This study focuses on evaluating how well spaceborne CPRs are able to document the properties of warm marine boundary layer (WMBL) clouds. We define WMBL clouds as cloudy columns with the highest cloud top below 5.5 km or 500 mbar and warmer than 0 ^∘C. This definition limits our analysis to WMBL regimes not associated with mid- or high-level clouds aloft but does not exclude periods where multiple WMBL cloud layers overlap.

The next subsections describe how we extracted cloud and precipitation information from raw CloudSat CPR data to evaluate its performance (Sect. 2.1), ARM measurements which provide a benchmark (Sect. 2.2), and how we forward-simulated alternative spaceborne radar configurations (Sect. 2.3).

Figure 1Hydrometeor radar reflectivity measured on 27 February 2016 (a) by the KAZR located at the Eastern North Atlantic (ENA) observatory over the course of 24 h and (b) by the CloudSat CPR when it overpassed the 200 km radius region around the KAZR between 15:05:21 and 15:06:07 UTC. In (a) the blue line marks the time when CloudSat overpassed the KAZR, the red dots represent the location of the ceilometer-determined cloud base and the black dots represent the boundaries of the KAZR radar echo; the latter coincides with the center of the first and last radar range gates containing signal (postprocessing). In (b) blue dots represent the boundaries of the CloudSat CPR radar echo; they coincide with the center of the first and last radar range gates containing signal (postprocessing). Also plotted are the CloudSat radar reflectivity (c) raw, (d) for significant returns (CPR_mask ≥5), (e) for echoes deemed very weak and stronger (CPR_mask ≥6), and (f) for echoes deemed weak and stronger (CPR_mask ≥20).

Download

2.3 Forward simulations based on ground-based KAZR observations

Forward simulations are conducted to improve our understanding of the CloudSat CPR limitations and to identify possible modifications which could lead to improvements in the detection of WMBL clouds (results in Sect. 5). We forward simulate seven radar architectures. The first four are based on the CloudSat CPR's current configuration gradually improving each of its capabilities until it matches the configuration of the EarthCARE CPR. The EarthCARE CPR design includes several improvements over CloudSat, namely

1. a new asymmetrical point target response,

2. enhanced sensitivity,

3. a smaller field of view and integration distance, and

4. increased range oversampling.

The EarthCARE CPR will also be the first spaceborne atmospheric radar capable of documenting the movement of hydrometeors. This capability has been evaluated in several publications such as Schutgens (2008), Battaglia et al. (2013), Kollias et al. (2014), Sy et al. (2014) and Burns et al. (2016) and is beyond the scope of this study. The last two architectures are based on propositions made in the context of the future Aerosol and Cloud, Convection and Precipitation (ACCP) mission of the National Aeronautics and Space Administration (NASA; https://science.nasa.gov/earth-science/decadal-accp, last access: 30 April 2020). They both have increased range resolution but reduced sensitivity. Specifications for each radar configuration are given in Table 1 and Fig. 2.

Table 1Specifications of the forward-simulated radar configurations including information about whether or not their pulse-weighting function is symmetrical (sym.) or asymmetrical (asym.) in either the vertical or the along-track dimension.

^* Shape of the range-weighting function is depicted in Fig. 2.

