1. Introduction
Since the launch of CloudSat (Stephens et al. 2002, 2008) and Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) (Winker et al. 2003, 2007, 2009) in April 2006, when they joined the A-Train constellation, they have provided detailed information about cloud vertical structures. Both satellites are equipped with active sensors; CALIPSO carries the Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP) lidar, and CloudSat carries the Cloud Profiling Radar (CPR; Mace and Wrenn 2013). Unlike passive sensors, CALIOP and CPR can directly measure cloud vertical profiles. Earlier studies have used CALIPSO and CloudSat measurements to obtain information of cloud profiles to validate passive-sensor cloud measurements or model simulations (e.g., Rossow and Zhang 2010; Kato et al. 2010; Stein et al. 2011; Cesana and Chepfer 2012; Ham et al. 2013; Li et al. 2015; Ham et al. 2017; Stephens et al. 2018; Yost et al. 2021). Most studies considered a relatively short period to examine seasonal variations or annual means of cloud properties, except a few studies (e.g., Stephens et al. 2018), since the data record of CALIPSO and CloudSat was relatively short in comparison with other passive-sensor cloud measurements.
CALIOP lidar was designed for a 3-yr life span, and CloudSat CPR was targeted for 22 months of observations (Stephens et al. 2018). These satellites have operated beyond their original life spans and remained within the A-Train constellation until the exit of CloudSat in February 2018 and CALIPSO in May 2018. Currently, over 11 years of CALIPSO and CloudSat measurements are available, from mid-2006 to early 2018, enabling us to examine relatively long-term cloud changes in comparison with the periods considered in earlier studies. Therefore, our objective is to use CALIPSO and CloudSat observations to understand how macrophysical cloud properties varied from 2007 to 2017. We will combine CALIPSO and CloudSat (CALCS) cloud boundaries and compute macrophysical properties based on the concept developed by Kato et al. (2010). Also, as a part of the A-Train constellation, the Aqua satellite flew equipped with a Moderate Resolution Imaging Spectroradiometer (MODIS) (Salomonson et al. 1989; Barnes et al. 1998) passive sensor. We will compare macrophysical cloud changes derived from CALCS and MODIS. This step is needed to answer the question of whether active and passive sensors generate consistent long-term cloud macrophysical changes.
We mainly focus on the tropical Pacific Ocean regions to understand how macrophysical cloud changes are associated with meteorological parameters such as temperature, humidity, and vertical motion perturbations. Over the Pacific, one of the major drivers of interannual variability of meteorological parameters is an El Niño–Southern Oscillation (ENSO) event, which in turn impacts the cloud macrophysical properties. The 11-yr CALCS period includes several major ENSO events; two El Niño events in 2009/10 and 2015/16, which are referred to as 2009/10 El Niño and 2015/16 El Niño, respectively, and one La Niña event lasted from 2010 to 2012. We investigate how the cloud macrophysical properties covary with meteorological conditions in different ENSO phases, and whether the consistent relations are obtained from MODIS and CALCS cloud observations. The observational-based relations shown in this study can be also used for validating cloud responses in model simulations.
Section 2 describes the datasets used in this study. Section 3 explains the method of how we generate CALCS and compute cloud area and volume. Mid- and high cloud changes are discussed in section 4, and low cloud changes are discussed in section 5. Different timings of cloud-top elevations and cloud area changes are discussed in section 6. Section 7 summarizes the findings of this study.
2. Data
a. A-Train satellite products
1) Cloud mask from CALIPSO, version 4, and CloudSat release-5 products
CALIOP aboard CALIPSO (Winker et al. 2003, 2007, 2009) is a dual-wavelength lidar, measuring 532-nm/1064-nm backscatter profiles. CALIPSO level-2 vertical feature mask (VFM) product (Vaughan et al. 2009) provides feature classifications such as cloudy, aerosol, or clear for each vertical bin at a 30-m vertical resolution below 8.2-km altitude and 60-m resolution above 8.2 km. The confidence level of the feature classification is provided as a cloud aerosol discrimination (CAD) score (Liu et al. 2009, 2019), where positive and negative values are for cloud and aerosol layers, respectively. The VFM profiles are provided for every single lidar shot (every 333 m along the track).
CPR aboard CloudSat (Stephens et al. 2002, 2008) measures 94-GHz radar reflectivity profiles at 125 layers at a 480-m vertical resolution with oversampling every 240 m. The CloudSat level-2B geometric profile product (2B-GEOPROF) provides cloud mask profiles, where the value of 0 means completely clear, and the value of 40 means completely cloudy (Marchand et al. 2008; Marchand and Mace 2018). The lowest five vertical CPR bins (up to 1.2-km altitude) are affected by the surface, and this makes the detection of near-surface clouds difficult due to the surface clutter (Marchand and Mace 2018; appendix A). In addition, precipitation significantly increases CloudSat radar signals, and these are flagged as clouds. However, its radiative impact on a visible channel is less noticeable and not detected by MODIS (Berry et al. 2020).
2) MODIS clouds from level-2 CERES Aqua SSF Edition 4 product
MODIS (Salomonson et al. 1989; Barnes et al. 1998) is a passive sensor with 36 channels covering from visible to infrared spectra. We use MODIS-derived cloud properties in the level-2 Clouds and the Earth’s Radiant Energy System (CERES) Aqua Single Scatter Footprint (SSF) Edition 4 product. In the SSF algorithm, MODIS cloud properties are retrieved at a MODIS 1-km pixel resolution, but these are averaged and stored up to two cloud types (i.e., upper and lower clouds) for every CERES footprint where they are assigned (Minnis et al. 2010, 2011a,b, 2021; Loeb et al. 2003). As a result, the MODIS cloud properties collocated with CERES footprints are only analyzed in this study. Note that the MODIS cloud properties in the CERES SSF product are different from MOD06/MYD06 cloud product (Platnick et al. 2017). While cloud properties in the two products generally agree well except at cloud edges or polar regions (Stubenrauch et al. 2013; Minnis et al. 2016; Chiriaco et al. 2007), cloud responses estimated from these two MODIS products can be significantly different (Yue et al. 2017).
CALIPSO, CloudSat, and MODIS aboard Aqua in A-Train observe Earth with an equator crossing time near 0130 (descending) and 1330 (ascending) local time. We use only daytime (1330 LT) observations (section 3c). Therefore, this study does not consider diurnal cloud variations in examining long-term macrophysical cloud changes.
b. Reanalysis data
We use Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2; Bosilovich et al. 2015; Gelaro et al. 2017), for analyzing meteorological conditions such as temperature, humidity, and vertical pressure velocity ω. When we relate the observed cloud properties with reanalysis meteorological conditions, the relation might be affected by the choice of reanalysis datasets. In appendix C, it is shown that ERA5 monthly anomalies from ERA5 are very close to MERRA-2 anomalies.
For relating the meteorological conditions with low clouds, the estimated inversion strength (EIS; Wood and Bretherton 2006) is computed using MERRA-2 temperature and humidity. Wood and Bretherton (2006) suggested that EIS explains the low cloud variability better than do the traditionally used lower tropospheric stability (LTS) (Klein and Hartmann 1993). However, in terms of anomalies, EIS and LTS are very similar (Fig. S1 in the online supplemental material). Therefore, we only use EIS for analyzing the impact of inversion on the low clouds.
3. Method
a. Combining CALIPSO and CloudSat cloud boundaries
Currently, two operational datasets provide CALIPSO–CloudSat merged cloud boundaries. The first is CALIPSO–CloudSat–CERES–MODIS (CCCM) release-B product (Kato et al. 2010), and the second is GEOPROF-lidar product (Mace et al. 2007, 2009; Mace and Zhang 2014). The CCCM product is only available for 4 years (2007–10), and it has a limitation in analyzing long-term anomalies. The GEOPROF-lidar is available for the entire CloudSat mission, but very thin (<120 m) boundary cloud layers can be missed because a high vertical resolution of CALIPSO cloud masks is regridded into CloudSat 240-m vertical bins (Ham et al. 2017). This does not happen in the CCCM algorithm since CALIPSO cloud boundaries are used without a regridding process. Meanwhile, the GEOPROF-lidar algorithm uses a higher threshold of confidence (CAD ≥ 70) for CALIPSO-detected clouds than the CCCM algorithm (CAD > 0). The higher CAD threshold was shown to remove uncertain thin cloud layers over the tropical ocean (Ham et al. 2017).
Therefore, in this study, we adopt the merging strategy from CCCM RelB algorithm (Kato et al. 2010) but the threshold for CAD score is increased to 70, as in the 2B-GEOPROF-lidar algorithm. Besides the threshold of CAD score, another difference from Kato et al. (2010) is that water clouds below 4 km detected by CALIPSO nonsingle lidar shot observations are excluded in this study because the strong lidar signal by liquid-phase particle generates false cloud detections during the spatial averaging process (D. Winker 2017, personal communication).
The first step of the merging process is collocating the nearest CloudSat pixel to the single-shot CALIPSO profile. The single-shot CALIOP profile is provided every 333 m along the track, while the horizontal resolution of CloudSat footprint is 1.9 km along the track and 1.4 km across the track. The closest CloudSat profile is collocated to each CALIPSO single-shot profile. The distance limit between CALIPSO and CloudSat profiles is set as 20 km, and the time difference is permitted up to 20 min.
Once the collocation is performed, cloud masks from CALIPSO and CloudSat are merged. CALIPSO clouds with CAD ≥ 70 and CloudSat clouds with the 2B-GEOPROF cloud mask value ≥ 30 are used. The threshold of 30 for the 2B-GEOPROF cloud mask gives 2.8% of false detections (Marchand and Mace 2018). Because the vertical resolution of CALIPSO lidar (30 m or 60 m) is finer than the resolution of CloudSat radar (480 m but oversampled every 240 m), cloud boundaries from CALIPSO are used as a primary source if they are available (Kato et al. 2010). Below the CALIPSO lidar attenuation level, in cases of optically thick clouds, CloudSat cloud boundaries are used. Note that if CloudSat and CALIPSO (CALCS) observations were not merged, CloudSat misses a significant fraction of clouds, particularly below 3-km and above 8-km altitudes (appendix A; Berry et al. 2020).
b. Computing cloud area and cloud volume
We use two parameters to compute cloud amounts from the satellite measurements. The first parameter is a cloud area fraction, hereinafter simply referred to as a cloud area (omitting “fraction”), which is defined as a fraction of two-dimensional (in the horizontal plane) cloud area. This parameter does not distinguish between geometrically thin or thick clouds when the area is the same. The second parameter is a three-dimensional cloud volume fraction for the given vertical bin, hereinafter simply referred to as a cloud volume. We consider 125 vertical bins from 0- to 20-km altitude with a 0.16-km depth to compute the three-dimensional volume coverage of the cloud layer.
