A new method for the retrieval of ice crystal number concentration (ICNC) from combined active remote-sensing measurements of Raman lidar, cloud radar and radar wind profiler is presented. We exploit – for the first time – measurements of terminal fall velocity together with the radar reflectivity factor and/or the lidar-derived particle extinction coefficient in clouds for retrieving the number concentration of pristine ice particles with presumed particle shapes. A lookup table approach for the retrieval of the properties of the particle size distribution from observed parameters is presented. Analysis of methodological uncertainties and error propagation is performed, which shows that a retrieval of ice particle number concentration based on terminal fall velocity is possible within 1 order of magnitude. Comparison between a retrieval of the number concentration based on terminal fall velocity on the one hand and lidar and cloud radar on the other shows agreement within the uncertainties of the retrieval.
Aerosols, clouds and precipitation are major components of Earth's climate system. The complex aerosol–cloud-dynamics interaction currently poses major challenges for the numerical modeling of climate and weather phenomena because the majority of rain formation on Earth happens through the ice phase
To date, ice nucleation has not been able to be observed directly in the atmosphere, but we are gaining the ability to retrieve ice crystal number concentration (ICNC, further designated as
Comparison of two clouds with different ice crystal number concentrations
Aircraft observations have been used frequently for measuring
Today, all remote-sensing approaches for retrieving
In the present work, we make use of these new measurements and present an alternative approach for the retrieval of
The paper is structured as follows. Section
The remote-sensing data used in the context of this work were acquired during the COLRAWI-2 campaign (Combined Observations with Lidar, Radar And Wind profiler) at the Richard Assmann Observatory (RAO) of the Deutscher Wetterdienst in Lindenberg, Germany, between 1 June and 30 September 2015
A Cloudnet
The retrieval presented in this paper is based on measurements of radar reflectivity factor Vertical air motions shift the Doppler spectrum in such a way that Turbulence and beam width effects broaden the Doppler spectrum
The shift induced by vertical air motions can be removed if the magnitude of the vertical air motion
Mie scattering effects are neglected in the context of this work because we aim on studying pristine ice particles. Signal attenuation by these ice particles is also negligible. Turbulence and beam width broadening are also introducing artifacts. A strongly broadened CR Doppler spectrum might contain unphysical negative terminal fall velocities even after the correction of mean vertical air motion. Such effects cannot be removed easily, but luckily they only affect the width and not the mean velocity of the spectrum, which is shown in Sect.
In the present work, a method for deriving
The basic measurement values that are used in the retrieval are the first three moments (
Illustration of the general idea of the retrieval.
Flowchart of the retrieval algorithm, illustrating the synthesis of data from the remote-sensing instruments with the simulated parameters.
Environmental factors affecting the shape of the CR Doppler spectra are taken into account during the computation of the lookup table. The signal strength of the CR is, e.g., affected by attenuation from water vapor and liquid water particles; air motion shifts the CR Doppler spectrum and turbulence broadens it. In the context of this work, water vapor attenuation is corrected with the method of Cloudnet
The modified gamma distribution from
PSDs are simulated for all particle types mentioned in Table
Ranges and step sizes for the computation of the lookup table.
The extensive properties
A CR Doppler spectrum of
Accordingly, the extensive variables normalized to 1 particle m
From the two latter equations, the reflectivity-to-extinction ratio
For means of completeness, the mean terminal fall velocity measured with a Doppler lidar is given as
Simulation of microphysical parameters is only done over the ranges in which the particle properties are valid, which are given in Table
As mentioned before, a proxy for particle size is the most crucial intensive parameter for the retrieval of
A CR Doppler spectrum is computed on the terminal-velocity grid by computing
Such spectra are computed for a variety of input parameters (
with the subscript L indicating that these vectors are members of the lookup table. This lookup table, containing the three vectors
The basis for the retrieval of
Relationship of
For retrieving
Step-by-step description of the retrieval principle.
Example of a retrieval of
For estimating the uncertainties introduced by a measurement value on the retrieved quantities, the retrieval is performed for a fixed set of input parameters, and afterwards each single parameter is varied by 1 standard deviation. The errors are an estimation of the maximum measurement accuracy that can be achieved currently. Table
Ratios between the results of the original and the disturbed retrieval. The row labeled mean gives the original input parameters, and the row labeled error is the range that has been used for variation.