Download Print Version | Download XLSX

Processed (i.e., filtered, corrected and calibrated) KAZR radar reflectivity observations (time–height) are used as input to the forward simulators. First, assuming a constant horizontal wind speed of 10 m s⁻¹, the KAZR time axis is converted to horizontal distance. Then, to emulate the surface reflectivity which is not seen by the KAZR, an artificial surface echo is added to the processed KAZR reflectivity field at 0 m altitude (see Appendix A for more information on how real CloudSat CPR observations were used to construct this surface echo). Each spaceborne radar configuration is simulated first by horizontally convolving the high-resolution (30 m×20 m) KAZR reflectivity fields using an along-track weighting function represented using a symmetrical Gaussian distribution covering a distance equivalent to 2 times the along-track field of view and then by vertically convolving the horizontally convolved reflectivity field using either of the two range-weighting functions depicted in Fig. 2. The asymmetrical range-weighting function is modeled after the point target response of the EarthCARE CPR which was obtained from prelaunch testing of the EarthCARE CPR (mission's engineering team personal communications, 2019). The symmetrical range-weighting function used (only) for the CloudSat_f forward simulation is modeled using a Gaussian distribution adjusted to produce a surface clutter echo profile similar to that observed by the CloudSat CPR postlaunch (more information in Appendix A). Finally, along-track integration is emulated by averaging the convolved profiles in sections dictated by the integration distance of each spaceborne radar without overlap between the sections. Note that these forward simulations are two dimensional and as such do not capture cross-track effects; also note that liquid attenuation and noise are not represented.

For cloud and precipitation characterization, the forward-simulated radar reflectivity fields are finally filtered for surface clutter (akin to reproducing the type of filtering achieved by applying the CPR_Cloud_mask). To do this, forward simulations of clear-sky conditions are used to estimate the vertical extent and intensity of surface clutter. For each radar configuration, for all heights affected by surface clutter, the clear-sky surface clutter reflectivity is removed from the forward-simulated radar reflectivity and only echoes with reflectivity at least 3 dB above the surface clutter reflectivity are conserved and deemed reliable. Otherwise, for all heights above the surface clutter, only those echoes with reflectivity below the radar MDS are filtered out.

Figure 1 illustrates examples of observations collected on 27 February 2016 near the ENA observatory. The ground-based KAZR radar and ceilometer detected the presence of a thin (up to ∼270 m) cloud layer whose properties varied throughout the day. Between 00:00 and 10:00 UTC (23:00 and 09:00 local time), cloud top height was observed to rise at a rate of roughly 21 m h⁻¹. Shortly after 10:00 UTC, the KAZR detected signs of drizzle below the ceilometer-detected cloud base height at 941 m. The vertical extent of this drizzle was observed to increase over the course of the day, until it eventually reached 87 m altitude (the lowest altitude at which the KAZR measures) around 20:00 UTC. Besides changes in cloud top and hydrometeor layer base height, the KAZR also measured changes in radar reflectivity over the course of the day with more intense radar reflectivity recorded coincidently with deeper drizzle shafts.

At 15:05 UTC, CloudSat overpassed within 200 km of the KAZR and ceilometer location (marked by the blue line in Fig. 1a). Although the subset of noise-and-clutter-filtered CloudSat CPR observations shows the presence of a hydrometeor layer, the hydrometeor layer detected by the CloudSat CPR had breaks, a higher top (1.28 vs. 1.07 km) and a higher base (1.15 vs. 0.51 km) than that detected by the KAZR, misleadingly making it appear thinner overall (Fig. 1b).

To illustrate how the aforementioned example is representative of the general picture of the WMBL cloud regimes at the ENA, we also compared statistics of hydrometeor layer properties estimated for all instances where CloudSat overpassed within 200 km of the ENA and boundary layer clouds were the dominant cloud type (Figs. 3 and 4; 103 out of the 138 overpasses). For this comparison, only KAZR and ceilometer observations taken within ±1 h of the overpass time are considered. The predominance of boundary layer clouds is established using KAZR observations. Instances with fewer than 30 % (in time) high or cold clouds are deemed dominated by boundary layer clouds; high or cold clouds present in these instances (if any) are filtered out of the analysis. This region size (for the spaceborne observations) and time period (for the ground-based observations) were selected to match those of Protat et al. (2009) and constitute a compromise between keeping the domain size small enough to maintain its homogeneity (∼99 % ocean by area) and capturing a number of cases large enough to reach statistical significance (103 overpasses).

Figure 2Symmetrical (blue) and asymmetrical (black) range-weighting functions for the forward-simulated radar architectures detailed in Table 1. Negative values are associated with the leading edge of the pulse in the direction of propagation.