The MODIS cloud areas and volumes are computed using the MODIS cloud-base and cloud-top heights from the level-2 SSF product. As mentioned in section 2a, MODIS cloud properties are retrieved at a 1-km MODIS pixel resolution, but these are averaged up to two cloud types in a CERES footprint. Therefore, we compute the cloud area and volume profile for each cloud type (upper or lower cloud layers) and then compute an area-weighted mean of them for every CERES footprint. Note that MODIS cloud effective height is first retrieved using a single-layer assumption. Then a cloud layer thickness is assumed depending on a cloud type to compute cloud-base and cloud-top heights (Minnis et al. 2011a). Yost et al. (2021) showed that the MODIS layer thickness tends to be underestimated, and the biases are larger for optically thinner ice clouds. The MODIS cloud boundary uncertainty is also significant for multilayer clouds since MODIS tends to detect a radiative center of these layers. In some cases, lower cloud layers are blocked by upper cloud layers, and the low cloud amounts are underestimated. The overlapping cloud effect in MODIS is discussed in appendix B.
The CALCS cloud areas and volumes are computed from the merged cloud boundaries at every CALIPSO–CloudSat collocated pixels (section 3a). Multiple cloud layers can be accurately detected by CALCS active sensors unless CALIPSO and CloudSat signals are fully attenuated.
After MODIS cloud areas and volumes are computed at a CERES footprint resolution, and CALCS cloud areas and volumes are computed at every CALIPSO–CloudSat collocated pixels, these are monthly averaged with a 1° interval. The monthly 1° gridded data are used for computing monthly anomalies for various domains. When we demonstrate geographical distributions of anomalies of cloud areas or volumes, larger spatial noises are shown in CALCS due to the narrow sampling along the track (appendix A). Therefore, we show smoothed anomalies with 10° gridded data.
Throughout the study, we use the terms low, mid-, and high clouds. The low, mid-, and high clouds are defined based on the range of cloud-top height (CTH), that is, 0–3, 3–10, and 10–18 km, respectively. Note that the low clouds and the 0–3-km clouds defined in this study are different (Fig. 1). The 0–3-km clouds are defined as all cloud layers present between 0 and 3 km, regardless of the CTH. The 0–3-km clouds, therefore, include bottom parts of mid- and high clouds with the cloud-base height (CBH) below 3 km. With this definition, the high clouds and 10–18-km clouds are identical if there are no clouds above the 18-km altitude, which is a reasonable assumption for most regions.
A schematic diagram of low, mid-, high, 0–3-km, 3–10-km, and 10–18-km clouds defined in this study. Low, mid-, and high clouds refer to the cloud layers with their CTHs at 0–3, 3–10, and 10–18 km, respectively. The 0–3-, 3–10-, and 10–18-km clouds refer to any cloud layers presenting at 0–3, 3–10, and 10–18 km regardless of CTHs, respectively.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
A schematic diagram of low, mid-, high, 0–3-km, 3–10-km, and 10–18-km clouds defined in this study. Low, mid-, and high clouds refer to the cloud layers with their CTHs at 0–3, 3–10, and 10–18 km, respectively. The 0–3-, 3–10-, and 10–18-km clouds refer to any cloud layers presenting at 0–3, 3–10, and 10–18 km regardless of CTHs, respectively.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
A schematic diagram of low, mid-, high, 0–3-km, 3–10-km, and 10–18-km clouds defined in this study. Low, mid-, and high clouds refer to the cloud layers with their CTHs at 0–3, 3–10, and 10–18 km, respectively. The 0–3-, 3–10-, and 10–18-km clouds refer to any cloud layers presenting at 0–3, 3–10, and 10–18 km regardless of CTHs, respectively.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
c. Methods for mitigating operational changes in CALIPSO and CloudSat
Since the launch of CloudSat in April 2006, there have been several operational events (https://cloudsat.atmos.colostate.edu/news/CloudSat_status). One of the major CloudSat events was a battery anomaly that occurred in April 2011, disabling the CloudSat operation for the remainder of 2011. In November 2011, the CloudSat power was restored, and Daylight-Only operations (DO-Op) started in June 2012 (Nayak et al. 2012). In February 2018, CloudSat exited the A-Train and entered the C-Train in May 2018, which is located 16.5 km below the A-Train track altitude (https://atrain.nasa.gov). To mitigate the CloudSat changes and issues, we only use daytime observations (1330 LT) from MODIS, CloudSat, and CALIPSO for the period from 2007 to 2017. The use of daytime clouds also helps to overcome large uncertainties in the nighttime passive-sensor (i.e., MODIS) cloud retrievals when visible channels are not available.
In addition, the sensitivity of CloudSat CPR has been decreasing (Stephens et al. 2018). In appendix D, it is shown that the impact of CPR sensor degradation on the cloud area anomalies is relatively small in comparison with the impact of the ENSO events if we focus on regional anomalies such as the northeastern (NE) Pacific. However, if we consider a larger domain such as 60°S–60°N, the decreasing trend of the cloud area is more obvious, particularly in CloudSat only observation. However, the decreasing trend is much less noticeable in the CALCS cloud area anomaly time series, suggesting that CALIPSO supplements the loss of CloudSat sensitivity over time.
There have been operational changes in CALIOP, but the impact of these factors on cloud detection seems to be smaller than that of CPR. Specifically, CALIOP consists of two redundant laser transmitters, and the switch of the laser operations caused a discontinuity in the laser pulse energy. For example, when the second laser replaced the first laser in March 2009, the laser pulse energy suddenly increased. The sudden increase in laser energy has a negligible effect on the normalized attenuated backscatter [Fig. 2b of Stephens et al. (2018)]. In addition, there was a low laser energy issue in CALIPSO after September 2016, but it mostly affected the South Atlantic anomaly region (CALIPSO Team 2018), which is out of the domains considered in this study.
Note that the distance between CloudSat and CALIPSO became longer after the CloudSat battery anomaly in 2011 (Mace and Zhang 2014). We do not find any noticeable differences in cloud area or volume anomaly time series over the Pacific Ocean before and after 2011, but further research is desired to quantify the effects of the distance changes between CALIPSO and CloudSat.
The merged CALCS clouds are not available for several months during the 2007–17 period when operational failures happened in CALIPSO and/or CloudSat. Except for the time series, we use common months of the datasets, where CALIPSO, CloudSat, and MODIS are available, for consistent samplings and comparisons.
4. Mid- and high cloud changes related to the RH
In this section, we examine the cloud changes related to relative humidity (RH). Figure 2 shows cross-sectional images of cloud volumes and their anomalies over the NE Pacific domain (150°–120°W, 0°–30°N). Comparing Figs. 2a and 2c, the variability of MODIS CTH is much smaller than that of CALCS. This is mainly because of the different sensitivity of CALIPSO and MODIS sensors to optically thin cirrus composed of small ice particles. CALIPSO is highly sensitive to those thin layers, but MODIS often misses them because of small extinction coefficients (appendix A). In addition, MODIS detects a radiative center (cloud effective height) below the actual CTH, and then CTH is inferred with the assumption of the cloud layer thickness (section 3b). The underestimated MODIS CTH, therefore, implies the underestimated MODIS cloud layer thickness. In the case of multilayer clouds, MODIS cloud effective heights are retrieved somewhere between lower and upper cloud layers, which also contributes to the biases in MODIS CTHs.
(top) MODIS and (bottom) CALCS (a),(c) monthly cloud volume profiles and (b),(d) their deseasonalized monthly anomalies over the NE Pacific domain (150°–120°W, 0°–30°N). To relate the cloud volume changes with ENSO events, multivariate ENSO index is provided at the top of the figure. Gray areas in (c) and (d) are data missing because of operational issues in CALIPSO and/or CloudSat.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(top) MODIS and (bottom) CALCS (a),(c) monthly cloud volume profiles and (b),(d) their deseasonalized monthly anomalies over the NE Pacific domain (150°–120°W, 0°–30°N). To relate the cloud volume changes with ENSO events, multivariate ENSO index is provided at the top of the figure. Gray areas in (c) and (d) are data missing because of operational issues in CALIPSO and/or CloudSat.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(top) MODIS and (bottom) CALCS (a),(c) monthly cloud volume profiles and (b),(d) their deseasonalized monthly anomalies over the NE Pacific domain (150°–120°W, 0°–30°N). To relate the cloud volume changes with ENSO events, multivariate ENSO index is provided at the top of the figure. Gray areas in (c) and (d) are data missing because of operational issues in CALIPSO and/or CloudSat.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Despite the difference in the cloud-top boundaries, general features of the cloud volume changes during ENSO events are comparable between MODIS (Fig. 2b) and CALCS (Fig. 2d). For example, from mid-2010 to the end of 2010, when a strong La Niña event happened, the cloud volume below 3 km increased, and the clouds volume above 3 km decreased. During the 2015/16 El Niño, MODIS and CALCS cloud volumes in the mid- and upper troposphere are largely increased. The exception is shown for low cloud anomalies during the 2015/16 El Niño, where negative low cloud MODIS anomalies were larger than CALCS. This will be discussed in section 5.
MERRA-2 meteorological conditions in Fig. 3 are consistent with cloud volume anomalies in Fig. 2. Figure 3a shows that negative ω anomalies occurred during the 2015/16 El Niño, strengthening the convergence of moisture and increasing the specific humidity in the troposphere (Fig. 3b). While the strong ω and specific humidity anomalies happened in the first half of the 2015/16 El Niño (May–October 2015), the positive midtropospheric temperature anomalies happened in the second half of the 2015/16 El Niño period (November 2015–April 2016) (Fig. 3c). This is related to the lagged response of the tropical tropospheric temperature to ENSO sea surface temperature (SST) forcing by 1–2 seasons, noted in earlier studies (Angell 1981; Trenberth et al. 2002; Fernández et al. 2004; Kumar and Hoerling 2003; Su et al. 2005). Figure 3c also shows that temperature anomalies near the tropopause (16–18 km) are anticorrelated with temperature anomalies at 5–12 km, which was interpreted as a signature of equatorial waves related to SST anomalies (Randel et al. 2000; Fernández et al. 2004; Trenberth and Smith 2006; Xie et al. 2012). The combination of specific humidity (Fig. 3b) and temperature (Fig. 3c) anomalies results in the RH anomalies in Fig. 3d. Specifically, many arch-shaped RH anomalies are shown in Fig. 3d. When the ERA5 reanalysis dataset is used to obtain anomalies of the meteorological parameters, similar features are found to those of MERRA-2 (appendix C).
