the XRD network by the University of Bonn (BoXPol), the Jülich Research Centre (JuXPol) (Diederich et al., 2015), and KIT (KiXPol) provided 3-D fields of polarimetric moments over the domain and precipitation estimates (Trömel and Simmer, 2012; Xie et al., 2016).
Within the sun photometer network (SUN), the vertically integrated aerosol characteristics and water vapour field at the three HOPE-Jülich supersites as well as at two more-remote sites (Aachen and Insel Hombroich; see Table 1) were derived. Except for the one operated within JOYCE at supersite JUE, all sun photometers were provided by NASA Goddard Space Flight Center (GSFC), Langley, USA, and operated by MPIM.
Additionally, two ground-based scanning spectral radiometers, SpecMACS from the Munich Institute for Meteorology (MIM) of the Ludwig Maximilian University (LMU) of Munich (Ewald et al., 2016) and EAGLE from Leipzig Institute of Meteorology (LIM) of the University of Leipzig (Jäkel et al., 2013), participated in the campaign. These instruments provide the solar radiation reflected at cloud sides from which vertical profiles of cloud microphysical properties shall be inferred.
The HOPE-Melpitz campaign basically combined the remote sensing of aerosol and cloud properties of the LACROS supersite with the in situ observations of the helicopter-borne Airborne Cloud Turbulence Observation System (ACTOS) (Siebert et al., 2013) (see Fig. 2). The follow-up campaign HOPE-Melpitz became necessary because of problems with the availability of a helicopter carrying ACTOS during HOPE-Jülich.
The Melpitz site (51.525
Topography around the location of the HOPE-Melpitz campaign.
Similar to HOPE-Jülich, during HOPE-Melpitz the LACROS instrumentation
comprised the polarization Raman lidar Polly
Measurements of the broadband irradiances at the surface were carried out with a mobile station following the recommendations of the Baseline Surface Radiation Network (McArthur, 2005) and can serve as high-quality reference for the pyranometer network. In addition, spectral irradiances were observed with a rotating shadowband radiometer of type GUVis-3511 (Witthuhn et al., 2017).
Detailed information on the ACTOS set-up are given in Siebert et al. (2013).
ACTOS provides dynamic, thermodynamic, and cloud and aerosol
microphysical properties of warm shallow boundary layer clouds. The standard
ACTOS instrumentation comprises sensors for the wind vector, temperature, and
humidity under clear and cloudy conditions. Observed microphysical parameters
of liquid clouds include the cloud droplet number–size distribution in the
range from 1 to 180
The two ground-based spectral radiometers EAGLE and SpecMACS from LIM and LMU, respectively, which were operated during HOPE-Jülich, were also deployed during HOPE-Melpitz. Besides ACTOS, airborne observations with spectral radiometers for cloud remote sensing from the Freie Universität Berlin (Schröder et al., 2004) were performed on some days.
HOPE-Jülich was conducted from 3 April to 31 May 2013 as this period in the year favours low-level cloud formation. Only the measurements of the PYR continued until end of July to capture high-sun conditions. An extensive operation plan documenting the daily availability of all central instruments of HOPE-Jülich can be found in the Supplement to this article.
The weather conditions during the campaign varied from several warm and cold front passages interrupted by a few high-pressure systems with high-level cirrus clouds at the beginning of the campaign and more low-level convective clouds later on. Since the campaign focused on the onset of clouds and precipitation, intensive observation periods (IOPs) have been called out whenever clear skies, boundary layer clouds, or precipitation-developing clouds were forecast. During IOPs, instruments requiring continuous human control were measuring in addition to autonomously operating instruments. Furthermore, radiosondes were launched more frequently at supersite Hambach, depending on the weather situation and its variability. Table 3 summarizes the IOPs during HOPE-Jülich and the corresponding weather conditions. IOPs with especially well-suited weather conditions have been labelled as “golden days” and have been more deeply analysed by all participating groups.
Summary of intensive observation periods during HOPE-Jülich. Bold typeface denotes “golden days”.
As an example, a detailed depiction of IOP7 (25 April 2013) consisting of a turbulently driven boundary layer development topped with afternoon single cumulus clouds in the afternoon can be found in Löhnert et al. (2015). There, it is demonstrated that a holistic view of the daily development of the boundary layer is only possible through the synergetic treatment of different ground-based remote sensors.
Weather conditions have not been optimal for the helicopter operations due to problems with low-level overcast clouds (no flight permit inside clouds) and icing conditions. During the 3 weeks of the campaign, five IOPs have been performed on which 10 ACTOS flights were performed, covering 15 h of measurements (Table 4). However, the helicopter flights captured a spectrum of different meteorological conditions as can be seen from Table 4.