Download

Figure 3For 103 instances where CloudSat overpassed the 200 km radius region centered on the ENA observatory, (a) fraction of observed profiles with cloud or rain (i.e., hydrometeor cover) and (b) hydrometeor fraction profile. Both estimated from CloudSat CPR observations within a 200 km radius of the ENA observatory (blue) and ground-based KAZR observations collected within ±1 h of the CloudSat overpass (black). Fractions are estimated based on the total number of observed profiles excluding those determined to contain high, deep or ice clouds.

Download

Figure 4For 103 instances where CloudSat overpassed the 200 km radius region centered on the ENA observatory, distribution of (a) echo top height and (b) echo base height, estimated from CloudSat CPR observations within a 200 km radius of the ENA observatory (blue) and ground-based KAZR observations collected within ±1 h of the CloudSat overpass (grey). For references are examples of hydrometeor radar reflectivity measured on (i) 11 February 2017 and (ii) 24 October 2016 by the ground-based KAZR within ±1 h of the CloudSat overpass and by the CloudSat CPR within 200 km of the KAZR location. Dots on these figures represent the boundaries of the radar echo (black and blue dots for the KAZR and the CloudSat CPR, respectively) and the location of the ceilometer-determined cloud base (red dots).

Download

First, agreement between the KAZR-reported cloud cover and the ceilometer-reported cloud cover confirms that the KAZR's sensitivity is sufficient for detecting even the most tenuous clouds forming in this marine boundary layer regime; this makes the KAZR an ideal sensor to document the properties of WMBL clouds and evaluate the CloudSat CPR's performance (Fig. 3a). Although not expected to perfectly match, the large hydrometeor cover discrepancy between the KAZR (48.1 %) and CloudSat CPR (27.2 %) suggests that the CloudSat CPR fails to detect clouds in more than a few (on the order of ∼40 %) of the atmospheric columns it samples (Fig. 3a). On the other hand, the CloudSat CPR seems to capture the shape and magnitude of the hydrometeor fraction profile above 1.0 km reasonably well (Fig. 3b). This suggests that the CloudSat CPR is able to detect the bulk of the thick hydrometeor layers controlling hydrometeor fraction above 1.0 km. This also leads us to believe that the CloudSat CPR's hydrometeor cover biases result either from its inability to detect clouds entirely located below 1.0 km and/or from its inability to detect thin and narrow hydrometeor layers that are negligible contributors to hydrometeor fraction. Detailed analysis of the location of individual cloud tops shows evidence supporting both of these postulations (Fig. 4a). Specifically (1) the distribution of KAZR-detected cloud top heights shows clouds below 0.6 km, most of which are undetected by the CloudSat CPR. We estimate that this near-surface cloud mode produces 4.5 % of the total cloud cover, and so its misdetection could explain nearly a quarter of the CloudSat CPR hydrometeor cover bias. (2) The distribution of KAZR-detected cloud top heights also shows the presence of cloud top modes near 1.2 km and frequent occurrences near 2.2 km that are only partially detected by the CloudSat CPR (Fig. 4a). These elevated cloud tops modes are likely related to the several echo bases between 1.5 and 2.0 km that nearly all went undetected by the CloudSat CPR (Fig. 4b). A figure showing time–height observations from 2 additional overpass days allows us to visualize these layers as generally thin, weakly reflective and broken (Fig. 4i and ii). We speculate that misdetection of such thin or tenuous clouds explains the remainder of the CloudSat CPR's cloud cover bias.