MERRA-2 monthly anomalies of (a) vertical pressure velocity (Pa s−1), (b) water vapor (WV) specific humidity (g kg−1), (c) temperature (K), and (d) RH (%) over the NE Pacific domain.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
MERRA-2 monthly anomalies of (a) vertical pressure velocity (Pa s−1), (b) water vapor (WV) specific humidity (g kg−1), (c) temperature (K), and (d) RH (%) over the NE Pacific domain.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
MERRA-2 monthly anomalies of (a) vertical pressure velocity (Pa s−1), (b) water vapor (WV) specific humidity (g kg−1), (c) temperature (K), and (d) RH (%) over the NE Pacific domain.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
The RH anomalies in Fig. 3d quite resemble the cloud volume anomalies in Figs. 2b and 2d. Particularly, CALCS cloud volume anomalies (Fig. 2d) are better correlated with the RH anomalies (Fig. 3d) than are those of MODIS (Fig. 2b). The arch-shaped cloud volume anomalies are only captured in CALCS (Fig. 2d). The close correlation between RH and cloud amounts is somewhat consistent with Chepfer et al. (2019), in which cloud fraction profiles from Cloud-Aerosol Transport System (CATS) spaceborne lidar sensor were related with RH variations. However, Chepfer et al. (2019) examined diurnal variations of the cloud fractions, while this study considers monthly cloud anomalies over the 11-yr period.
As implied in Figs. 2 and 3, the probability of the cloud formation should increase with RH (e.g., MERRA-2 gridded RH), for which a similar principle had been often used in model cloud schemes (e.g., Sundqvist 1988; Tiedtke 1993; Teixeira 2001). Figure 4 illustrates the composites of the cloud volume as a function of the MERRA-2 RH at each altitude using a monthly 5° gridded dataset over the NE Pacific (150°–120°W, 0°–30°N) domain. The cloud volume profiles are from MODIS, CALCS, or MERRA-2.
Composites of (a) MODIS, (b) CALCS, and (c) MERRA-2 cloud volumes (%) as a function of altitude and relative humidity (%). The interval of the relative humidity is 10% between 0% and 100%. Vertical bins are defined from 0 to 20 km with a 0.16-km interval. MERRA-2 relative humidity is used for (a), (b), and (c). (d) The occurrences of the MERRA-2 relative humidity. The cloud volume composites are obtained from the monthly 5° gridded datasets over the NE Pacific domain. For consistent sampling, only months from 2007 to 2017 that have MODIS, CALCS, and MERRA-2 data are used.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Composites of (a) MODIS, (b) CALCS, and (c) MERRA-2 cloud volumes (%) as a function of altitude and relative humidity (%). The interval of the relative humidity is 10% between 0% and 100%. Vertical bins are defined from 0 to 20 km with a 0.16-km interval. MERRA-2 relative humidity is used for (a), (b), and (c). (d) The occurrences of the MERRA-2 relative humidity. The cloud volume composites are obtained from the monthly 5° gridded datasets over the NE Pacific domain. For consistent sampling, only months from 2007 to 2017 that have MODIS, CALCS, and MERRA-2 data are used.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Composites of (a) MODIS, (b) CALCS, and (c) MERRA-2 cloud volumes (%) as a function of altitude and relative humidity (%). The interval of the relative humidity is 10% between 0% and 100%. Vertical bins are defined from 0 to 20 km with a 0.16-km interval. MERRA-2 relative humidity is used for (a), (b), and (c). (d) The occurrences of the MERRA-2 relative humidity. The cloud volume composites are obtained from the monthly 5° gridded datasets over the NE Pacific domain. For consistent sampling, only months from 2007 to 2017 that have MODIS, CALCS, and MERRA-2 data are used.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Generally, all cloud datasets show that the cloud volume increases with the RH. However, MODIS cloud volume does not increase with RH above the 13-km altitude (Fig. 4a) because MODIS clouds above 13 km are underestimated as shown in Fig. 2. Meanwhile, when the RH between 50% and 70% at 4–10-km altitudes, MODIS cloud volumes (Fig. 4a) are larger than CALCS cloud volumes (Fig. 4b). This also implies that the MODIS cloud volumes at these 4–10-km altitudes are overestimated due to the low-biased CTHs for thin cirrus clouds. Below the 3-km altitude, RH is often greater than 80% (Fig. 4d). In this altitude range, CALCS cloud volume increases with RH (Fig. 4b), while MODIS cloud volume does not strictly increase with RH. Specifically, MODIS cloud volume is larger than 20% for the RH between 20% and 80%, and MODIS cloud volume is smaller than 20% for RH > 80% (Fig. 4a). The weaker correlation between RH and MODIS cloud volume below 3 km indicates that the MODIS cloud altitude might be misplaced in this altitude range. CALCS shows increasing cloud volumes with the RH for the wider range of altitude (0–16 km) than MODIS (3–13 km), and the relationship is less dependent on the altitude. This confirms more accurate cloud boundaries from active sensors than passive sensors.
In Fig. 4c, when MERRA-2 cloud volumes are used, the dependence of the MERRA-2 cloud volume on the RH is weaker than the relationships obtained from the observed cloud volumes (Figs. 4a,b). MERRA-2 cloud volumes below 8 km are substantially smaller than the observed cloud volumes, implying underestimation of both mid- and low cloud amounts in MERRA-2. These indicate that cloud volume and RH in the same dataset do not give a better correlation. When using ERA5, a similar conclusion is obtained; ERA5 cloud amounts below 10 km seem to be underestimated and the correlation is not better than those from observed cloud volumes (Fig. S2 in the online supplemental material). This indicates that there is room for improving the cloud parameterization scheme in the reanalysis datasets, which is left for future investigation.
When different domains such as the western Pacific (110°–140°E, 15°S–15°N) (Fig. S3 in the online supplemental material) or the tropics (30°S–30°N) (Fig. S4 in the online supplemental material) are considered, the frequency distributions of the RH differ between domains; the western Pacific has a higher frequency of a large RH than over the NE pacific. Despite the different RH distributions, similar relationships between the cloud volume and RH are noted, indicating that the relationship is less affected by the domain.
The close relationship between cloud amounts and RHs also appears when using the 3–10-km cloud area anomalies, where 3–10-km cloud area is defined as the area of cloud layers present between 3 and 10 km (Fig. 1). Each point in Fig. 5 is obtained with 3-month-running means over the NE Pacific domain. The 3–10-km cloud area anomalies are positively correlated with 500-hPa RH (RH500) anomalies. They are also negatively correlated with 500-hPa ω (ω500) anomalies because the negative ω500 anomalies are associated with a stronger convergence of moisture, increasing RH500.
Anomalies of 3–10-km cloud area as a function of MERRA-2 (left) 500-hPa relative humidity (RH500) and (right) 500-hPa pressure velocity (ω500) anomalies. The 3–10-km cloud area is defined as the area of all clouds present between 3 and 10 km (Fig. 1), obtained from (a),(b) MODIS, (c),(d) CALCS, and (e),(f) MERRA-2. Each point is obtained from 3-month-running-mean anomalies over the NE Pacific domain.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Anomalies of 3–10-km cloud area as a function of MERRA-2 (left) 500-hPa relative humidity (RH500) and (right) 500-hPa pressure velocity (ω500) anomalies. The 3–10-km cloud area is defined as the area of all clouds present between 3 and 10 km (Fig. 1), obtained from (a),(b) MODIS, (c),(d) CALCS, and (e),(f) MERRA-2. Each point is obtained from 3-month-running-mean anomalies over the NE Pacific domain.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Anomalies of 3–10-km cloud area as a function of MERRA-2 (left) 500-hPa relative humidity (RH500) and (right) 500-hPa pressure velocity (ω500) anomalies. The 3–10-km cloud area is defined as the area of all clouds present between 3 and 10 km (Fig. 1), obtained from (a),(b) MODIS, (c),(d) CALCS, and (e),(f) MERRA-2. Each point is obtained from 3-month-running-mean anomalies over the NE Pacific domain.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
A similar analysis is performed including other domains such as the western, central, and southeastern (SE) Pacific, and the results are summarized in Fig. 6. To quantify the impacts of RH and vertical motion velocity, multivariate linear regression is performed using the equation ΔC = a1ΔRH500 + a2Δω500 + ε is used, where ΔC, ΔRH500, and Δω500 are the 3-month running means of 3–10-km cloud area, RH500, ω500 anomalies, respectively. The positive a1 [=(∂C/∂RH500)ω500] and negative a2 [=(∂C/∂ω500)RH500] values are consistent with Fig. 5. Relative to a2, the value of a1 (the regression slope over RH500) is more consistent among the CALCS, MODIS, and MERRA-2 cloud datasets. Slightly larger correlation coefficients between ∆C and ∆RH500 or between ∆C and Δω500 are found over the deep convective regions such as the western Pacific, relative to eastern or central Pacific. This also results in larger coefficients of determination (Fig. 6c) of the regression model over the western Pacific.
(a) Regression slopes (a1 and a2) from the multivariate linear regression using the equation ΔC = a1ΔRH500 + a2Δω500 + ε, where ΔC, ΔRH500, and Δω500 are 3-month-running-mean anomalies of 3–10-km cloud area, RH500, and ω500, respectively. (d) The regression is performed over the western (here WP; 110°–140°E, 15°S–15°N), central (here CP; 180°–150°W, 15°S–15°N), northeastern (here NEP; 150°–120°W, 0°–30°N), and southeastern (here SEP; 110°–70°W, 30°–10°S) Pacific domains. The 3–10-km cloud area anomalies come from CALCS (red), MODIS (blue), or MERRA-2 (green) datasets. Vertical bars in a1 or a2 represent standard errors in the regression results. (b) Correlation coefficients between ΔRH500 and ΔC or Δω500 and ΔC, and (c) coefficient of determinant R2 of the multivariate linear regression model. The value of R2 is defined as a ratio of the sum of squares regression to the sum of squares total, explaining how much of variability of ΔC is explained by the multivariate linear regression model.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(a) Regression slopes (a1 and a2) from the multivariate linear regression using the equation ΔC = a1ΔRH500 + a2Δω500 + ε, where ΔC, ΔRH500, and Δω500 are 3-month-running-mean anomalies of 3–10-km cloud area, RH500, and ω500, respectively. (d) The regression is performed over the western (here WP; 110°–140°E, 15°S–15°N), central (here CP; 180°–150°W, 15°S–15°N), northeastern (here NEP; 150°–120°W, 0°–30°N), and southeastern (here SEP; 110°–70°W, 30°–10°S) Pacific domains. The 3–10-km cloud area anomalies come from CALCS (red), MODIS (blue), or MERRA-2 (green) datasets. Vertical bars in a1 or a2 represent standard errors in the regression results. (b) Correlation coefficients between ΔRH500 and ΔC or Δω500 and ΔC, and (c) coefficient of determinant R2 of the multivariate linear regression model. The value of R2 is defined as a ratio of the sum of squares regression to the sum of squares total, explaining how much of variability of ΔC is explained by the multivariate linear regression model.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(a) Regression slopes (a1 and a2) from the multivariate linear regression using the equation ΔC = a1ΔRH500 + a2Δω500 + ε, where ΔC, ΔRH500, and Δω500 are 3-month-running-mean anomalies of 3–10-km cloud area, RH500, and ω500, respectively. (d) The regression is performed over the western (here WP; 110°–140°E, 15°S–15°N), central (here CP; 180°–150°W, 15°S–15°N), northeastern (here NEP; 150°–120°W, 0°–30°N), and southeastern (here SEP; 110°–70°W, 30°–10°S) Pacific domains. The 3–10-km cloud area anomalies come from CALCS (red), MODIS (blue), or MERRA-2 (green) datasets. Vertical bars in a1 or a2 represent standard errors in the regression results. (b) Correlation coefficients between ΔRH500 and ΔC or Δω500 and ΔC, and (c) coefficient of determinant R2 of the multivariate linear regression model. The value of R2 is defined as a ratio of the sum of squares regression to the sum of squares total, explaining how much of variability of ΔC is explained by the multivariate linear regression model.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
5. Low cloud changes related to the SST and EIS
In this section, we discuss low cloud changes related to the SST or EIS, and how common and different features are obtained from MODIS and CALCS. In Fig. 7, low cloud (CTH at 0–3 km) area anomalies are obtained for the four periods defined in Table 1. During the 2010 La Niña period, both MODIS and CALCS low cloud anomalies were largely positive over the central and eastern Pacific. In contrast, MODIS and CALCS low cloud anomalies were negative over the central Pacific during El Niño events. In comparing 2009/10 and 2015/16 El Niño events, larger negative low cloud anomalies were noted in 2015/16 El Niño. These are related to two different types of El Niño. The 2009/10 El Niño belongs to the Modoki or central Pacific El Niño type, while 2015/16 El Niño belongs to the canonical or eastern Pacific El Niño type (Ashok et al. 2007; Kug et al. 2009; Yeh et al. 2009; Kao and Yu 2009). The Modoki El Niño type is weaker and the warm SST anomalies are limited to the central Pacific, while the canonical El Niño type involves warm SST anomalies over a broader region and lasts longer. The low cloud changes in the two different types of El Niño are similarly captured by MODIS and CALCS.
