1. Introduction
Climate models use the parameter cloud fraction (CF) to determine the radiative fluxes through the atmosphere and at the surface. A small increase of 4% in the area of low level stratus clouds would be sufficient to offset the 2–3 K predicted rise in global mean temperature due to a doubling of the atmospheric CO
concentration [
1]. Arrays of ground based sensors located in diverse climate regimes have been used for more than 20 years by the Atmospheric Radiation Monitoring (ARM) Program to study the impact of clouds on the radiation budget (
https://www.arm.gov/about/history) [
2]. Climatologies of CF are considered for the choice of suitable sites for solar power plants [
3]. Further, supercooled liquid water clouds are a serious hazard for aviation [
4]. For all these reasons, long-term monitoring of CF of liquid water clouds is required. An intercomparison of different measurement techniques for thin liquid water clouds and radiative fluxes was given by [
5]. They emphasized that large differences in retrievals of liquid water amount and droplet size still must be resolved.
Ground-based microwave radiometers have the advantage that they continuously monitor the integrated liquid water (ILW) along a line of sight throughout the cloud under all weather conditions during day- and nighttime [
6,
7]. The measurement noise of integrated liquid water can be as small as 0.77
m (or 0.77 g/m
) as determined by [
7] for the TROWARA radiometer at Bern, Switzerland. Contrary to [
7], the present study utilizes the infrared brightness temperature of the
m channel of TROWARA in addition to the microwave channels for the data interpretation. This enhanced measurement information permits us to separate for thick supercooled liquid water clouds and thick warm liquid water clouds. In addition, the small measurement error of TROWARA supports the reliable detection of thin liquid water clouds with ILW between 0.0023 and 0.03 mm.
In the following, we give a brief overview of past studies on the climatology and trends of CF of liquid water clouds. A 10-year cloud fraction climatology was derived from TROWARA data from 2004 to 2014 [
6]. The CF values varied from 40% during summer to 60% during winter [
7]. The authors of [
6] found that the regional analysis of the Weather Research and Forecasting (WRF) model underestimates the regional cloud fraction at Bern (17% for WRF and 40% for TROWARA in summer 2012 at Bern). The High Resolution Infrared Radiometer Sounder (HIRS) onboard the NOAA satellites observed a quite constant cloud cover over an interval of 22 years with a cloud fraction of 75% [
8]. However, cloud fraction of high clouds in the upper troposphere showed a significant trend of +0.2%/yr. Cloud fraction from the Terra and Aqua Moderate Resolution Imaging Spectroradiometers (MODIS) showed a nice agreement with ground-based cloud fraction observations at several Swiss stations from 2000 to 2012 [
9]. The mean cloud fraction of MODIS was between 59% and 65% at the Swiss sites. The bias of MODIS with respect to the ground-based observations of CF was between −2.5% and +5%.
Data from the International Satellite Cloud Climatology Project (ISCCP) showed negative trends in CF (−0.3%/yr) during dry season and positive trends (0.1%/yr) during wet season in Amazonia over a time interval of two decades [
10]. Based on ISCCP data from 60
S to 60
N, it was found that CF of upper level clouds over land decreased by 1.5% from 1971 to 1996 [
11]. In [
12], the authors compared cloud parameters from 12 different satellite data sets. The global mean of CF ranged between 55% and 73%. The last IPCC report (Intergovernmental Panel on Climate Change) concluded that substantial ambiguity and therefore low confidence remains in the surface-based and satellite-based observations of global-scale cloud variability and trends [
13]. The ambiguities can be caused by different spatio-temporal sampling, changes in the aspect angles and other reasons. In the following, we show that ground-based microwave radiometry is a stable and objective method for long-term monitoring of CF.
2. Instrument, Measurement Technique and Data Analysis
The study is based on the measurements of the TROpospheric WAter RAdiometer (TROWARA). TROWARA is a dual-channel microwave radiometer built by [
14]. It provides vertically-integrated water vapour (IWV) and vertically-integrated cloud liquid water (ILW), also known as liquid water path (LWP). TROWARA is located inside a temperature-controlled room on the roof of the EXWI building of the University of Bern (46.95
N, 7.44
E, 575 m a.s.l.). Since TROWARA is operated indoors, it is capable to measure IWV even during rainy periods.
The two microwave channels are at 21.4 GHz (bandwidth 100 MHz) and 31.5 GHz (bandwidth 200 MHz). The lower frequency is more sensitive to microwaves from water vapour, and the higher frequency is more sensitive to microwaves from atmospheric liquid water.
The radiative transfer equation of a plane-parallel atmosphere is
where
the observed brightness temperature of the
i-th frequency channel is (e.g., 21 GHz).
is the opacity along the line of sight of the radiometer, and
is the contribution of the cosmic microwave background.
is the effective mean temperature of the troposphere [
15,
16].
From Equation (
1) we can derive the opacities
where the radiances
are measured by TROWARA.