The measurement errors of the parameters
In this analysis, only methodological errors and random measurement errors have been assumed. Also the influence of uncertainties in the calculation of fall speeds and choice of particles are left out of the estimation of uncertainties here. The assessment of their influences is actually not straightforward.
Figure
The retrieval is done with three forms of the measurement vector which is used for the retrieval.
In the
Results of the retrieval based on
In the
A
The same as Fig.
In all three modes, uncertainties are derived from the distribution of
The distribution of retrieved results for
Averaged and median results for
Means, median and errors for the retrieval of
A method has been demonstrated to retrieve the size and shape parameters of PSDs from a combination of
The method has its limitations only in the signal thresholds of the instruments (CR, lidar and/or RWP) and the a priori knowledge about the shape of the observed particles. In the current study, the CR has a signal threshold of about
The paper presents a first step towards the usage of the unique direct measurements of
The presented method is essentially applicable to all remote-sensing facilities that provide a lidar particle extinction coefficient and CR (e.g., from the European Aerosols, Clouds and Trace Gases Infrastructure or the US program for Atmospheric Radiation Measurement). Combinations of lidar and radar prove most robust in terms of retrieval uncertainties. However, an error of less than 0.1 m s
The forward model used here is transparent and instructive, but other forward-iteration methods might be used in the future as well. In the context of this work, the lookup table approach is used primarily for an analysis of retrieval uncertainties due to input measurement errors. Typing of the pristine particles on the basis of radar depolarization measurements is crucial for the methodology presented here. It is currently a field of intensive research
Several issues need a solution for successful application of the method in future.
Automatic particle typing must be improved. Recently developed methods employing scanning techniques Uncertainties in CR calibration have not been taken into account here because those errors are essentially unknown. However, great effort is being made to come up with a solution for this problem. Matching between the CR and lidar beam has to be improved in order to avoid artifacts under complex situations. Direct information about local turbulence has to be taken into account to avoid errors in the estimation of the shape parameter
Given the downside of less flexibility, there are distinct advantages to the lookup table approach over classic forward-iteration methods; e.g., all possible results within the uncertainty range of the input variables are found at once. There is no risk that the method gets stuck in a local minimum. A lookup table approach also has the distinct advantage that numerical forward modeling and the actual retrieval are fully separable. Challenges to the approach are the extensive memory needs and the need for more effort in the evaluation of the results. The method is transparent, and it can be implemented easily in a numerically very efficient way. The computation of a case study as shown, e.g., in Fig.
The cloud radar data used in this work are available via the database of the Aerosol, Clouds and Trace Gases Research Infrastructure (ACTRIS) at
Table
Values for
Combined particles types used in the context of this work.
The lookup table used in this paper is computed of all particle types with a single parameterization from Table
This appendix gives a detailed description for the actual calculation of
The density of the air,
The two constants
With these results,
Parameters for velocity calculations
JB initiated the COLRAWI measurements, conceived the retrieval method and wrote the paper. PS and MR contributed methods for data evaluation and interpretation. HB evaluated the Raman lidar data. AA supervised the work and supported the preparation of the paper.
The authors declare that they have no conflict of interest.
This article is part of the special issue “BACCHUS – Impact of Biogenic versus Anthropogenic emissions on Clouds and Climate: towards a Holistic UnderStanding (ACP/AMT/GMD inter-journal SI)”. It is not associated with a conference.
We thank Volker Lehmann, Ulrich Görsdorf and Ronny Leinweber of MOL/RAO Lindenberg for their cooperation and for performing the measurements with RWP, CR and DL. The Polly
This research has been supported by the Deutsche Forschungsgemeinschaft (grant no. 398285025), the European Union's Framework Programme for Research and Innovation, Horizon 2020 (ACTRIS-2 (grant no. 654109)), the former European Commission Seventh Framework Programme FP7/2007–2013 (ACTRIS (grant no. 262254); BACCHUS (grant no. 603445)) and the European Union (Cloudnet (project no. EVK2-2000-00611)). The publication of this article was funded by the Open Access Fund of the Leibniz Association.
This paper was edited by Johannes Schneider and reviewed by two anonymous referees.