Deseasonalized low cloud area anomalies derived from (left) MODIS and (left center) CALCS for the four periods shown in Table 1. The low clouds are defined when their CTH is below 3 km (Fig. 1). Deseasonalized anomalies of MERRA-2 (right center) EIS (K) and (right) skin temperature (K). The anomalies are computed with a 10°-grid resolution. Note that skin temperature is identical to SST over the ocean.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Deseasonalized low cloud area anomalies derived from (left) MODIS and (left center) CALCS for the four periods shown in Table 1. The low clouds are defined when their CTH is below 3 km (Fig. 1). Deseasonalized anomalies of MERRA-2 (right center) EIS (K) and (right) skin temperature (K). The anomalies are computed with a 10°-grid resolution. Note that skin temperature is identical to SST over the ocean.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Deseasonalized low cloud area anomalies derived from (left) MODIS and (left center) CALCS for the four periods shown in Table 1. The low clouds are defined when their CTH is below 3 km (Fig. 1). Deseasonalized anomalies of MERRA-2 (right center) EIS (K) and (right) skin temperature (K). The anomalies are computed with a 10°-grid resolution. Note that skin temperature is identical to SST over the ocean.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Four periods of interest related to ENSO events. Since the 2015/16 El Niño lasted longer than the 2009/10 El Niño, the 2015/16 El Niño is divided into two periods to examine evolution of the El Niño. Multivariate ENSO index (MEI) for each period is also provided.
MODIS and CALCS low cloud anomalies are well explained by MERRA-2 EIS or SST anomalies in Fig. 7. Specifically, the low cloud area anomalies are positively correlated with EIS and negatively correlated with SST anomalies, well understood from earlier studies (Klein and Hartmann 1993; Wood and Bretherton 2006; Bond et al. 2015; Hartmann 2015; Di Lorenzo and Mantua 2016; Tseng et al. 2017).
Since low cloud area anomalies increase with EIS and decrease with SST anomalies, impacts of SST and EIS are competing if the same signs of SST and EIS anomalies happen. These are examined in Fig. 8 using scatterplots of 3-month means of anomalies over the NE pacific. As seen in the third column of Fig. 8, EIS and SST anomalies are generally negatively correlated, and thus they are mostly in opposite signs. The exceptions are appeared in the first quadrant (upper-right section) in the third column of Fig. 8, meaning that positive EIS and SST anomalies happened coincidently. This occurs when the midtroposphere warms more than the low troposphere (ΔT700 > ΔT1000 > 0). In this case, the low cloud area anomalies are close to zero (green color). This implies that when SST anomalies are positive, low cloud amounts do not increase with the positive EIS anomalies, which are commonly shown in MODIS, CALCS, and MERRA-2. In other words, when the warm dry air aloft is dominant over the strong inversion, the low cloud amounts do not increase, suggesting further investigations for the cloud parameterization. The results are also consistent with earlier studies showing strong cancellation of EIS and SST on low clouds (Myers and Norris 2016; Klein et al. 2017).
Low cloud area anomalies as a function of MERRA-2 (left) SST anomalies and (center) EIS anomalies over the NE Pacific. All data points are computed with 3-month-running-mean anomalies. Red and blue dots in the left column are for, respectively, EIS anomalies < 0 and ≥ 0 K. Red and blue dots in the center column are for, respectively, SST anomalies < 0 K and ≥ 0 K. (right) Low cloud area anomalies are plotted as a function of EIS and SST anomalies. The low cloud area anomalies are computed with (top) MODIS, (middle) CALCS, and (bottom) MERRA-2 datasets. The low cloud area is defined as the area of all clouds with CTHs below 3 km.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Low cloud area anomalies as a function of MERRA-2 (left) SST anomalies and (center) EIS anomalies over the NE Pacific. All data points are computed with 3-month-running-mean anomalies. Red and blue dots in the left column are for, respectively, EIS anomalies < 0 and ≥ 0 K. Red and blue dots in the center column are for, respectively, SST anomalies < 0 K and ≥ 0 K. (right) Low cloud area anomalies are plotted as a function of EIS and SST anomalies. The low cloud area anomalies are computed with (top) MODIS, (middle) CALCS, and (bottom) MERRA-2 datasets. The low cloud area is defined as the area of all clouds with CTHs below 3 km.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Low cloud area anomalies as a function of MERRA-2 (left) SST anomalies and (center) EIS anomalies over the NE Pacific. All data points are computed with 3-month-running-mean anomalies. Red and blue dots in the left column are for, respectively, EIS anomalies < 0 and ≥ 0 K. Red and blue dots in the center column are for, respectively, SST anomalies < 0 K and ≥ 0 K. (right) Low cloud area anomalies are plotted as a function of EIS and SST anomalies. The low cloud area anomalies are computed with (top) MODIS, (middle) CALCS, and (bottom) MERRA-2 datasets. The low cloud area is defined as the area of all clouds with CTHs below 3 km.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
While common features of low cloud response derived from MODIS and CALCS exist, there are also differences. These differences are mostly attributed to two limitations of passive-sensor measurements: 1) detection of overlapping clouds and 2) estimation of cloud layer thicknesses.
First, we discuss the impact of overlapping clouds. When the differences between MODIS and CALCS low cloud area anomalies are obtained (Fig. 9), the differences (left column) are anticorrelated with mid+high (CTH at 3–18 km) cloud area anomalies (middle and right columns). It suggests that MODIS low cloud area anomalies have a larger anticorrelation with mid+high cloud area anomalies. A similar feature also appears in Fig. 2—MODIS low cloud anomalies are negative, and they are anticorrelated with upper mid- and high cloud anomalies during the 2015/16 El Niño (Fig. 2b), while CALCS shows a much weaker correlation (Fig. 2d). The anticorrelation between MODIS low and mid+high clouds is explained by overlapping cloud effects in passive measurements (appendix B). This is because the detection of underlying low clouds is limited when upper overlapping clouds are optically thick. When the MODIS low cloud area is corrected using the random overlap assumption, MODIS and CALCS low cloud anomalies are more consistent (appendix B).
(left) Differences between MODIS and CALCS low cloud (CTHs at 0–3 km) area anomalies in Fig. 7. Area anomalies of mid-/high clouds (CTHs at 3–18 km) are obtained from (center) MODIS and (right) CALCS. Numbers in parentheses include multivariate ENSO index and 80°S–80°N domain averages of low and mid+high cloud area anomalies for each period.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(left) Differences between MODIS and CALCS low cloud (CTHs at 0–3 km) area anomalies in Fig. 7. Area anomalies of mid-/high clouds (CTHs at 3–18 km) are obtained from (center) MODIS and (right) CALCS. Numbers in parentheses include multivariate ENSO index and 80°S–80°N domain averages of low and mid+high cloud area anomalies for each period.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(left) Differences between MODIS and CALCS low cloud (CTHs at 0–3 km) area anomalies in Fig. 7. Area anomalies of mid-/high clouds (CTHs at 3–18 km) are obtained from (center) MODIS and (right) CALCS. Numbers in parentheses include multivariate ENSO index and 80°S–80°N domain averages of low and mid+high cloud area anomalies for each period.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Second, MODIS cloud layer (geometrical) thicknesses of mid- and high clouds are thinner than those from CALCS, inducing smaller impacts of mid/high clouds on the lower troposphere (<3 km). One way to estimate the impact of mid- and high clouds on the lower troposphere is to obtain the differences between the 0–3-km cloud and low cloud area anomalies. The differences are whether bottom parts of mid- and high clouds, residing below 3 km, are included or not (Fig. 1). Figure 10 shows that the inclusion of mid- and high clouds hardly affects MODIS 0–3-km cloud area anomalies (third column). In contrast, the CALCS 0–3-km cloud area anomalies are significantly changed, especially over the central and western Pacific where deep convections occur. As a result, CALCS and MODIS 0–3-km cloud area anomalies are often in opposite signs over the central and western Pacific (first and second columns). Because of the issues related to overlapping clouds and layer thicknesses, MODIS cloud amounts below the 3-km altitude have larger uncertainties over deep convective regions.
Area anomalies of 0–3-km clouds obtained from (left) MODIS and (left center) CALCS for the four periods defined in Table 1. The differences between the 0–3-km cloud and low cloud area anomalies are obtained from (right center) MODIS and (right) CALCS.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Area anomalies of 0–3-km clouds obtained from (left) MODIS and (left center) CALCS for the four periods defined in Table 1. The differences between the 0–3-km cloud and low cloud area anomalies are obtained from (right center) MODIS and (right) CALCS.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Area anomalies of 0–3-km clouds obtained from (left) MODIS and (left center) CALCS for the four periods defined in Table 1. The differences between the 0–3-km cloud and low cloud area anomalies are obtained from (right center) MODIS and (right) CALCS.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Considering surface longwave downward radiation is sensitive to the emission from the lowermost cloud layer (e.g., Kato et al. 2011), surface longwave anomalies could be different depending on MODIS or CALCS clouds being used. For example, when CALCS indicates a decrease of low clouds and the increase of mid+high clouds, low and mid+high cloud changes would be largely canceled in the low troposphere (<3 km). In contrast, impacts of mid+high clouds on the lower troposphere (<3 km) are weaker in MODIS. Furthermore, the increased mid+high clouds reduce detections of underlying low clouds in MODIS (overlapping cloud effects). These two factors would result in a larger decrease of MODIS 0–3-km cloud amounts relative to CALCS. In this case, longwave surface radiation would largely decrease with MODIS clouds, but not with CALCS. Quantification of surface longwave radiation anomalies derived with MODIS and CALCS clouds is left for future studies.