The opacity is closely related to IWV and ILW by a quasi-linear relationship
where the coefficients
and
are not really constant since they can partly depend on the air pressure which is the case at 31 GHz. The authors of [
16] show that these coefficients can be statistically derived by means of coincident measurements of radiosondes.
is the mass absorption coefficient of cloud water. It depends on temperature (and frequency), but not on pressure. It is derived from the physical expression of Rayleigh absorption by clouds [
16]. Once the coefficients are determined, combined opacity measurements at 21 and 31 GHz permit the retrieval of IWV and ILW from Equation (
3). Thus, a dual channel microwave radiometer can monitor IWV and ILW with a time resolution of 6–11 s and nearly all-weather capability during day and nighttime.
An infrared radiometer channel is operated at
m, which measures the physical temperature at the cloud base when the cloud is optically thick (ILW > 0.03 mm). TROWARA’s antenna coil has a full width at half power of 4
and is pointing towards the sky at an zenith angle of 50
towards south-east. All the time, the view direction is constant, and the microwave and infrared channels of TROWARA observe the short-term temporal variations of the brightness temperature in the same volume of the atmosphere. This contributes to the high sensitivity of TROWARA for cloud detection. Further details of the measurement instrument and retrieval technique are given in [
7,
16].
TROWARA has been operated since 1994, and it has delivered an almost uninterrupted time series of ILW since 2004, with a time resolution of 11 s until end of 2009 and 6 s afterwards. The cloud detection in the line of sight of TROWARA is performed with the same time resolution, and the criterion is that ILW
mm. The authors of [
7] determined the instrumental noise
mm of TROWARA from the noise of ILW during 245 days in which the sky was free of clouds. If a ILW value exceeds the
level, then we are confident by 99.7% that the ILW value was generated by a cloud and not by instrumental noise. We emphasize that this is a remarkable sensitivity for a microwave radiometer since 0.0023 mm corresponds to the small mass of 2.3 gram water per square meter. Contrary to the ILW series, the time series of IWV have been used since 1994 for trend analyses, as has been shown by [
17,
18]. So far, a trend analysis has not been performed for the TROWARA ILW and CF data. CF (cloud fraction) can be easily determined in the time domain, for example, CF is the quotient of the time intervals when ILW
mm and the total observation time. The time intervals are as small as 6 s for ILW data after 2009 and 11 s for ILW data before 2009. Thus, we set the cloud flag with a high temporal resolution (6 or 11 s) which is required because of the high spatio-temporal variability of clouds floating through the fixed line-of-sight of TROWARA. Further, TROWARA’s coincident ILW and infrared brightness temperature measurements permit to separate the liquid water clouds in four classes:
thin liquid water clouds ( mm < ILW < 0.03 mm),
thick supercooled liquid water clouds (ILW > 0.03 mm and K),
thick warm liquid water clouds (ILW > 0.03 mm and K),
all liquid water clouds (ILW > 0.0023 mm).
Quite similar criteria for the separation of supercooled liquid water clouds were described in the Section 3.2.2 Radiometer method of [
19]. The critical point is that the derived cloud distributions are possibly biased towards the low level clouds since the infrared channel mainly sees the cloud base of thick clouds. The authors of [
20] avoided this bias by using additional satellite data for the cloud-top temperature.
Thin liquid water clouds were in the focus of the study by [
21]. They derived the microphysical and optical properties of thin liquid water clouds and emphasized that these clouds should be considered in climate studies since these clouds are frequent and they change the radiative forcing of the climate system. Measurements indicated that the downwelling infrared radiance of a thin liquid water cloud is increased by about 60% compared to clear sky. The authors of [
21] reported that thin liquid water cloud areas are often located at the edges of and in the inter-region between clouds (
twilight zone of clouds).
Since TROWARA is not sensitive to ice clouds, CF of TROWARA is in general smaller compared to synoptic observations. In [
7], the authors found a CF difference of about 17% between TROWARA and synoptic observations in the same region over a period of 6 years. In addition, some of the very thin and tenuous clouds which are still visible by eye might be not seen by TROWARA. However, CF of different classes of liquid water clouds is a new data product and of high value for evaluation of weather and climate models. The trend analysis is based on the determination of the trend and its uncertainty by means of linear regression. The differences of the CF trend and uncertainty values with respect to the alternative bootstrap method [
18] are less than 0.01%/yr in the present study and negligible.
The present study is not a cloud type study which would require the evaluation of coincident observations by ceilometer, lidar, radiosonde and hemispherical sky camera. In our study, the terms
thin and
thick refer to the magnitude of the optical depth at microwave frequencies which is proportional to the liquid-water path. The terms should not be misunderstood by the geometrical thickness of the clouds which is not measured by the microwave radiometer. A statistical cloud type study was performed by [
22] for Payerne. Payerne is representative for Bern since Payerne is located just 40 km eastward of Bern in the Swiss plateau. In [
22], the authors found that the cloud type low-level stratocumulus (Sc) is most common at Payerne. About 70% of the classified clouds have the type Sc while about
are stratus-altostratus (St-As). The authors of [
22] also found that the relationship between the ILW value and the cloud type is ambiguous. There is only a tendency that a small ILW value is most likely connected to the occurrence of stratocumulus.