1. Introduction
Cloud feedback is a major source of uncertainty in our understanding of the climate evolution under increased greenhouse gases in the atmosphere (Bony et al. 2015). High clouds in the tropics exert a significant contribution to the overall cloud feedback (Zelinka et al. 2012). The importance of the cloudiness induced by the tropical deep convection for the radiation budget is indeed recognized for a long time because of their effect in both the shortwave (SW) and longwave (LW) radiation (Webster and Stephens 1980; Del Genio and Kovari 2002). This cloud feedback is associated with various physical impacts ranging from stabilizing (Ramanathan and Collins 1991), destabilizing (Hartmann 2002; Zelinka and Hartmann 2010), or neutral (Del Genio et al. 2005) feedbacks prompting the need for further investigations.
In the tropics, upper-level cloudiness is mainly associated with mesoscale convective systems (MCS; Del Genio and Kovari 2002; Roca et al. 2017), which are an important aspect of climate sensitivity (Bony et al. 2016). On the modeling front, the representation of these MCS in CMIP-like climate models, as subgrid-scale processes, is problematic (Del Genio et al. 2012). Various research efforts have been proposed to overcome this problem and were directed either toward incorporating mesoscale organization into these climate model parameterizations (Donner et al. 2001; Elsaesser et al. 2017), via superparameterization (Kooperman et al. 2016), or by turning to global cloud resolving modeling framework (Palmer and Stevens 2019). The latter framework is designed to resolve the mesoscale convective systems. A better appreciation of the organized nature of tropical convection is accompanied with a renewed perspective on the role of these clouds on climate sensitivity to different physical perturbations. For instance, in radiative convective experiments, Hohenegger and Stevens (2016) found that climate is less sensitive to imposed increase in insolation when MCS are resolved. Similarly, the identification of self-aggregation of convection, one form of convective organization, in idealized simulations has been promoted as an interesting framework to explore the climate sensitivity to a doubling CO2 (Wing 2019). The importance of the organization of convective cloudiness for the radiation budget arises from the strong scale dependence of the radiative properties of the tropical cloud clusters (Roca and Ramanathan 2000; Roca et al. 2005). Organized cloud systems span a wide range of scales from 10 to 1000 km (Houze 2004) and the larger the cloud shield the lower the associated outgoing longwave radiation (OLR) and the brighter the system (high albedo) (Del Genio and Kovari 2002). The largest clusters are also the strongest contributors to the LW CRE (Roca et al. 2002). Analyzing deep convective cloud objects, Eitzen et al. (2009) found similar results with the larger objects having the highest albedos (lowest OLR) and corresponding to higher values in cloud optical depth, condensate water path, and low cloud-top temperature. Anvil cloud radiative properties are directly related to the cloud macrophysics (Vaillant de Guélis et al. 2017) and microphysics in the uppermost part of anvil clouds (Bouniol et al. 2010; Bouniol et al. 2016). The spatial area of the cloud clusters is driven by the life cycle of the MCS (Roca et al. 2017). Yet such an important aspect of the convection-induced cloudiness remains underappreciated in the abovementioned studies on climate sensitivity and high clouds in the tropics.
Numerous studies investigated the statistical relationships among the various radiation-related parameters as well as the way the environment drives these relationships but fewer studies investigated the relation to the deep convective cloud life cycle. Futyan and Del Genio (2007) focused on the West African and eastern Atlantic climate and use GERB satellite observations to reveal the dependence of the mean OLR and albedo of deep convective systems to the life cycle of the system. The system’s albedo exhibits a growing and decaying pattern centered on the mature phase of the system while the OLR is characterized by an inverted pattern with the minimum OLR being associated with the mature phase of the system. Using ground-based radar and geostationary observations, Feng et al. (2011) investigated the evolution of radiative properties of MCS as a function of the diurnal cycle over the United States. Their results suggested that newly formed MCS in the late afternoon are optically thicker than MCS dissipating in the morning. Using a complete set of A-Train and geostationary data and distinguishing the different MCS subregions (convective, stratiform, and cirriform), Bouniol et al. (2016) found a similar behavior as Futyan and Del Genio (2007), but additional investigation on cloud-related parameters allowed us to interpret the differences in terms of convection intensity (continental vs oceanic) and subsequent microphysical properties: denser hydrometeors and higher cloud top for continental convection. Wall et al. (2018) examined the evolution of anvil cloud properties derived from geostationary measurements over the MCS life cycle. They found a decrease of cloud-top pressure with time and the highest albedo just before the middle of the life cycle. A modeling study under idealized settings further indicated that the life cycle eventually yields to a near neutral cloud radiative effect (Hartmann et al. 2018) involving interaction between radiative, microphysical and turbulence processes. These studies all point toward the importance of the time evolution of the macro- and microphysical properties of the storms to produce the overall near neutral net radiative effects of the cloud system at the top of the atmosphere (TOA).