Summary of intensive observation periods during HOPE-Melpitz. On these days a total of 15 h of observations with ACTOS were performed. Cu: cumulus; Sc: stratocumulus. Bold typeface denotes “golden days”.
All officially participating partners have been submitting their
quality-controlled data in a common format to the HD(CP)
One central goal of HOPE was the characterization of the turbulent structure of the ABL. To capture this feature, both the surface energy budget components and the wind fields near the surface and in the lower boundary layer are required. The set of instruments available during HOPE-Jülich provided a unique opportunity to compare and to correlate vertical-velocity variances from different locations. Maurer et al. (2016) made use of a triangular set-up of three KITcube Doppler lidar systems deployed approximately 3 km apart from each other. This distance was assumed to be sufficient to ensure that the lidars do not monitor the same convective cells at the same time. Nevertheless, they found persistent similar statistical properties of velocity variances measured along the wind direction in contrast to measurements across the wind direction. This indicates that local organized structures of turbulence can dominate turbulence characteristics and that single turbulence measurements may not be representative for a larger domain.
In a similar approach Träumner et al. (2015) investigated correlation patterns of near-surface wind fields from a dual-Doppler lidar set-up scanning at low elevation angles together with available in situ wind vectors from ground-based stations. As a measure for anisotropy, integral length scales were defined for the along-stream and the cross-stream wind components. Integral scales provide a measure of the spatial or temporal dimension of turbulent eddies (Wyngaard, 2004). The authors confirmed previous findings of streak-like structures elongated and aligned in the wind direction. Also periodic behaviour in the horizontal wind fields has been identified occasionally. Interestingly, the mean structural pattern could be related to the background wind speed and the atmospheric stability. Still, individual wind fields can vary strongly for the same external forcing. Thus, a characterization of coherence patterns in the otherwise turbulent boundary layer requires extensive spatiotemporal averaging.
Eder et al. (2015) investigated the complete surface energy budget and tested the hypothesis of whether so-called turbulent organized structures (TOS), low-frequency structures that fill the entire ABL, are a major cause for the frequently unclosed surface energy balances as they contribute to the vertical energy fluxes. In fact, by means of data from horizontally and vertically scanning Doppler lidars the authors could show that TOS with timescales larger than 30 min extend deep into the surface layer. This finding implies that future turbulent energy exchange studies require the full 3-D field of humidity, temperature, and velocity in high spatiotemporal resolution, which was also pointed out and elaborated in Wulfmeyer et al. (2016).
Based on the autonomous pyranometer network described in Madhavan et
al. (2016), the representativeness of a single station measurement for
spatially extended domains with different area sizes has been investigated
(Madhavan et al., 2017). This is an important aspect for the evaluation of
model results with observations, where point measurements are mostly compared
to grid-box means and are thus implicitly assumed to have similar
statistical properties. Spatial and temporal smoothing has been quantified,
which limits the representativeness of a point measurement for its surrounding
domain size and period. Spatial averaging acts as a low-pass filter and
reduces or even completely removes high-frequency spatiotemporal variations.
This is illustrated in Fig. 5a, which shows a wavelet-based power
spectrum obtained from 99 pyranometer stations and corresponding estimates
of the power spectra for three areas ranging from 1
Spatiotemporal characteristics derived from the pyranometer network
under broken-cloud conditions during HOPE-Jülich. This figure illustrates
the origin of deviations between a point measurement (labelled as var(
Also based on the horizontally high-resolved measurements of the irradiance from the PYR performed by TROPOS, Lohmann et al. (2016) analysed the statistics of spatiotemporal irradiance fluctuations with a strong application-oriented focus on photovoltaic power systems. They specifically calculated single-point statistics and two-point correlation coefficients for clear, overcast, and mixed skies. The statistics for clear and overcast skies show similar behaviour as in previously published work; see Lohmann et al. (2016) for references. In order to account for conditions for a distributed PV system, they defined so-called irradiance increments as changes in transmissivities over specified intervals of time and showed that the magnitude of increments is more strongly reduced by spatial averaging than that of the fluctuations. By conditioning the sky type – which can easily be done from the irradiance measurements themselves – they demonstrated that the probability for strong irradiance increments is twice as high compared to increment statistics computed without distinguishing between different sky types.
As clouds impose the largest short-term variability in solar irradiance at the surface, the analysis of cloud advection and subsequent extrapolation represents a reasonable approach for short-term irradiance forecasts. Schmidt et al. (2016) made use of time series of hemispheric sky images to predict the surface irradiance by means of mapping the cloud position, which in turn is translated into shadow maps at the surface. The temporal evolution of such shadow maps is calculated from cloud motion vectors that were calculated from subsequent sky images. Irradiance forecasts of up to 25 min have been produced and were validated against the network of pyranometers described in Madhavan et al. (2016). Although these sky-imager-based forecasts do not outperform a simple persistence forecast on average, improved forecast skill was found for convective cloud conditions with high cloud and irradiance variability. This finding may provide useful application in photovoltaic electricity production.