Beyond its inability to detect all cloudy columns, the CloudSat CPR also severely underestimates the presence of hydrometeors below 0.75 km because it suffers from surface echo contamination; this creates an artificial enhancement in the number of apparent hydrometeor layer bases estimated from the CloudSat CPR near 0.75 km and is not representative of the true height of the base of either clouds or virga (Fig. 4b). We believe that the surface echo limits the CloudSat CPR's ability to observed true cloud base in ∼52 % of the cloudy columns it detects and true virga base in ∼80 %; in other words, the CloudSat CPR often provides an incomplete view of even the WMBL cloud systems it does detect. This approximation is made based on the subset of cloudy columns observed by the KAZR, whose tops are above the CloudSat CPR surface clutter echo (1.0 km) and are likely of sufficient thickness (250 m) and reflectivity (Z > −28 dBZ) to be detected by the CloudSat CPR.

Figure 5From ground-based KAZR observations collected between October 2015 and November 2017, (a) profile of cloud (solid black line) and sub-cloud-layer rain (dotted black line) fraction and the fraction of either cloud (solid red line) or sub-cloud-layer rain (dotted red line) echoes located below a certain height. Fractions are estimated based on the total number of observed profiles excluding those determined to contain high, deep or ice clouds. (b) Fraction of hydrometeor (cloud or rain) echoes with reflectivity larger than a given reflectivity threshold (colormap) with superimposed surface clutter profile as simulated for the CloudSat_f (royal blue line), EarthCARE (magenta line), ACCP₂₅₀ (red line) and ACCP₁₀₀ (green line) CPR configurations and as observed by the CloudSat CPR between May 2010 and November 2017 (broken black line marks the median; dotted black lines mark the interquartile range). (c) Median profile of hydrometeor layer mean reflectivity as a function of thickness (black) and the fraction hydrometeor layers thinner than a certain thickness (red).

Download

Although these 103 CloudSat overpasses are reasonably representative of the properties of the WMBL hydrometeor systems found in the vicinity of the Eastern North Atlantic facility, considering the entire set of measurements collected by the KAZR between October 2015 and November 2017 (719 d) provides additional insight on the challenges associated with measuring the properties of these hydrometeor systems (Fig. 5).

Analysis of the ground-based observations suggests that WMBL cloud fraction exceeds 5 % at all heights between 320 m and 2.09 km, with cloud fraction peaking at 1.13 km (Fig. 5a; solid black curve). On the other hand, rain tends to be found in the subcloud layer below 1.28 km altitude, occupying the largest fractional area between 100 m and 1.1 km (Fig. 5a; dotted black curve). The low height at which WMBL clouds and precipitation are found is especially challenging for spaceborne radar systems which are known to suffer from contamination from the surface return. We estimate that roughly 20 % of the cloud echoes and 52 % of the rain echoes recorded by the KAZR fall within the CloudSat CPR's surface echo region which extends at best only to 0.75 km (Fig. 5a; red curves).

The intensity (in terms of radar reflectivity) of cloud and precipitation also largely affects their ability to be detected by radars. Using KAZR observations, we characterize the intensity of the hydrometeor echoes observed at each height and report in Fig. 5b (colormap) the fraction of echoes with a reflectivity above a given threshold at each height. Generally, cloud and precipitation that produce radar reflectivity above a radar MDS can be detected. Thus, we would expect that the CloudSat CPR, with its −27 dBZ MDS (observed performance depicted by the broken black line in Fig. 5b), should have the capability to detect at best 80 % of all cloud and/or echoes forming at any given height, de facto missing at least 20 % of hydrometeor echoes. Radar performance degrades within the surface clutter region. In the clutter region, only those hydrometeor echoes whose intensity is larger than the surface echo intensity can be detected. To reflect this and for reference, we overlaid on Fig. 5b the median reflectivity recorded by the CloudSat CPR on clear-sky days between 2010 and 2016 as well as its variability as quantified by the interquartile range (broken and dashed black lines, respectively). Over that time interval, the CloudSat CPR's median surface echo varied from 37 dBZ at the surface to −27 dBZ at 0.75 km. Using this curve, we estimate that at a height of 0.5 km, the CloudSat CPR would miss at least 80 % of the echoes detected by the KAZR because their reflectivity is below that of the surface clutter, and that is based simply on sensitivity and does not account for other effects.