In Fig. 11, the impact of EIS and SST anomalies on low cloud area anomalies are quantified for western, central, NE, and SE Pacific domains. Following the approach of earlier studies (Qu et al. 2014, 2015), the equation ΔC = b1ΔEIS + b2ΔSST + ε is used where ΔC, ΔEIS, and ΔSST are, respectively, low cloud area, EIS, and SST anomalies. The 3-month running anomalies are used for the regression. Since we use low cloud area anomalies, impacts of mid- and high clouds are excluded in this analysis. For MODIS low cloud area, the random overlap correction (appendix B) is also applied, and the results are given as circle symbols. The MODIS results without the random overlap correction are given as x symbols. CALCS regression slopes b1 and b2 (red symbols) fall between two MODIS regression slopes with and without the random overlap correction (blue o and x symbols) over the western, central, and NE Pacific domains. Therefore, the actual overlap is smaller than the random overlap, which is also discussed in appendix B. It is also shown that even though the absolute MERRA-2 low cloud amounts are underestimated (Fig. 4), the regression slopes are comparable with satellite-measured (MODIS or CALCS) regression slopes. For all CALCS, MODIS, and MERRA-2, the regression slopes, b1 and b2, are larger in the eastern Pacific than in the western or central Pacific, with higher correlations coefficients (Fig. 11b) and higher coefficients of determination (Fig. 11c). This is consistent with earlier studies (Scott et al. 2020). Overall magnitudes of regression slopes in Fig. 11 are comparable with earlier studies (Qu et al. 2015; McCoy et al. 2017), while the slight differences are explained by differences in satellite cloud algorithms, domains, the definition of low clouds, and the number of the cloud controlling factors in the regression.
(a) A multivariate linear regression is performed using the equation ΔC = b1ΔEIS + b2ΔSST + ε, where ΔC, ΔEIS, and ΔSST are 3-month-running-mean anomalies of low cloud area, EIS, and SST, respectively. The regression is performed over the western, central, northeastern, and southeastern Pacific domains in Fig. 6d. The low cloud area anomalies come from CALCS (red), MODIS (blue), or MERRA-2 (green) datasets. The results using MODIS low cloud area anomalies with no overlap assumptions are given as crosses. The MODIS results with the random overlap correction are given as dots. Vertical bars in b1 or b2 in (a) represent standard errors in the regression results. (b) Correlation coefficients between ΔEIS and ΔC, and between ΔSST and ΔC. (c) Coefficients of determination R2 of the multivariate regression model are also provided for each cloud dataset.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(a) A multivariate linear regression is performed using the equation ΔC = b1ΔEIS + b2ΔSST + ε, where ΔC, ΔEIS, and ΔSST are 3-month-running-mean anomalies of low cloud area, EIS, and SST, respectively. The regression is performed over the western, central, northeastern, and southeastern Pacific domains in Fig. 6d. The low cloud area anomalies come from CALCS (red), MODIS (blue), or MERRA-2 (green) datasets. The results using MODIS low cloud area anomalies with no overlap assumptions are given as crosses. The MODIS results with the random overlap correction are given as dots. Vertical bars in b1 or b2 in (a) represent standard errors in the regression results. (b) Correlation coefficients between ΔEIS and ΔC, and between ΔSST and ΔC. (c) Coefficients of determination R2 of the multivariate regression model are also provided for each cloud dataset.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(a) A multivariate linear regression is performed using the equation ΔC = b1ΔEIS + b2ΔSST + ε, where ΔC, ΔEIS, and ΔSST are 3-month-running-mean anomalies of low cloud area, EIS, and SST, respectively. The regression is performed over the western, central, northeastern, and southeastern Pacific domains in Fig. 6d. The low cloud area anomalies come from CALCS (red), MODIS (blue), or MERRA-2 (green) datasets. The results using MODIS low cloud area anomalies with no overlap assumptions are given as crosses. The MODIS results with the random overlap correction are given as dots. Vertical bars in b1 or b2 in (a) represent standard errors in the regression results. (b) Correlation coefficients between ΔEIS and ΔC, and between ΔSST and ΔC. (c) Coefficients of determination R2 of the multivariate regression model are also provided for each cloud dataset.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
6. Different timings of cloud-top elevations and cloud area changes
Figure 3 shows the arch-shaped positive RH anomalies in the upper troposphere at the later stage of the 2009/10 and 2015/16 El Niño events. This is coincident with the elevated CTHs of high clouds at the beginnings of 2010 and 2016 (Fig. 2). In this section, we discuss the timings of cloud-top elevations and cloud area changes during ENSO events, and how these are observed by CALCS and MODIS.
In Fig. 12c, time series of 3-month running means of CTH anomalies over the NE Pacific are obtained for high clouds (CTH at 10–18 km). As expected from Fig. 2, the CALCS CTH anomalies of the high clouds are significantly larger than MODIS (Fig. 12c). However, timing-wise, the MODIS and CALCS produce consistent CTH responses to ENSO events. The CTH of the high clouds (Fig. 12c) elevated after 2–3 months after SST peaks (Fig. 12a), where the maximum peaks of SST anomalies during the 2009/10 and 2015/16 El Niño events are displayed as vertical solid lines.
Time series of 3-month-running-mean anomalies of (a) MERRA-2 temperature (K), (b) MERRA-2 EIS (K), (c) CTHs of high clouds, (d) CTHs of low clouds, (e) high cloud area, and (f) low cloud area over the NE Pacific domain. High clouds in (c) and (e) are those with the CTH between 10- and 18-km altitudes. Low clouds in (d) and (f) are those with CTH is below 3-km altitude. Solid vertical lines are local maxima of SST anomalies, and dashed vertical lines are local minima of EIS anomalies during 2009/10 and 2015/16 El Niño events.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Time series of 3-month-running-mean anomalies of (a) MERRA-2 temperature (K), (b) MERRA-2 EIS (K), (c) CTHs of high clouds, (d) CTHs of low clouds, (e) high cloud area, and (f) low cloud area over the NE Pacific domain. High clouds in (c) and (e) are those with the CTH between 10- and 18-km altitudes. Low clouds in (d) and (f) are those with CTH is below 3-km altitude. Solid vertical lines are local maxima of SST anomalies, and dashed vertical lines are local minima of EIS anomalies during 2009/10 and 2015/16 El Niño events.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Time series of 3-month-running-mean anomalies of (a) MERRA-2 temperature (K), (b) MERRA-2 EIS (K), (c) CTHs of high clouds, (d) CTHs of low clouds, (e) high cloud area, and (f) low cloud area over the NE Pacific domain. High clouds in (c) and (e) are those with the CTH between 10- and 18-km altitudes. Low clouds in (d) and (f) are those with CTH is below 3-km altitude. Solid vertical lines are local maxima of SST anomalies, and dashed vertical lines are local minima of EIS anomalies during 2009/10 and 2015/16 El Niño events.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Meanwhile, the magnitudes of low cloud CTH anomalies are more consistent between MODIS and CALCS (Fig. 12d). This is explained by optically thicker low clouds, whose boundaries are better detected by MODIS (Yost et al. 2021). Ahead of the 2015/16 El Niño, both MODIS and CALCS indicated that the CTH of low clouds was elevated. The increase of CTHs of low clouds did not occur during the 2009/10 El Niño, probably due to a weaker El Niño strength relative to the 2015/16 El Niño.
While the timings of the CTH elevations of high clouds (Fig. 12c) and low clouds (Fig. 12d) are different, timings of high and low cloud area changes (Figs. 12e,f) are coincident. Specifically, during the 2015/16 El Niño, high cloud area anomalies increased, and low cloud area anomalies decreased at almost the same time the minimum of EIS anomalies occurred, where these are marked as dashed lines. Throughout the period, low cloud area anomalies (Fig. 12f) are well correlated with EIS anomalies (Fig. 12b), except when warm SST anomalies occurred in 2017. MODIS and CALCS show close agreement in high cloud area anomalies (Fig. 12e), even though the absolute values of them are different (appendix A). In the comparison of low cloud area anomalies (Fig. 12f), MODIS low cloud area anomalies are slightly larger than CALCS. The random overlap correction (appendix B) to MODIS (cyan color) further improves the agreement between MODIS and CALCS.
In summary, we found the coincidence of low and high cloud area anomalies, but CTH elevations of low and high clouds happen with a time lag over the NE Pacific. The lag can be explained by the lagged midtropospheric warming and the corresponding RH changes. However, it is not clear from this study that such a feature is generally shown in El Niño events over the eastern Pacific. Further investigation is desired to generalize our findings.
7. Summary
We use daytime cloud observations from MODIS and CALCS from 2007 through 2017 to examine cloud macrophysical changes over the Pacific. Common cloud features are found from MODIS and CALCS observations associated with meteorological changes during ENSO events. Both observations indicate that mid/high cloud anomalies are well correlated with RH anomalies over the tropics, and low cloud area anomalies are correlated with EIS and SST anomalies over the central and eastern Pacific. EIS and SST anomalies are generally negatively correlated, affecting low cloud area anomalies in the same way. Although positive SST and EIS anomalies rarely occur, when they occur, low cloud area anomalies do not increase with positive EIS anomalies.
While common features of the cloud changes are found from MODIS and CALCS observations, there are also differences between the two observations:
Relative to MODIS, CALCS cloud volume anomalies are correlated better with MERRA-2 RH anomalies for the wider range of altitudes in the troposphere. This confirms more accurate cloud boundary detections from CALCS observations relative to MODIS.
Clouds above 13 km are underestimated and clouds in 3–10 km are overestimated in MODIS observations, relative to CALCS observations. In the later stage of El Niño events, a smaller increase of cloud-top boundaries of high clouds (>10 km) is noted in MODIS observations.
For multilayer clouds, passive MODIS sensor has a limitation in detecting underlying low clouds. As a result, low cloud area anomalies are anticorrelated with mid/high cloud area anomalies. When the random overlap assumption is used to correct MODIS low cloud area anomalies, a better agreement is shown with CALCS.
According to CALCS, mid/high cloud changes also affect cloud amounts in the low troposphere (<3 km) since their CBHs are often below 3 km. As a result, there are partial cancellations between mid/high and low cloud changes below 3 km according to CALCS observations. This feature is missing in MODIS observations.