The goal of the HD(CP)
The increased model resolution puts new requirements on evaluation
techniques. The HOPE campaigns provided an optimum test bed for novel
applications to derive boundary layer fluxes and turbulence characteristics.
Observations of the turbulent fluxes of thermodynamic properties in the planetary boundary layer (PBL),
such as of temperature and water vapour, provide detailed information on the
minimum resolution required by a model to capture the turbulence spectrum
down to the inertial sub-range and consequently to resolve the major part of
the turbulent fluctuations. This value is here introduced as the integral
scale. During HOPE-Jülich, based on TRRL observations it was possible to derive the statistics of turbulent temperature fluctuations and thus of
the integral scale of this parameter in the PBL (Behrendt et al., 2015). In
addition to commercially available Doppler lidar systems, which provide
turbulent wind fluctuations, three water vapour research lidars were deployed
during HOPE-Jülich, which provide turbulent humidity fluctuations that
were documented by Di Girolamo et al. (2017) and Muppa et
al. (2016). As the authors of the above-mentioned studies note,
HOPE-Jülich provided for the first time data to observe the turbulence
characteristics of the PBL, more specifically the CBL, up to the fourth statistical moment, i.e. the mean, standard
deviation, variance, skewness, and kurtosis of the spatiotemporal water
vapour and temperature. Examples of the relationship between the integral
scales (introduced in Sect. 3.1) of humidity and temperature fluctuation
and height above ground within the CBL for the 20 April 2013 (IOP 5),
11:30–13:30 UTC (only temperature fluctuations; see Di Girolamo et
al., 2017), and 24 April 2013 (IOP 6), 11:00–12:00 UTC (temperature and
humidity fluctuation; see Behrendt et al., 2015 and Muppa et al., 2016),
respectively, are depicted in Fig. 6. A decrease in the integral length
scale of the water vapour mixing ratio with height in the upper part of the
CBL was found at the HAM site similar to previous observations (Couvreux et
al., 2005; Wulfmeyer et al., 2010). A similar decrease was found for
temperature at the same site. The temperature observations from JUE site show
a more complex structure. The reasons for this are still under investigation.
The decrease of the integral length scale toward the top of the CBL can be
explained by the decrease in the size of the turbulent eddies with height
resulting from the entrainment of dry free-tropospheric air at the CBL top
(Couvreux et al., 2005), which is also characterized by an increase in the
variance of the temperature or water vapour toward CBL top. Converting the
observed timescales shown in Fig. 6 to spatial scales assuming horizontal
and vertical wind velocities of 5 and 1 m s
Integral scales of the temperature fluctuations (black) and humidity
fluctuations (red) in the convective boundary layer derived from
high-resolved observations obtained between 11:30 and 13:30 on 20
April 2013 (IOP 5) and 11:00 and 12:00 UTC on 24 April 2013 (IOP 6) during
HOPE-Jülich. Heights are normalized with respect to the height of the
convective boundary layer
Detailed CBL turbulence characteristics from HOPE and further field campaigns (Wulfmeyer et al., 2016) showed that the combination of active temperature, humidity and wind profiling applied during HOPE-Jülich sufficiently resolves the turbulence structure of the CBL and lays the groundwork for new boundary layer turbulence parameterizations.
In addition to turbulent fluxes in the cloud-free planetary boundary layer,
the turbulence characteristics of a stratocumulus layer were investigated
simultaneously with ACTOS and the Doppler WiLi of the LACROS site on 22
September 2013 during HOPE Melpitz. The intercomparison shown in Fig. 7
presents a histogram of the vertical velocities observed with ACTOS (red) and
WiLi (blue); further insights into the microphysical properties of the cloud
layer are given in Sect. 3.4 and Fig. 12. The variability of the vertical
velocities (with the mean adjusted to 0 m s
Furthermore, a combination of lidar and microwave radiometer data has been used to infer the height of the stable nocturnal boundary layer from aerosol-induced lidar backscatter variance and microwave-radiometer-derived potential temperature profiles (Saeed et al., 2016).
Besides wind vectors, profiles of atmospheric temperature and humidity are
the main drivers of numerical weather forecast models and key for the
verification of climate and Earth system models. An overview of their
importance and the requirements set to observing systems is presented in
Wulfmeyer et al. (2015). For models explicitly resolving turbulent processes
(such as the HD(CP)
Simultaneous observation of the vertical-velocity variations in a stratocumulus layer performed in-cloud with ACTOS (red) and at cloud base with Doppler wind lidar WiLi of LACROS (blue) on 22 September 2013 during HOPE-Melpitz. The mean vertical velocity of both observations was set to zero to correct for large-scale vertical motions. Adapted from Seifert et al. (2017).