Adding to the challenge is the fact that boundary layer systems are shallow. Based on KAZR observations, 53 % of WMBL systems (cloud and rain) that form at ENA are shallower than 500 m, 33 % shallower than 250 m and 16 % shallower than 100 m (Fig. 5c; red line). Sampling hydrometeor layers using radar pulses longer than the hydrometeor layer thickness inherently produces partial-beam-filling issues, which lead to a weakening of the returned power. This results in an underestimation of the reflectivity of the thin echoes sampled and may even lead to their misdetection if the resulting reflectivity is below the radar MDS. There is also an unfortunate relationship between hydrometeor layer thickness and mean reflectivity such that thin layers not only suffer from more partial beam filling but also have weaker reflectivities. The black curve in Fig. 5c shows the median hydrometeor layer mean reflectivity as a function of hydrometeor layer thickness. From this figure we can estimate that 500 m thick hydrometeor layers typically have a mean reflectivity of −21 dBZ, 250 m thick layers −26 dBZ and 100 m thick layers −33 dBZ.

Improving our ability to detect boundary layer clouds and precipitation could likely be achieved through radar system modifications including the following (not necessarily in order of importance):

1. Alter the point target response (which dictates the shape of the forward-simulated range-weighting function).

2. Decrease the minimum detectable signal (MDS).

3. Reduce the horizontal field of view.

4. Increase the vertical oversampling.

5. Reduce the transmitted pulse length.

Figure 6Based on KAZR observations of the hydrometeor layer of 27 February 2016, forward-simulated radar reflectivity (colormap) and estimated hydrometeor layer boundaries (colored dots) for (a) CloudSat_f (royal blue dots), (b) CloudSat_a which is CloudSat operating with the EarthCARE asymmetrical range-weighting function (cyan dots), (d) CloudSat_a+es which alternatively has an enhanced sensitivity equivalent to the EarthCARE (purple dots), (c) EarthCARE which additionally operates with a factor of 5 vertical oversampling (magenta dots), (e) ACCP₂₅₀ which instead has a 250 m range resolution (red dots) and (f) ACCP₁₀₀ which instead has a 100 m range resolution (green dots). For reference, the corresponding KAZR-observed radar reflectivity is depicted in Fig. 1a and echo boundaries identified by the KAZR are overlaid on each subpanel using black dots.

Download

We emulate the impact of these radar modifications by constructing forward simulations for seven radar configurations, each of which has been gradually improved by the aforementioned radar modifications (described in Sect. 2.3, Table 1 and Fig. 2). Quantitative assessment of the performance of the forward-simulated radar configurations is estimated based on a set of 719 forward simulations constructed from KAZR observations collected between October 2015 and November 2017. As carried out for the real CloudSat CPR observations in Sect. 3, performance is evaluated in terms of how well hydrometeor cover and hydrometeor fraction are captured (Fig. 7) as well as in terms of how accurately the boundaries of hydrometeor layers are detected (Fig. 8). However, since all forward simulations presented in this section are based on the same KAZR observations, we expect a perfect match and interpret any deviations from the KAZR observations as a bias. To help visualize the performance of the seven radar configurations, we present output from forward simulations of the 27 February 2016 hydrometeor layer. The KAZR's view of this hydrometeor layer was depicted and described in Fig. 1a and Sect. 3; for reference the KAZR's detected echo top and base are overlaid on each forward simulation in Fig. 6 using black dots.