This study shows that CALCS cloud boundaries can be reliably used for studying cloud macrophysical changes over a decade. It is also noted that, relative to CloudSat, the combination of CloudSat and CALIPSO brings significant benefits in detecting low clouds and mitigating CPR sensor degradation. These underscore the importance of sensor synergy of space-born lidar and radar instruments in the future, such as Atmospheric Lidar (ATLID) and CPR in the EarthCARE mission (Illingworth et al. 2015).
Acknowledgments
This research has been supported by the NASA CERES project and NASA ROSES CloudSat/CALIPSO science team funding to the Langley Research Center.
Data availability statement
The multivariate ENSO index was obtained from the Physical Science Laboratory (https://psl.noaa.gov/enso/mei/). CERES Aqua level-2 SSF data are available in the CERES ordering tool (https://ceres.larc.nasa.gov/data/). CloudSat release-5 datasets are available in the CloudSat data processing center (http://www.cloudsat.cira.colostate.edu). CALIPSO version-4 datasets are available in the Atmospheric Science Data Center (https://asdc.larc.nasa.gov). MERRA-2 datasets were obtained through the NASA Goddard Earth Sciences (GES) Data and Information Services Center (DISC) (https://disc.gsfc.nasa.gov), and ERA5 datasets were obtained through the Copernicus Climate Change Service (C3S) Climate Data Store (https://cds.climate.copernicus.eu).
APPENDIX A
Cloud Areas and Volumes from MODIS, CloudSat, and CALCS
In this section, we compare cloud areas and volumes from MODIS, CloudSat, and CALCS. Note that the spatial coverage of MODIS cloud properties in the SSF product is better than that of CALIPSO and CloudSat observations since the product includes off-nadir MODIS observations from cross-track scans, while CALIPSO and CloudSat only perform along-track scans with a narrow swath. This implies that larger viewing zenith angles are included in the MODIS observations, relative to the CALIPSO and CloudSat observations. It is well known that the satellite-observed cloud area increases with the satellite viewing zenith angle (Minnis 1989). When we selected nadir-view MODIS observations in the SSF product, which is a consistent way to CALCS observations, the global mean of the cloud area is reduced by 3%. However, once the consistent sampling of viewing angles is used over time, the range of the viewing zenith angles should not affect interannual variabilities, which are mainly examined in this study. Therefore, we use all available MODIS observations in the SSF product to reduce sampling noises, with bearing the 3% cloud area global mean difference.
In Fig. A1, zonal distributions of the cloud areas are compared among MODIS, CloudSat, and CALCS. In this figure, CloudSat 0–3-km cloud area is significantly smaller than those from MODIS and CALCS (Fig. A1a), indicating that CloudSat misses low clouds because of the surface clutter (Marchand and Mace 2018). In contrast, MODIS and CALCS 0–3-km areas agree well except over Antarctica.
Zonal distributions of (a) 0–3-km cloud area, (b) 3–10-km cloud area, (c) 10–18-km cloud area, and (d) 0–18-km cloud area, defined in section 3b. One year of data in 2008 is used for computations. The numbers in parentheses are mean values over the 80°S–80°N region.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Zonal distributions of (a) 0–3-km cloud area, (b) 3–10-km cloud area, (c) 10–18-km cloud area, and (d) 0–18-km cloud area, defined in section 3b. One year of data in 2008 is used for computations. The numbers in parentheses are mean values over the 80°S–80°N region.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Zonal distributions of (a) 0–3-km cloud area, (b) 3–10-km cloud area, (c) 10–18-km cloud area, and (d) 0–18-km cloud area, defined in section 3b. One year of data in 2008 is used for computations. The numbers in parentheses are mean values over the 80°S–80°N region.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
In addition, MODIS and CloudSat 10–18-km cloud areas are significantly smaller than CALCS (Fig. A1c). This is because CALIPSO is sensitive to thin cirrus clouds, while MODIS and CloudSat often miss them. Despite different 10–18-km cloud areas, total cloud areas (0–18-km cloud areas) from CALCS and MODIS are more consistent in Fig. A1d. This is because the differences at 3–10 and 10–18 km are compensating. As discussed earlier in this section, if we select nadir MODIS observations in the CERES SSF product, MODIS 0–18-km cloud area would be around 62% (down from 65%) with the nadir view sampling, and it is still closer to CALCS, relative to CloudSat observations.
In Fig. A2, geographical distributions of 0–18-km cloud areas are obtained from MODIS, CloudSat, and CALCS. Because of the smaller sampling numbers in CloudSat and CALCS, larger spatial noises are shown in CloudSat and CALCS cloud areas, relative to MODIS. Despite the different degrees of spatial noises, the 0–18-km cloud areas from MODIS and CALCS generally agree well. In contrast, CloudSat underestimates 0–18-km cloud areas, and the differences from other datasets are 23%–24% globally.
Global distributions of 0–18-km cloud areas from (a) MODIS, (b) CloudSat, and (c) CALCS using 1 yr of datasets in 2008. The numbers in parentheses are mean values over the 80°S–80°N region.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Global distributions of 0–18-km cloud areas from (a) MODIS, (b) CloudSat, and (c) CALCS using 1 yr of datasets in 2008. The numbers in parentheses are mean values over the 80°S–80°N region.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Global distributions of 0–18-km cloud areas from (a) MODIS, (b) CloudSat, and (c) CALCS using 1 yr of datasets in 2008. The numbers in parentheses are mean values over the 80°S–80°N region.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Even though MODIS and CALCS show similar distributions of the 0–18-km cloud areas, as inferred in Fig. A1, if we focus on the cloud areas at a specific range of altitude (e.g., 3–10 or 10–18 km), larger differences appear. For the same reason, cloud volume profiles from MODIS, CloudSat, and CALCS are different (Fig. A3), while a similar comparison was made in Kato et al. (2019). MODIS cloud heights of deep convective clouds and cirrus are lower than those from CALCS over the tropics (30°S–30°N) because CALIPSO is more sensitive to the ice cloud-top boundaries. The differences between MODIS and CALCS cloud volumes have both negative and positive signs (Fig. A3d), and these are largely compensating in computing total cloud areas in Fig. A2.
Zonal cross-sectional cloud volume profiles obtained from (a) MODIS, (b) CloudSat, and (c) CALCS. The differences between (d) MODIS and CALCS and (e) CloudSat and CALCS are also obtained. One year of datasets in 2008 is used.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Zonal cross-sectional cloud volume profiles obtained from (a) MODIS, (b) CloudSat, and (c) CALCS. The differences between (d) MODIS and CALCS and (e) CloudSat and CALCS are also obtained. One year of datasets in 2008 is used.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Zonal cross-sectional cloud volume profiles obtained from (a) MODIS, (b) CloudSat, and (c) CALCS. The differences between (d) MODIS and CALCS and (e) CloudSat and CALCS are also obtained. One year of datasets in 2008 is used.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
CloudSat (Fig. A3b) and CALCS (Fig. A3c) cloud volumes are generally consistent in terms of the cloud altitude, but as inferred in Fig. A1, the overall magnitude in CloudSat volume is smaller than CALCS. Particularly, a large fraction of low cloud volumes below 1 km is missing in CloudSat observations because of the surface clutter (Marchand and Mace 2018). Moreover, high cirrus cloud volumes above 13 km are missing in CloudSat observations, showing negative differences between CloudSat and CALCS for most of altitudes and latitudes (Fig. A3e).
APPENDIX B
Correction of the MODIS Clouds with the Overlap Assumptions
When multilayer clouds exist, CALCS active sensors can detect all layers unless the lidar and radar signals are fully attenuated by cloud particles. In contrast, the passive MODIS sensor retrieves a radiative center of these layers with a single-layer assumption. If upper cloud layers are optically thick enough, the passive sensor detects the upper cloud layers, and the lower cloud amount is underestimated. To correct the issue, earlier studies (Kato et al. 2019; Scott et al. 2020) used the random overlap assumptions. We revisit this approach using MODIS and CALCS in this section.
In Fig. B1, we correct cloud overlap effects in MODIS measurements by deriving the true low cloud area c2 and its anomaly
(left) Low cloud areas and (right) their anomalies over the (top) NE (150°–120°W, 0°–30°N), (top middle) SE (110°–70°W, 30°–10°S), (bottom middle) central (180°–150°W, 15°S–15°N), and (bottom) western (110°–140°E, 15°S–15°N) Pacific domains. MODIS low cloud areas and their anomalies with no overlap correction are indicated by the blue lines. The results with the overlap correction using β = 0.5 and β = 1 in Eq. (B2) are indicated by the cyan and red lines, respectively. CALCS low cloud area and their anomalies are indicated by the black lines.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(left) Low cloud areas and (right) their anomalies over the (top) NE (150°–120°W, 0°–30°N), (top middle) SE (110°–70°W, 30°–10°S), (bottom middle) central (180°–150°W, 15°S–15°N), and (bottom) western (110°–140°E, 15°S–15°N) Pacific domains. MODIS low cloud areas and their anomalies with no overlap correction are indicated by the blue lines. The results with the overlap correction using β = 0.5 and β = 1 in Eq. (B2) are indicated by the cyan and red lines, respectively. CALCS low cloud area and their anomalies are indicated by the black lines.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
(left) Low cloud areas and (right) their anomalies over the (top) NE (150°–120°W, 0°–30°N), (top middle) SE (110°–70°W, 30°–10°S), (bottom middle) central (180°–150°W, 15°S–15°N), and (bottom) western (110°–140°E, 15°S–15°N) Pacific domains. MODIS low cloud areas and their anomalies with no overlap correction are indicated by the blue lines. The results with the overlap correction using β = 0.5 and β = 1 in Eq. (B2) are indicated by the cyan and red lines, respectively. CALCS low cloud area and their anomalies are indicated by the black lines.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
When the MODIS low cloud area anomalies
The overestimate of s2 in Eq. (B2) explains the overestimate of c2 with β > 0 in Fig. B1. In addition, if s2 is overestimated, a larger value of β is needed to produce unbiased
APPENDIX C
Impact of the Reanalysis Datasets
Throughout the study, MERRA-2 meteorological datasets are used to relate to the cloud macrophysical anomalies. The reanalysis dataset is produced by assimilating observational data in the numerical model. As a result, the reanalysis dataset is closer to observations relative to model-only results. However, there still exist differences between observations and reanalysis data, as indicated by earlier comparison studies (e.g., Hearty et al. 2014; Reichle et al. 2017). This implies that the results might be affected by the choice of the reanalysis dataset. We found that the monthly anomaly time series (Fig. C1) from European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric reanalysis of the global climate (ERA5) are consistent with those from MERRA-2 (Fig. 3) over the NE Pacific. The same conclusions are obtained over the western or central Pacific (not shown). The close agreement of monthly anomalies between MERRA-2 and ERA5 indicates that our analysis using monthly anomaly time series is less affected by the choice of the reanalysis datasets.