From the multi-sensor observations available for the HOPE-Jülich
experiment, Steinke et al. (2015) investigated the comparability and range of
applicability of various sensors for the determination of the integrated
water vapour (IWV). As can be seen in Fig. 8, in general a good agreement
was found between the IWV observations from Global Positioning System (GPS)
stations (Gendt et al., 2001), microwave radiometer, sun photometer, and
radiosonde. The systematic difference and standard deviation were derived to
be approximately 0.4 and 1 kg m
Observation of the integrated water vapour (IWV) during HOPE-Jülich for a large suite of different instruments. Right panel shows the frequency distribution of the IWV values recorded with the different techniques. Bottom panel shows the accumulated amount of precipitation. Adapted from Steinke et al. (2015).
Calibrated night-time observations at KRA of the water vapour mixing
ratio for April 2013 during HOPE-Jülich obtained from Polly
A technique that is considered to provide accurate, continuous,
height-resolved observations of the water vapour mixing ratio is the Raman
lidar. Nevertheless, the stability of the system calibration is still the subject
of research and may depend on the design of specific systems. Based on
observations with the Raman polarization lidar Polly
Aerosol target classification for the HOPE-Jülich period from 24
to 26 April 2013 (IOPs 6–8) based on continuous observations of the
multi-wavelength polarization lidar Polly
Based on scanning measurements with the water vapour DIAL of IPM made during
HOPE-Jülich, Späth et al. (2016) (see Sect. 2.1.1) presented a
detailed study of the 3-D structure of the water vapour field
between the supersites HAM, KRA, and JUE with a range resolution of 30–300 m
and a temporal resolution in the range of 10 s for each profile. Full conical
scans (360
The retrieval and evaluation of microphysical properties of aerosols, clouds, and precipitation from ground-based remote-sensing observations is a crucial task. In situ observations do provide much higher accuracy but for the long-term evaluation of the performance of operational weather forecast models and the microphysical parameterizations therein continuous datasets are required. In particular the HOPE-Melpitz campaign provided the opportunity to relate in situ observations of warm-cloud microphysical properties and of aerosol properties from ACTOS to the respective parameters observed with ground-based observations of the LACROS facility. Case studies are presented in the following that document the simultaneous ground-based remote sensing and in situ observations of a stratocumulus layer and of the aerosol properties in the lower troposphere, respectively.
Aerosol particles act as nuclei for cloud droplets and ice crystals and are
thus a prerequisite for the formation of clouds. Lidar is a promising tool to
provide estimates of the concentration of CCN and ice nucleating particles (Mamouri and Ansmann, 2016). During
HOPE-Jülich and HOPE-Melpitz the Raman polarization lidar Polly
Retrievals of microphysical aerosol properties, such as CCN concentration,
from lidar observations as well as retrievals of the ambient scattering
properties of an aerosol population measured in situ are still subject to
large uncertainties. In situ observations of aerosol properties are usually
performed under dry conditions and inlets are limited by a maximum cut-off
size of an aerosol distribution. During HOPE-Melpitz, both in situ aerosol
observations and lidar observations of Polly
Correlation between the particle extinction coefficient derived from
Mie modelling and hygroscopic-growth correction of in situ measurements of
ACTOS with the respective ones measured with Polly
During HOPE-Jülich the availability of CCN was investigated using an aerosol model. The approach presented by Hande et al. (2016) used the COSMO-MUSCAT model to simulate the generation and transportation of aerosols over Germany during the campaign. From the simulation results, a parameterization of the CCN concentration was derived which can be applied also to other climatological regions and different aerosol regimes. Even though the simulated aerosol properties were evaluated against in situ observations of aerosol particle size distributions at Melpitz, no evaluation of the CCN parameterization against measurements was performed. This emphasizes the need to improve remote-sensing techniques for the retrieval of CCN profiles as the one of Mamouri and Ansmann (2016).