First, we validate our forward simulation framework by simulating the CloudSat CPR's current configuration (results depicted in royal blue and designated as CloudSat_f for short). The forward simulations of the CloudSat_f show similar biases to the real CloudSat CPR when compared to the KAZR, indicating that the forward simulator captures enough of the radar's characteristics to reasonably emulate its performance. In a nutshell, the CloudSat_f underestimates hydrometeor cover by more than 10 % (Fig. 7a), likely owing to its misdetection of an important fraction of clouds with tops between 750 m and 1.75 km (Fig. 8a) and its inability to detect the small fraction of clouds forming entirely below 500 m. Just like the real CloudSat CPR, the CloudSat_f performs well in capturing hydrometeor fraction between 750 m and 3 km but poorly below that height since it suffers from contamination by surface clutter (Fig. 7b).

Prelaunch testing of the EarthCARE CPR showed that its particular transmitter and receiver filter generate an asymmetrical point target response. This means that, unlike the CloudSat CPR, the EarthCARE CPR must be represented by an asymmetrical range-weighting function (Fig. 2). The range-weighting function of the EarthCARE CPR's pulse has a rapid cutoff at a factor of 0.5 times the pulse length on its leading edge and a longer taper extending to 1.5 times the pulse on its trailing edge. To isolate performance changes resulting strictly from this change in point target response, we contrast the result of forward simulations performed with the CloudSat CPR's original configuration (CloudSat_f results depicted in royal blue) with a CloudSat-like configuration with the EarthCARE CPR's asymmetrical range-weighting function (CloudSat_a; results depicted in cyan). A time-series comparison of CloudSat_a (Fig. 6b) and CloudSat_f (Fig. 6a) reflectivity shows that the asymmetrical point target response causes a reduction in the vertical extent of the surface clutter echo, allowing for the detection of a larger fraction of hydrometeor at 500 m. Over the entire set of 719 forward simulations, this leads to improvements in the representation of the hydrometeor fraction profile (Fig. 7b) and of the echo base height distribution (not shown) around 500 m. However, differences in the echo base height from the KAZR (black dots) and from CloudSat_a (cyan dots) suggest that changes in the shape of the pulse point target response alone are insufficient to accurately detect the base of the precipitating WMBL systems found at the ENA (Fig. 6b). We also note that the change in shape of the point target response alone only marginally improves the ability of CloudSat_f to determine hydrometeor cover (improvement from 27.9 % to 28.2 % compared to 39.1 % reported by the KAZR); the reason for this is that hydrometeor cover is controlled by thin, tenuous clouds and clouds located entirely below 0.5 km. As a potential drawback, the asymmetrical point target response seems to lead to slightly more vertical stretching of cloud top signals (on average 37 m) such as those visible by comparing the examples in Fig. 6a and b and in Fig. 8a. When compounded over the entire ensemble of forward-simulated clouds, this leads to a 0.24 % overestimation of hydrometeor fraction at all heights between 0.75 and 3.00 km (Fig. 7b). The vertical stretching of cloud tops results from additional power being focused between a factor of 0.0 and 0.5 times the pulse length on the leading edge of the pulse (comparing the range-weighting function of EarthCARE CPR to that of the CloudSat CPR; the black and blue line in Fig. 2, respectively).

Figure 7For 719 forward-simulated days, (a) fraction of observed profiles containing either cloud or rain (i.e., hydrometeor cover); also, for the KAZR only, using complementary ceilometer observations, we estimate the fraction of all observed profiles containing rain in the subcloud layer. (b–d) Hydrometeor fraction profile estimated for all the forward-simulated radar architectures. All acronyms and colors are defined in Fig. 6 with the exception of CloudSat${}_{\mathrm{nps}+\mathrm{es}+\mathrm{hf}}$ which is the CloudSat CPR operating with EarthCARE's asymmetrical range-weighting function, enhanced sensitivity and half the horizontal field of view (gold).