As in Fig. 3, but from ERA5.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
As in Fig. 3, but from ERA5.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
As in Fig. 3, but from ERA5.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
APPENDIX D
Impact of the Degradation of CloudSat CPR
During the operation of CloudSat, the CPR sensor sensitivity has degraded with time. As a result, the minimum detectable values of the equivalent radar reflectivity factor (ZdB) decreased from −30 to −26 dB [Fig. 2a of Stephens et al. (2018)]. The sensor degradation makes the detection of clouds with weak radar signals more difficult. In other words, more clouds composed of small particles are getting missed over time because the radar reflectivity is proportional to the sixth order of particle size.
This implies that the CPR degradation affects the cloud volume or area anomalies when we use CloudSat observations. More specifically, a larger portion of the cloud layers with weak radar signals (ZdB < −26 dB) are getting missed with time, and thus a decreasing trend is expected. In Fig. D1, anomalies of the 0–18-km cloud area are obtained for two domains. When the anomalies are obtained for the NE Pacific domain, the 0–18-km cloud area anomalies from MODIS, CloudSat, and CALCS are consistent, and the decreasing trend is not noticeable in CloudSat nor CALCS. This is because the 0–18-km cloud area anomalies are mainly driven by ENSO events, and the impact of ENSO events is larger than the impact of CPR degradation. However, when a larger domain (60°S–60°N) is considered to obtain the 0–18-km cloud area anomalies (Fig. D1b), the decreasing trend is more obvious in CloudSat observations (blue lines), as the ENSO signals are smoothed out. In contrast, the 0–18-km cloud area anomalies from MODIS do not show decreasing trends, while CALCS anomalies show a much weaker sign of decreasing trend after mid-2016 relative to CloudSat. Therefore, CALIPSO observations seem to largely supplement the CPR degradation, but the quantification of the degradation and its correction are left for future studies.
Time series of the 3-month-running-mean anomalies of the 0–18-km cloud area over the (a) NE Pacific and (b) 60°S–60°N domains. The 0–18-km cloud areas are obtained from MODIS (black lines), CloudSat (blue lines), and CALCS (red lines).
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Time series of the 3-month-running-mean anomalies of the 0–18-km cloud area over the (a) NE Pacific and (b) 60°S–60°N domains. The 0–18-km cloud areas are obtained from MODIS (black lines), CloudSat (blue lines), and CALCS (red lines).
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
Time series of the 3-month-running-mean anomalies of the 0–18-km cloud area over the (a) NE Pacific and (b) 60°S–60°N domains. The 0–18-km cloud areas are obtained from MODIS (black lines), CloudSat (blue lines), and CALCS (red lines).
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0226.1
REFERENCES
Angell, J. K., 1981: Comparison of variations in atmospheric quantities with sea surface temperature variations in the equatorial eastern Pacific. Mon. Wea. Rev., 109, 230–243, https://doi.org/10.1175/1520-0493(1981)109<0230:COVIAQ>2.0.CO;2.
Ashok, K., S. K. Behera, S. A. Rao, H. Weng, and T. Yamagata, 2007: El Niño Modoki and its possible teleconnection. J. Geophys. Res., 112, C11007, https://doi.org/10.1029/2006JC003798.
Barnes, W. L., T. S. Pagano, and V. V. Salomonson, 1998: Prelaunch characteristics of the Moderate Resolution Imaging Spectroradiometer (MODIS) on EOS-AM1. IEEE Trans. Geosci. Remote Sens., 36, 1088–1100, https://doi.org/10.1109/36.700993.
Berry, E., G. G. Mace, and A. Gettleman, 2020: Using A-Train observations to evaluate east Pacific cloud occurrence and radiative effects in the Community Atmosphere Model. J. Climate, 33, 6187–6203, https://doi.org/10.1175/JCLI-D-19-0870.1.
Bond, N. A., M. F. Cronin, H. Freeland, and N. Mantua, 2015: Causes and impacts of the 2014 warm anomaly in the NE Pacific. Geophys. Res. Lett., 42, 3414–3420, https://doi.org/10.1002/2015GL063306.
Bosilovich, M., and Coauthors, 2015: MERRA-2: Initial evaluation of the climate. NASA Tech. Rep. Series on Global Modelling and Data Assimilation NASA/TM–2015-104606, Vol. 43, 145 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Bosilovich803.pdf.
CALIPSO Team, 2018: CALIPSO low laser energy technical advisory for data users. NASA Doc., 7 pp., https://www-calipso.larc.nasa.gov/resources/calipso_users_guide/advisory/advisory_2018-10-10-CALIPSO_Laser_Energy_Technical_Advisory_Ver03.pdf.
Cesana, G., and H. Chepfer, 2012: How well do climate models simulate cloud vertical structure? A comparison between CALIPSO-GOCCP satellite observations and CMIP5 models. Geophys. Res. Lett., 39, L20803, https://doi.org/10.1029/2012GL053153.
Chepfer, H., H. Brogniez, and V. Noel, 2019: Diurnal variations of cloud and relative humidity profiles across the tropics. Sci. Rep., 9, 16045, https://doi.org/10.1038/s41598-019-52437-6.
Chiriaco, M., and Coauthors, 2007: Comparison of CALIPSO-like, LaRC, and MODIS retrievals of ice-cloud properties over SIRTA in France and Florida during CRYSTAL-FACE. J. Appl. Meteor. Climatol., 46, 249–272, https://doi.org/10.1175/JAM2435.1.
Di Lorenzo, E., and N. Mantua, 2016: Multi-year persistence of the 2014/15 North Pacific marine heatwave. Nat. Climate Change, 6, 1042–1047, https://doi.org/10.1038/nclimate3082.
Fernández, N. C., R. R. García, R. G. Herrera, D. G. Puyol, L. G. Presa, E. H. Martín, and P. R. Rodríguez, 2004: Analysis of the ENSO signal in tropospheric and stratospheric temperatures observed by MSU, 1979–2000. J. Climate, 17, 3934–3946, https://doi.org/10.1175/1520-0442(2004)017<3934:AOTESI>2.0.CO;2.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
Ham, S.-H., B.-J. Sohn, S. Kato, and M. Satoh, 2013: Vertical structure of ice cloud layers from CloudSat and CALIPSO measurements and comparison to NICAM simulations. J. Geophys. Res., 118, 9930–9947, https://doi.org/10.1002/jgrd.50582.
Ham, S.-H., and Coauthors, 2017: Cloud occurrences and cloud radiative effects (CREs) from CERES-CALIPSO-CloudSat-MODIS (CCCM) and CloudSat radar-lidar (RL) products. J. Geophys. Res., 122, 8852–8884, https://doi.org/10.1002/2017JD026725.
Hartmann, D. L., 2015: Pacific sea surface temperature and the winter of 2014. Geophys. Res. Lett., 42, 1894–1902, https://doi.org/10.1002/2015GL063083.
Hearty, T. J., and Coauthors, 2014: Estimating sampling biases and measurement uncertainties of AIRS/AMSU-A temperature and water vapor observations using MERRA reanalysis. J. Geophys. Res., 119, 2725–2741, https://doi.org/10.1002/2013JD021205.
Illingworth, A. J., and Coauthors, 2015: The EarthCARE Satellite: The next step forward in global measurements of clouds, aerosols, precipitation, and radiation. Bull. Amer. Meteor. Soc., 96, 1311–1332, https://doi.org/10.1175/BAMS-D-12-00227.1.
Kao, H.-Y., and J.-Y. Yu, 2009: Contrasting eastern-Pacific and central-Pacific types of ENSO. J. Climate, 22, 615–632, https://doi.org/10.1175/2008JCLI2309.1.
Kato, S., S. Sun-Mack, W. F. Miller, F. G. Rose, Y. Chen, P. Minnis, and B. A. Wielicki, 2010: Relationships among cloud occurrence frequency, overlap, and effective thickness derived from CALIPSO and CloudSat merged cloud vertical profiles. J. Geophys. Res., 115, D00H28, https://doi.org/10.1029/2009JD012277.
Kato, S., and Coauthors, 2011: Improvements of top-of-atmosphere and surface irradiance computations with CALIPSO-, CloudSat-, and MODIS-derived cloud and aerosol properties. J. Geophys. Res., 116, D19209, https://doi.org/10.1029/2011JD016050.
Kato, S., and Coauthors, 2019: Radiative heating rates computed with clouds derived from satellite-based passive and active sensors and their effects on generation of available potential energy. J. Geophys. Res., 124, 1720–1740, https://doi.org/10.1029/2018JD028878.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587–1606, https://doi.org/10.1175/1520-0442(1993)006<1587:TSCOLS>2.0.CO;2.
Klein, S. A., A. Hall, and J. R. Norris, 2017: Low-cloud feedbacks from cloud-controlling factors: A review. Surv. Geophys., 38, 1307–1329, https://doi.org/10.1007/s10712-017-9433-3.
Kug, J.-S., F.-F. Jin, and S.-I. An, 2009: Two types of El Niño events: Cold tongue El Niño and warm pool El Niño. J. Climate, 22, 1499–1515, https://doi.org/10.1175/2008JCLI2624.1.
Kumar, A., and M. P. Hoerling, 2003: The nature and causes for the delayed atmospheric response to El Niño. J. Climate, 16, 1391–1403, https://doi.org/10.1175/1520-0442-16.9.1391.
Li, J., J. Huang, K. Stamnes, T. Wang, Q. Lv, and H. Jin, 2015: A global survey of cloud overlap based on CALIPSO and CloudSat measurements. Atmos. Chem. Phys., 15, 519–536, https://doi.org/10.5194/acp-15-519-2015.
Liu, Z., and Coauthors, 2009: The CALIPSO lidar cloud and aerosol discrimination: Version 2 algorithm and initial assessment of performance. J. Atmos. Oceanic Technol., 26, 1198–1213, https://doi.org/10.1175/2009JTECHA1229.1.
Liu, Z., and Coauthors, 2019: Discriminating between clouds and aerosols in the CALIOP version 4.1 data products. Atmos. Meas. Tech., 12, 703–734, https://doi.org/10.5194/amt-12-703-2019.
Loeb, N. G., N. Manalo-Smith, S. Kato, W. F. Miller, S. K. Gupta, P. Minnis, and B. A. Wielicki, 2003: Angular distribution models for top-of-atmosphere radiative flux estimation from the Clouds and the Earth’s Radiant Energy System instrument on the Tropical Rainfall Measuring Mission satellite. Part I: Methodology. J. Appl. Meteor., 42, 240–265, https://doi.org/10.1175/1520-0450(2003)042<0240:ADMFTO>2.0.CO;2.
Mace, G. G., and F. J. Wrenn, 2013: Evaluation of the hydrometeor layers in the east and west Pacific within ISCCP cloud-top pressure–optical depth bins using merged CloudSat and CALIPSO data. J. Climate, 26, 9429–9444, https://doi.org/10.1175/JCLI-D-12-00207.1.