At the beginning of the first phase of HD(CP)
The implemented Frisch-2002 (Frisch et al., 2002) retrieval of cloud droplet effective radius and the Cloudnet retrieval of the adiabatically scaled LWC were evaluated against in situ observations of ACTOS for a stratocumulus deck observed simultaneously by ACTOS and LACROS during the HOPE-Melpitz campaign on 22 September 2013 (IOP 22) from 09:59 to 10:16 UTC, as is shown in Fig. 12. During the time period, ACTOS constantly flew horizontal legs of 2 km length in cross-wind direction in a distance of about 500 m upwind of the LACROS site. Time–height cross sections from the continuous LACROS observations as shown in Fig. 12a and b will be available in the SAMD database (Sect. 2.2.3) for all of HOPE-Jülich and HOPE-Melpitz. The comparisons of the average vertical profiles of LWC and cloud droplet effective radius observed with ACTOS and retrieved with LACROS are shown in Fig. 12. It can be seen that ACTOS probed mainly the mid-upper part of the cloud layer. Both the observations of the LWC of the cloud droplet effective radius of ACTOS and LACROS (Fig. 12a) are within the range of one standard deviation, as is shown by the horizontal error bars. Beside the found absolute differences, the profiles of LWC and effective radius retrieved from the LACROS observations deviate more strongly from those of ACTOS toward cloud top. A possible explanation for the observed discrepancies is the temporal variability of the LWC and effective radius in the cloud-top region as is shown in Fig. 12a and b. Also, ACTOS was not flying directly above the LACROS site. Considering the applied retrieval of Eq. (5) in Frisch et al. (2002), the assumption of a certain shape of the size distribution and of a cloud droplet number concentration can introduce biases. The application of the co-located observations of ACTOS and LACROS for the evaluation of ground-based retrievals will be discussed in an upcoming publication (Seifert et al., 2017).
Stratocumulus observation at the Melpitz site on 22 September 2013.
Time–height cross sections of
Relationship between mean ice water content (IWC) and ice-to-liquid mass ratio as a function of cloud-top temperature of all thin supercooled stratiform clouds detected during HOPE-Jülich. The colours represent the different radar linear depolarization ratios.
The accurate representation of the ice phase in numerical models is a crucial
task since cold rain is the main driver of precipitation formation at
midlatitudes (Mülmenstädt et al., 2015). The continuous observations
of the LACROS supersite during HOPE-Jülich enabled us to obtain statistical
information about the primary ice production in stratiform midlevel
mixed-phase cloud layers. Figure 13 shows an overview about the ice water
content (IWC) and ice-to-total mass ratio of all mixed-phase cloud layers that were
identified from the HOPE-Jülich observations. In these plots the method
for measurement of ice formation efficiency of Bühl et al. (2016) is
used, which selects supercooled thin stratiform cloud layers with a turbulent
mixed-phase (liquid-dominated) cloud top of a vertical extent of less than
380 m. In this way, non-linear ice formation effects like ice multiplication
or splintering are avoided and, thus, do not affect the statistics. IWC is
measured 60 m below the base of the mixed-phase layer, where an observation
of the falling ice particles is possible without influence of water droplets
or turbulent motions. LWCs are mean values of the scaled-adiabatic approach
(Merk et al., 2016) averaged over the complete height of the shallow
mixed-phase top layer of the cloud where liquid water is present. As shown in
Fig. 13, the IWC of clouds with top temperatures above
The combination of scanning polarimetric X-band Doppler rain radars,
vertically pointing micro rain radars, and a ground-based network of
disdrometers and rain gauges provided an excellent opportunity to validate
the Doppler rain radar's ability to infer the spatial variability of
quantitative precipitation properties from polarimetric radar reflectivities.
Xie et al. (2016) performed a detailed analysis of all precipitation
observations under different synoptic conditions. As an example, Fig. 14
shows a time series of the surface precipitation rates estimated from
measurements of three Doppler rain radar compared to the in situ observations
from seven disdrometers (partly from TR32 and TERENO projects), averaged over
the disdrometer locations. The authors note that rainfall accumulations at
the daily and even hourly scale were surprisingly consistent between the
different observations of rain gauges, disdrometers, and X-band radar, at
least for the low-intensity rainfall events (of 0.5–20 mm day
Time series of rain rates derived from observations of seven disdrometers (including those from the TR32 programme) and the three polarimetric radars on 29 May 2013. The shaded grey area indicates the range of rain rates observed by the disdrometers with 1 min temporal resolution in the HOPE area, while the rain rate from the three polarimetric radar observations is calculated at the radar gates that are coincident with disdrometer locations and also averaged over the disdrometer locations. From Xie et al. (2016).
Ground-based cloud photography provides the most detailed qualitative information on cloud patterns at high spatial and temporal resolution. Consequently, up to six sky imagers were operated in the SKY network during HOPE-Jülich. The combination of several imagers allows also for a quantitative retrieval of the spatial cloud structure. Beekmans et al. (2016) presented an approach for a spatial cloud reconstruction by using two hemispheric sky imagers in a stereoscopic set-up. They combined a dense stereo correspondence technique and a large-scale stereo set-up to derive 3-D cloud geometries. Obviously, such a stereoscopic cloud reconstruction is best suited for convective clouds that exhibit strong 3-D spatial features. Important aspects of such a technique include an accurate camera calibration (internal projection and camera orientation in space), precise synchronization, similar radiometric properties, and successful stereo matching on the rather fuzzy (diffuse) cloud images. As an example, Fig. 15 shows the determination of a cross section (panel d) from a reconstruction from a cumulus cloud (panel a). It was found that the near-zenith cloud base height is very well reproduced in comparison to lidar observations, yielding errors between 5 to 10 % for low- to mid-altitude cumuliform clouds. In general, Beekmans et al. (2016) provided a complete approach including geometric and radiometric corrections to obtain the spatial cloud envelope geometry for the cloud sides facing the sky imagers. Together with 3-D cloud information from scanning active systems such data will be very valuable for cloud reconstruction and radiation closure studies.