Download

Besides having an asymmetrical point target response, the EarthCARE CPR will also operate with an MDS of −35 dBZ which is 7 dB more sensitive than the CloudSat CPR. To isolate performance changes resulting strictly from this sensitivity enhancement, we contrast the result of forward simulations performed with a CloudSat-like configuration with the asymmetrical point target response (CloudSat_a; results depicted in cyan) with that of a CloudSat-like configuration with both an asymmetrical point target response and enhanced sensitivity (CloudSat_a+es; results depicted in purple). A time-series comparison of CloudSat_a+es (Fig. 6d) and CloudSat_a (Fig. 6b) reflectivity shows that the sensitivity enhancement allows for the detection of hydrometeors in previously undetected columns such as the broken hydrometeor segments at around a 100 km distance along the forward-simulated track. Quantitatively, the more sensitive CloudSat CPR configuration detects 8 % more cloudy columns than either of the other two CloudSat CPR configurations discussed so far (i.e., with or without the asymmetrical point target response), reporting only 2.4 % less cloudy columns than the KAZR reports (Fig. 7a). This implies that, if an important mission objective is detecting even tenuous cloudy columns, improving the MDS is crucial. That being said, we advise against accomplishing this by transmitting a longer pulse (e.g., as done in the first 4 years of operation of the Global Precipitation Measurement (GPM) CPR) since there are two main drawbacks to transmitting a long pulse with a higher sensitivity, both caused by partial beam filling. Firstly, the enhanced sensitivity leads to additional vertical stretching of cloud boundaries, an effect visible between 400 and 800 km along-track when comparing Fig. 6d to b. This is because the signal from cloud boundaries measured away from their location by the edges of the radar range weighing function can then exceed the MDS. Secondly, the enhanced sensitivity also leads to previously undetected thin layers becoming detectable, but it stretches them vertically at least to the vertical extent of the radar pulse length. From changes in the location of the cloud top height distribution peak shown in Fig. 8a, we estimate that enhancing the sensitivity of a 3.3 µs long pulse from −28 to −35 dBZ would lead to a 250 m bias in detected cloud top height for the WMBL clouds that form at the ENA. Moreover, because it both vertically stretches clouds and detects more real clouds, the highly sensitive CloudSat_a+es overestimates hydrometeor cover by up to 7 % at all heights between 500 m and 3.0 km (Fig. 7b).

Since EarthCARE will travel at an altitude closer to the Earth's surface, it will also have half the horizontal field of view of CloudSat. Our results suggest that halving the CloudSat CPR's horizontal field of view and halving its integration distance would lead to a slight reduction in its estimated hydrometeor cover (1.7 % less). We take this as an indication that the larger horizontal field of view of the CloudSat CPR only marginally artificially broadens broken clouds (see CloudSat${}_{\mathrm{a}+\mathrm{es}+\mathrm{hf}}$; results depicted in gold in Fig. 7). That being said, note that this result, like all the others presented here, is based on a 2-D forward simulation and as such it does not take into account cross-track effects which may also generate biases especially in sparse broken cloud fields.

Another interesting radar configuration proposed by the EarthCARE mission advisory group concerns the amount of vertical oversampling of the radar pulse. Radar signals are typically oversampled by a factor of 2 effectively halving the vertical spacing between available measurements. The EarthCARE CPR will use a factor of 5 oversampling to increase its vertical range sampling to 100 m while still operating at a 500 m vertical resolution. While oversampling may be appealing because it creates a smoother view of cloud fields, it does not effectively improve the vertical resolution because of the correlations between the oversampled measurements. Evaluating the impact of these correlations on the observed radar reflectivity field is beyond the scope of this study which instead focuses on evaluating the impact of oversampling on accurately locating cloud and precipitation boundaries. The time series of EarthCARE (Fig. 6c) reflectivity shows that increased oversampling will allow for a more precise characterization of the variability in echo base and top height (also see the echo top height distribution presented in Fig. 8c). Comparison of the ensemble of EarthCARE (magenta) and CloudSat${}_{\mathrm{a}+\mathrm{es}+\mathrm{hf}}$ (gold) forward simulations indicates that this precision can be achieved without causing significant biases in hydrometeor cover (Fig. 7a) or hydrometeor fraction (Fig. 7c).