Mace, G. G., and Q. Zhang, 2014: The CloudSat radar-lidar geometrical profile product (RL-GeoProf): Updates, improvements, and selected results. J. Geophys. Res., 119, 9441–9462, https://doi.org/10.1002/2013JD021374.
Mace, G. G., R. Marchand, Q. Zhang, and G. L. Stephens, 2007: Global hydrometeor occurrence as observed by CloudSat: Initial observations from summer 2006. Geophys. Res. Lett., 34, L09808, https://doi.org/10.1029/2006GL029017.
Mace, G. G., Q. Zhang, M. Vaughan, R. Marchand, G. L. Stephens, C. Trepte, and D. Winker, 2009: A description of hydrometeor layer occurrence statistics derived from the first year of merged CloudSat and CALIPSO data. J. Geophys. Res., 114, D00A26, https://doi.org/10.1029/2007JD009755.
Marchand, R., and G. G. Mace, 2018: Level 2 GEOPROF product process description and interface control document. NASA Doc., 27 pp., http://www.cloudsat.cira.colostate.edu/sites/default/files/products/files/2B-GEOPROF_PDICD.P1_R05.rev0__0.pdf.
Marchand, R., G. G. Mace, T. Ackerman, and G. L. Stephens, 2008: Hydrometeor detection using CloudSat—An Earth-orbiting 94-GHz cloud radar. J. Atmos. Oceanic Technol., 25, 519–533, https://doi.org/10.1175/2007JTECHA1006.1.
McCoy, D. T., R. Eastman, D. L. Hartmann, and R. Wood, 2017: The change in low cloud cover in a warmed climate inferred from AIRS, MODIS, and ERA-Interim. J. Climate, 30, 3609–3620, https://doi.org/10.1175/JCLI-D-15-0734.1.
Minnis, P., 1989: Viewing zenith angle dependence of cloudiness determined from coincident GEOS East and GOES West data. J. Geophys. Res., 94, 2303–2320, https://doi.org/10.1029/JD094iD02p02303.
Minnis, P., and Coauthors, 2010: CERES Edition 3 cloud retrievals. 13th Conf. on Atmospheric Radiation, Portland, OR, Amer. Meteor. Soc., 5.4, https://ams.confex.com/ams/pdfpapers/171366.pdf.
Minnis, P., and Coauthors, 2011a: CERES Edition-2 cloud property retrievals using TRMM VIRS and Terra and Aqua MODIS data—Part I: Algorithms. IEEE Trans. Geosci. Remote Sens., 49, 4374–4400, https://doi.org/10.1109/TGRS.2011.2144601.
Minnis, P., and Coauthors, 2011b: CERES Edition-2 cloud property retrievals using TRMM VIRS and Terra and Aqua MODIS data—Part II: Examples of average results and comparisons with other data. IEEE Trans. Geosci. Remote Sens., 49, 4401–4430, https://doi.org/10.1109/TGRS.2011.2144602.
Minnis, P., K. Bedka, T. Qing, C. R. Yost, S. T. Bedka, B. A. Scarino, K. Khlopenkov, and M. M. Khaiyer, 2016: A consistent long-term cloud and clear-sky radiation property dataset from the Advanced Very High Resolution Radiometer (AVHRR). NOAA Climate Algorithm Theoretical Basis Doc., 159 pp., https://www.ncdc.noaa.gov/sites/default/files/cdr-documentation/CDRP-ATBD-0826%20AVHRR%20Cloud%20Properties%20-%20NASA%20C-ATBD%20(01B-30b)%20(DSR-1051).pdf.
Minnis, P., and Coauthors, 2021: CERES MODIS cloud product retrievals for Edition 4—Part I: Algorithm changes. IEEE Trans. Geosci. Remote Sens., 59, 2744–2780, https://doi.org/10.1109/TGRS.2020.3008866.
Myers, T. A., and J. R. Norris, 2016: Reducing the uncertainty in subtropical cloud feedback. Geophys. Res. Lett., 43, 2144–2148, https://doi.org/10.1002/2015GL067416.
Nayak, M., and Coauthors, 2012: CloudSat anomaly recovery and operational lessons learned. SpaceOps 2012, Stockholm, Sweden, AIAA, https://doi.org/10.2514/6.2012-1295798.
Platnick, S. E., and Coauthors, 2017: The MODIS cloud optical and microphysical products: Collection 6 updates and examples from Terra and Aqua. IEEE Trans. Geosci. Remote Sens., 55, 502–525, https://doi.org/10.1109/TGRS.2016.2610522.
Qu, X., A. Hall, S. A. Klein, and P. M. Caldwell, 2014: On the spread of changes in marine low cloud cover in climate model simulations of the 21st century. Climate Dyn., 42, 2603–2626, https://doi.org/10.1007/s00382-013-1945-z.
Qu, X., A. Hall, S. A. Klein, and A. M. DeAngelis, 2015: Positive tropical marine low-cloud cover feedback inferred from cloud controlling factors. Geophys. Res. Lett., 42, 7767–7775, https://doi.org/10.1002/2015GL065627.
Randel, W. J., F. Wu, and D. J. Gaffen, 2000: Interannual variability of the tropical tropopause derived from radiosonde data and NCEP reanalysis. J. Geophys. Res., 105, 15 509–15 523, https://doi.org/10.1029/2000JD900155.
Reichle, R. H., C. S. Draper, Q. Liu, M. Girotto, S. P. P. Mahanama, R. D. Koster, and G. J. M. de Lannoy, 2017: Assessment of MERRA-2 land surface hydrology estimates. J. Climate, 30, 2937–2960, https://doi.org/10.1175/JCLI-D-16-0720.1.
Rossow, W. B., and Y. Zhang, 2010: Evaluation of a statistical model of cloud vertical structure using combined CloudSat and CALIPSO cloud layer profiles. J. Climate, 23, 6641–6653, https://doi.org/10.1175/2010JCLI3734.1.
Salomonson, V. V., W. L. Barnes, P. W. Maymon, H. E. Montgomery, and H. Ostrow, 1989: MODIS: Advanced facility instrument for studies of the earth as a system. IEEE Trans. Geosci. Remote Sens., 27, 145–153, https://doi.org/10.1109/36.20292.
Scott, R. C., T. A. Myers, J. R. Norris, M. D. Zelinka, S. A. Klein, M. Sun, and D. R. Doelling, 2020: Observed sensitivity of low-cloud radiative effects to meteorological perturbations over the global oceans. J. Climate, 33, 7717–7734, https://doi.org/10.1175/JCLI-D-19-1028.1.
Stein, T. H. M., J. Delanoë, and R. J. Hoan, 2011: Comparison among four different retrieval methods for ice-cloud properties using data from CloudSat, CALIPSO, and MODIS. J. Appl. Meteor. Climatol., 50, 1952–1969, https://doi.org/10.1175/2011JAMC2646.1.
Stephens, G. L., and Coauthors, 2002: The CloudSat mission and the A-Train. Bull. Amer. Meteor. Soc., 83, 1771–1790, https://doi.org/10.1175/BAMS-83-12-1771.
Stephens, G. L., and Coauthors, 2008: CloudSat mission: Performance and early science after the first year of operation. J. Geophys. Res., 113, D00A18, https://doi.org/10.1029/2008JD009982.
Stephens, G. L., D. Winker, J. Pelon, C. Trepte, D. Vane, C. Yuhas, T. L’Ecuyer, and M. Lebsock, 2018: CloudSat and CALIPSO within the A-Train: Ten years of actively observing the Earth system. Bull. Amer. Meteor. Soc., 99, 569–581, https://doi.org/10.1175/BAMS-D-16-0324.1.
Stubenrauch, C. J., and Coauthors, 2013: Assessment of global cloud datasets from satellites: Project and database initiated by the GEWEX Radiation Panel. Bull. Amer. Meteor. Soc., 94, 1031–1049, https://doi.org/10.1175/BAMS-D-12-00117.1.
Su, H., D. Neelin, and J. E. Meyerson, 2005: Mechanisms for lagged atmospheric response to ENSO SST forcing. J. Climate, 18, 4195–4215, https://doi.org/10.1175/JCLI3514.1.
Sundqvist, H., 1988: Parametrization of condensation and associated clouds in models for weather prediction and general circulation simulation. Physically-Based Modelling and Simulation of Climate and Climate Change, M. E. Schlesinger, Ed., Kluwer, 433–461.
Teixeira, J., 2001: Cloud fraction and relative humidity in a prognostic cloud fraction scheme. Mon. Wea. Rev., 129, 1750–1753, https://doi.org/10.1175/1520-0493(2001)129<1750:CFARHI>2.0.CO;2.
Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev., 121, 3040–3061, https://doi.org/10.1175/1520-0493(1993)121<3040:ROCILS>2.0.CO;2.
Trenberth, K. E., and L. Smith, 2006: The vertical structure of temperature in the tropics: Different flavors of El Niño. J. Climate, 19, 4956–4973, https://doi.org/10.1175/JCLI3891.1.
Trenberth, K. E., J. M. Caron, D. P. Stepaniak, and S. Worley, 2002: Evolution of El Niño-Southern Oscillation and global atmospheric surface temperatures. J. Geophys. Res., 107, 4065, https://doi.org/10.1029/2000JD000298.
Tseng, Y.-H., R. Ding, and X. Huang, 2017: The warm blob in the northeast Pacific—The bridge leading to the 2015/16 El Niño. Environ. Res. Lett., 12, 054019, https://doi.org/10.1088/1748-9326/aa67c3.
Vaughan, M., and Coauthors, 2009: Fully automated detection of cloud and aerosol layers in the CALIPSO lidar measurements. J. Atmos. Oceanic Technol., 26, 2034–2050, https://doi.org/10.1175/2009JTECHA1228.1.
Winker, D. M., J. Pelon, and M. P. McCormick, 2003: The CALIPSO mission: Spaceborne lidar for observation of aerosols and clouds. Proc. SPIE, 4893, https://doi.org/10.1117/12.466539.
Winker, D. M., B. H. Hunt, and M. J. McGill, 2007: Initial performance assessment of CALIOP. Geophys. Res. Lett., 34, L19803, https://doi.org/10.1029/2007GL030135.
Winker, D. M., M. A. Vaughan, A. Omar, Y.-X. Hu, K. A. Powell, Z. Liu, W. H. Hunt, and S. A. Young, 2009: Overview of the CALIPSO mission and CALIOP data processing algorithms. J. Atmos. Oceanic Technol., 26, 2310–2323, https://doi.org/10.1175/2009JTECHA1281.1.
Wood, R., and C. S. Bretherton, 2006: On the relationship between stratiform low cloud cover and lower-tropospheric stability. J. Climate, 19, 6425–6432, https://doi.org/10.1175/JCLI3988.1.
Xie, F., J. Li, W. Tian, J. Feng, and Y. Huo, 2012: Signals of El Niño Modoki in the tropical tropopause layer and stratosphere. Atmos. Chem. Phys., 12, 5259–5273, https://doi.org/10.5194/acp-12-5259-2012.