Three-dimensional reconstruction of a cumulus tower from a stereographic photograph
from 24 July 2014, 11:32:00 UTC.
Shown are
In the previous section, results of the HOPE observations were presented by
means of a summary of the different studies covering a large range of
meteorological processes from land-surface–atmospheric boundary layer
exchange and cloud and precipitation processes to the sub-grid variability and
microphysical properties of clouds and precipitation. Within this section the
application of these results for the evaluation of the newly developed ICON
model in LES mode as well as other LES and small-scale GCMs will be
summarized. A detailed overview about the set-up of the different models can
be found in Heinze et al. (2017). In general, ICON was run in LES mode on a
daily basis. Thus, usually the model was initialized at 00:00 UTC and
calculations were performed for a period of 24 h. The lateral boundaries
for the ICON runs were provided by the COSMO-DE model (Baldauf et al., 2011),
which is one of the operational models of the DWD. Within the boundaries of COSMO-DE, covering full Germany and the
Netherlands as well as parts of the other neighbouring countries, three ICON
domains, only slightly smaller than the COSMO-DE domain (47.6–54.6
Given the requirements on computational time and storage space the simulation days were chosen according to the appropriateness of the present weather conditions for the evaluation goals. A list of the HOPE days for which ICON runs are already available is provided in Table 5. It should be noted that the number of modelled HOPE days is subject to change in the future and that ICON runs for dates not covered by HOPE were also already performed but are not shown in here. The HOPE days selected for ICON runs cover a wide range of meteorological conditions, from clear-sky days for the evaluation of convective processes in the planetary boundary layer to days on which frontal passages accompanied by large-scale precipitation occurred. Most evaluation efforts were so far performed in a study of Heinze et al. (2017), but also others already made use of the extensive observational dataset. The studies available so far are discussed below.
Days of HOPE for which runs of the ICON model are available.
The observational studies presented in Sect. 3 demonstrate well that large efforts are being taken to make observations suitable for the initialization and the evaluation of numerical weather prediction (NWP) models and to provide process studies that are essential for their improvement. The high temporal resolution of the HOPE dataset allows an analysis beyond the mean, which offers new opportunities to improve the simulation of boundary layer dynamics. Vertical profiles of higher-order moments (variances and turbulent fluxes) can be derived (Behrendt et al., 2015; Van Weverberg et al., 2016), which are essential to advance higher-order closure parameterizations of turbulent transport schemes in numerical models. Recent LES studies analysed the underlying sources and sinks of such prognostic higher-order moment equations for the cloud-topped boundary layer (Heinze et al., 2015) and precipitating shallow cumulus regime (Schemann and Seifert, 2017). While these studies underline the importance, more robust conclusions are achieved by combining synoptically realistic model simulations with accompanying observational studies.
Nevertheless, operating a forecast model at scales that are small enough to
resolve the different supersites of the HOPE-Jülich campaign puts certain
requirements on the capabilities of the model. When the model resolution is
between LESs (with resolved energy-containing turbulence)
and mesoscale simulations (no turbulence resolved), the model is operating in
the so-called “grey zone” where more-sophisticated physical
parameterizations (e.g. for boundary layer turbulence or cloud microphysics)
might be needed. To what extent the parameterization of turbulence and
shallow convection is still necessary has been one of the key subjects of
HD(CP)
Temporal evolution of the boundary layer depth
A major goal of HD(CP)
The evaluation of actual LES simulations of the HOPE-Jülich area was done by Heinze et al. (2016) who performed simulations with PALM and UCLA-LES (University of California Los Angeles Large-Eddy Simulation model; Stevens et al., 2005) at up to 50 m horizontal resolution over the HOPE domain for a 19-day time period in order to capture a variety of different atmospheric and especially boundary layer conditions. The general weather pattern was reproduced in 80 % of the cases. Also cloud types usually agree well with observations. Resulting turbulence characteristics and boundary layer heights have been compared to observations from active remote sensing (Doppler lidar and aerosol lidar) and from in situ radiosonde observations as proposed by Schween et al. (2014). Figure 16 exemplarily shows the temporal evolution of the boundary layer height as derived from different model runs and from observations. The 2-hour (12:00–14:00 UTC) mean boundary layer depth derived with the PALM model agreed within 400 m to the different observation methods and to the COSMO-DE run at 2.8 km resolution. The found differences point to problems in the representation of ABL features in the LES and should be subject of further investigations. Please note that the criterion of model-based ABL depth is also subject to uncertainties which are explained further by Milovac et al. (2016), who found similar deviations between measurements and observations as found by Heinze et al. (2016). Heinze et al. (2016) further compared the observed turbulence characteristics of the ABL with the LES model. Observed and modelled profiles of the vertical-velocity variance agreed in their shape with the modelled values being in the range of uncertainty of the observations and showing slightly higher values throughout the boundary layer. Modelled profiles of potential temperature variances were found to be lower than the TRRL observations. For humidity variance, agreement within the uncertainty range was found in the lower and mid-CBL between measurements and LES models. But the modelled variance peaks at the CBL top showed an underestimation when compared with observations. Significant differences with respect to results from coarser-resolved COSMO simulations were not reported. This might in part be due to the so-called semi-idealized set-up with periodic boundary conditions and a homogeneous surface forcing. The authors also conclude that the long-wave and short-wave surface fluxes simulated with the LES model can be seen as representative in comparison to respective observations at five different sites in the HOPE area. The peak short-wave heat flux in the LES and COSMO-DE tends to be overestimated compared to the weighted average, whereas the long-wave heat flux tends to be underestimated.