Figure 8For 719 forward-simulated days, distribution of echo top height observed by the KAZR (grey) and estimated from the forward-simulated radar architectures. Results are estimated at various range sampling resolutions according to the capability of each spaceborne sensor configuration. All acronyms and colors are defined in Fig. 6.

Download

Although the EarthCARE CPR's performance is significantly better than that of the CloudSat CPR when it comes to detecting thin, tenuous and broken clouds as well as clouds and precipitation near 500 m, its configuration still does not allow for the detection of all WMBL clouds and precipitation. Remaining detection limitations occur below 500 m within the region of the surface clutter echo. Additional reduction of the vertical extent of the surface clutter can be achieved by reducing the pulse length. This, however, comes at the expense of reduced sensitivity. Comparing EarthCARE (results depicted in magenta), ACCP₂₅₀ (results depicted in red) and ACCP₁₀₀ (results depicted in green) simulations allows us to see the gain and penalty incurred from shortening the radar vertical range resolution from 500 to 250 to 100 m at the cost of reducing sensitivity from −35 to −26 and −17 dBZ. In alignment with our previous conclusion that a high sensitivity is necessary for detecting all cloudy columns, reducing the radar pulse length and sensitivity reduces the fraction of cloudy columns which can be detected by the ACCP configurations (Fig. 7a). For instance, the ACCP₂₅₀ configuration, which is nearly as sensitive as CloudSat (−26 dB versus −28 dB), performs very similarly in terms of the number of cloudy columns it is able to detect (Fig. 7a) and in terms of how well it can capture the vertical distribution of hydrometeors between 500 m and 3.0 km (Fig. 7d), which we determined is influenced by the deeper more reflective clouds rather than the thin and tenuous ones. The ACCP₂₅₀ configuration does, however, have the advantage of providing information on the base of clouds and/or precipitation down to 250 m which is much more than the CloudSat CPR can achieve (Fig. 7d). The short pulse of ACCP₂₅₀ also helps mitigate the amount of cloud stretching related to partial-beam-filling issues, thus providing a more precise characterization of cloud top height (Fig. 8c; effects also visible in Fig. 6e). So generally speaking, reducing vertical pulse length reduces the fraction of detected cloudy columns but improves the characterization (both in terms of echo top and echo base location) of those cloudy columns which are detected.

Results also suggest that radars with shorter less sensitive pulses would be more suitable for the characterization of surface rain and virga, which are more reflective targets. In fact, we estimate that ACCP₁₀₀ would detect 18 % of the rainy columns compared to the 26.2 % detected by the KAZR (Fig. 7a). ACCP₁₀₀ would also do reasonably well at capturing the vertical distribution of drizzle and rain; comparisons of rain fraction profiles estimated from the KAZR (subcloud layer only) suggest that ACCP₁₀₀ would miss < 2 % of the virga forming at each height below 750 m and would be able to detect the presence of rain as close as 25 m above the surface.

All CloudSat CPR observations were obtained from the CloudSat data processing center (http://www.cloudsat.cira.colostate.edu/, last access: 12 June 2019, CloudSat, 2019). All ARM observations were obtained from the ARM archive (https://www.archive.arm.gov/discovery/, last access: 12 June 2019, ARM Data Discovery, 2019). Output of all forward simulations is fully reproducible from the information given.

KL coordinated the project, extracted the ground-based measurement files from the ARM archive, performed the data analysis and produced the final manuscript draft. PK extracted the CloudSat CPR GEOPROF product files from the data processing center and provided feedback on the forward simulator. AB provided feedback on the analysis methods as well as on the manuscript draft. SP performed exploratory data analysis and provided feedback on the manuscript draft.

The authors declare that they have no conflict of interest.

This research has been supported by the U.S. Department of Energy Biological and Environmental Research program (grant nos. DE-SC0016344 and DE-SC0017967).

This paper was edited by Vassilis Amiridis and reviewed by three anonymous referees.