Furthermore, within the synthesis module of HD(CP)
Regarding the application of HOPE observations for the initialization of NWP models, a first attempt was recently reported by Adam et al. (2016) who concentrated on 24 April 2013 (IOP 6). In their study the authors assimilated lower-tropospheric temperature profiles from the TRRL, reaching from about 500 to 3000 m above ground, into the Weather Research and Forecasting (WRF; Skamarock et al., 2008) model using a 3-D-variational method (Barker et al., 2004). The WRF model covered central Europe with 57 vertical levels and 3 km horizontal resolution. The assimilation of the temperature profiles from the TRRL and the assimilation of conventional data including zenith total delay integrated water vapour field from the Global Navigation Satellite System and operational radiosonde data were found to improve the agreement of measured boundary layer height and temperature gradient to the modelled values. Nine hours after the assimilation of TRRL data was initialized, an area of 100 km in radius around the HOPE-Jülich area was already affected, showing a temperature deviation from the conventional run of up to 2.5 K at 2.5 km height above sea level. Similar impacts can also be expected for the assimilation of profiles of water vapour mixing ratio from continuous lidar observations, as found in an earlier study of Grzeschik et al. (2008).
The HD(CP)
With the large number of in situ and Doppler wind lidar instruments, coherent structures in the surface near-boundary layer wind fields and characteristic integral scales have been identified and have been related to the type of external forcing. For the first time to our knowledge, TRRL demonstrated its capability to resolve the temperature inversion layer at the top of the ABL during daytime, which is key information for future process studies. Similarly, vertical temperature fluctuations have been observed for the first time by means of rotational Raman lidar measurements. It turned out that a temporal resolution of 10 s was sufficient to resolve turbulence structures down to the inertial sub-range from the mixed layer to the entrainment zone. Observed statistics of vertically resolved temperature fluctuations up to the forth-order moment provide important information on boundary layer dynamics and thermodynamics. The combination of daytime temperature and humidity profiles from Raman lidar and water vapour DIAL measurements with Doppler lidar measurements was used to obtain turbulent flux profiles in the convective boundary layer. In general, the combination of vertically resolved (lidar) and vertically integrated (microwave radiometer) and in situ (radiosondes) measurements of the atmospheric humidity has produced a unique 3-D field that together with wind and temperature measurements will serve as a solid constraint for the evaluation of high-resolution models. These results confirm the importance of high-resolution thermodynamic profiles for weather and climate research as demonstrated in Wulfmeyer et al. (2015). Surface solar and thermal radiation budget measurements complement the energy budget observations. A high-resolution pyranometer network produced statistics on spatiotemporal solar irradiance correlations for different sky conditions.
A comparison of turbulence measurements near cloud top from aircraft in situ measurements and from cloud base by lidar measurements revealed similar statistical properties, which points to a vertically homogeneous turbulence structure inside stratocumulus clouds.
Continuous operation of most of the instruments for 2 months made it possible to identify atmospheric variability from the micro- to the mesoscale. A long-term comparison of integrated water vapour from radiosondes and from ground-based and satellite remote sensing shows a generally good agreement but also revealed a bias of the spaceborne measurements towards lower values. Lidar observations of the aerosol profiles have been translated into the dominant aerosol type within each measurement volume. Such aerosol target classifications showed the hygroscopic growth of spherical aerosol particles under humid conditions as well as the presence of large non-spherical dust particles that were emitted from nearby sources. It turned out that the closure of in situ observations and remote sensing of aerosol microphysical properties is feasible when an extensive aerosol in situ characterization is available. A respective closure of cloud microphysical properties remains challenging due to uncertainties stemming from required assumptions on the particle size distribution and from spatiotemporal averaging. Cloud liquid water content profiles derived in situ and with remote sensing, however, were found to agree well. Continuous observations of mixed-phase clouds from a combination of active and passive remote sensing show that the ratio of ice to liquid water increases with decreasing cloud-top temperature, which serves as important information for the evaluation of ice formation parameterizations in cloud modelling.
Macrophysical cloud structures like cloud vertical dimension, cloud cover, cloud type, and precipitation fields have been continuously observed with lidar, radar, and sky imagers. Large-scale precipitation patterns together with the dominant process type for precipitation formation were observed with polarimetric Doppler precipitation radars. Three-dimensional cloud morphology has been retrieved from sky imagers in a stereoscopic set-up. Thus, a uniquely high-resolved dataset on cloud structural properties has been achieved during HOPE.
With the completion of the high-resolution ICON LES model a vast number of
model evaluation work is currently in progress. First evaluation studies
based on HOPE data have shown general agreement between observed and modelled
boundary layer height, turbulence characteristics, and cloud coverage, and
they
also point to significant differences that deserve further investigations
from both the observational and the modelling perspective. Although the
meteorological conditions which were prevalent during HOPE-Jülich and
HOPE-Melpitz enabled the collection of a broad set of observations, it is
obvious that the experimental coverage of the ABL
requires ongoing measurement efforts. In particular the continuous
observations from the German supersites will contribute to these efforts. The
supersites JOYCE, KIT, and LACROS that have been deployed during
HOPE-Jülich continue their long-term measurements at their base
institutes and will contribute to further process and model evaluation
studies in conjunction with further national and international supersites
like Barbados (13.2
Future work will take advantage of the synergy of the different active and passive remote-sensing measurements. For instance, Doppler lidar and polarimetric radar measurements may link dynamical forcing (up and downdrafts) with microphysical processes (riming, coagulation, ice formation). The cloud radars of JOYCE, KITcube, and LACROS were occasionally operating in a synchronized scan mode. Together with vertically pointing and scanning microwave radiometer data, three-dimensional distributions of cloud liquid water may be constructed and may get even further refined from cloud structure stereoscopy from synchronized sky imager data. Radiation closure studies will be performed based on observed and modelled spatial cloud structures and observed surface radiation budget measurements. High-resolution irradiance data can be used to build stochastic irradiance simulators for specific cloudy-sky conditions, which in turn can be used to construct realistic cloud-induced solar radiation variability. Combined measurements of temperature, humidity, and vertical wind fluctuations in the PBL under different meteorological conditions will provide important statistical information for improved turbulence parameterizations. HOPE also demonstrated the future potential of the synergy of scanning wind, temperature, and water vapour lidar systems for 3-D studies of land–atmosphere exchange and ABL entrainment in heterogeneous terrain. HOPE data may also reveal to what extent variations in aerosol concentrations and thus in CCN and IN concentrations have an effect on cloud and ice formation compared to dynamical forcing.
In future, HOPE data will continue to contribute to the development,
evaluation, and improvement of high-resolution NWP and LES models because the
data will be available via the SAMD database, which fulfils the needs of model experts.
Focused on the ICON development and the collection of observational data
for model evaluation, phase 1 of HD(CP)
Thanks to the valuable efforts of the community of observers during the HOPE campaigns and given its open-access availability in the SAMD database (see Sect. 2.2.3), the HOPE dataset can serve as excellent tool for the model evaluation and initialization community.
Data availability is discussed in Sect. 2.2.3.
The authors declare that they have no conflict of interest.
The work summarized in this review was mainly carried out in the project
HD(CP)
HOPE is particularly grateful to the research centre Jülich and RWE Power AG (Hambach) that provided generous logistic support during the Jülich campaign. We thank the Transregional Collaborative Research Centre 32 “Patterns in Soil-Vegetation-Atmosphere Systems – Monitoring, Modelling and Data Assimilation” for contributing their valuable rain observation research infrastructures to the Jülich campaign. The authors thank the German Climate Computing Center (DKRZ) for supporting the model simulations.
The universities of Cologne and Bonn as well as TROPOS secured intense radiosonde observations from internal budgets. Raman lidar system BASIL was funded on the basis of a specific cooperation agreement between Scuola di Ingegneria – Università degli Studi della Basilicata, TROPOS, and MPI Hamburg.
We appreciated the provision of four sun photometers for HOPE-Jülich and one device for HOPE-Melpitz by Goddard Space Flight Center, Greenbelt, MD, USA. Edited by: H. Russchenberg Reviewed by: two anonymous referees