Here we propose to revisit the radiative properties of mesoscale convective systems in the tropics over their life cycle using satellite observations. A large observational dataset of MCS and the dedicated radiation measurements of the Scanner for Radiation Budget-3 (ScaRaB) instrument on board the Megha-Tropiques (MT) mission are used for this endeavor and the whole tropical belt is documented. The MT satellite is located on a precessing orbit, allowing us to include the consideration of the diurnal cycle in our analysis. The paper is structured as follows: First, data and methods for the merging of the two sets of observations are presented. The particularity of this paper is to consider the whole tropical belt including both summer and winter period and continental and oceanic environments. Evolution of radiative properties are then documented and contrasted in various environments. In the discussion section, impact of changes within the life cycle is assessed. A discussion and concluding remarks close the paper.
2. Data
a. Radiation at the top of the atmosphere
We use on the ScaRaB instrument on board the Indo-French MT satellite (Roca et al. 2015) to document the evolution of these MCS properties throughout their life cycle and over the diurnal cycle for the whole tropical belt. The low inclination orbit at 20° on the equator, the high altitude of flights, and the relatively wide swath (about 2200 km) enhance the sampling of the tropical regions, relative to typical low-Earth observing platforms (Roca et al. 2015). The precession cycle is responsible for the complex aliasing between the MT orbit and the diurnal cycle. This makes ScaRaB measurements the only dataset covering the whole tropics and sampling all local times for an entire year since ERBS (Rutan et al. 2014).
ScaRaB measures the total energy between 0.2 and 100 μm and the SW channel (0.2–4 μm) is subtracted to obtain the LW part of the spectrum. ScaRaB has been designed to measure Earth radiation components at the TOA with an accuracy below 1% (Kandel et al. 1994). The ScaRaB radiometer performs regular onboard calibration using a calibration module (Duvel et al. 2001). Cross comparison of measured radiances as well as OLR and OSR is also regularly performed with CERES instrument on board Terra showing that both technologies produce consistent estimates of outgoing radiation at the TOA (Trémas et al. 2015). ScaRaB on MT has a square-shaped pixel footprint of 41.6-km side length at nadir with pixel overlapping by 29.6 km in the cross-track direction and 36.6 km along track. The radiances are converted to hemispheric flux at the TOA after determination of the scene type using the additional infrared (IR) channel 10.5–12.5 μm and application of angular distribution models (ADMs). These ADMs have been built using artificial neural networks trained on CERES on board TRMM for which all the solar zenith angles (SZA) are sampled (Viollier et al. 2009).
The ScaRaB flux estimates used in this study is the level 2B product and corresponds to the projection of the original square-shaped ground pixels of the radiometer on a 0.5° × 0.5° regular grid considering the point spread function of the instrument (Gif et al. 2011). To avoid undersampling at the swath edge only viewing zenith angles lower than 50° are considered in the present analysis. Since the calibration of the instrument is in very good agreement with CERES and the algorithm shares the ADM in the radiance to flux conversion, CERES and ScaRaB fluxes also agree within a few percent with each other (Roca et al. 2015; Trémas et al. 2015). The error budget of the CERES products provides a reference to discuss the error budget of the ScaRaB fluxes. The overall uncertainty of the TOA instantaneous all-sky flux is about 1.8% in the LW and 3% in the SW (Loeb et al. 2007). When only pixels overcast by upper-level ice clouds are considered the budget reads between 2.3% and 4.1% for the SW and 4.6% and 5.4% for the LW. These rough estimates will serve as a baseline to interpret the sensitivity experiments detailed in section 5.
b. Morphology of mesoscale convective systems
MCS are identified and tracked by the Tracking of Organized Convection Algorithm through a 3D Segmentation (TOOCAN, Fiolleau and Roca 2013a), which is applied to IR images collected by geostationary platforms distributed around the equator. This algorithm is based on an original 3D image processing technique that aims at segmenting the high cold cloudiness delineated in the spatiotemporal domain by a 235-K threshold into MCS. This 235-K threshold is a good compromise to include as much anvil clouds as possible, avoiding the contribution of low and midlevel clouds that may develop under thinner cirrus clouds (Bouniol et al. 2016). The main skills of the TOOCAN algorithm are to overcome the traditional split and merge artifacts (Fig. 1a), to track MCS from their earliest stages of initiation up to their latter stages of dissipation, and to identify MCS regardless of their degree of organization. A global tropical coverage by the geostationary satellites is achieved by using data from Meteosat Second Generation (MSG-1, MSG-2, MSG-3), Himawari-8, GOES-13 and GOES-15, Meteosat-7, and Multifunctional Transport Satellite 2 (MTSAT-2) during the 2012–16 period. The IR imagers on board these platforms all have different spatial resolutions, spectral response functions, and calibration drifts. These data have been homogenized and intercalibrated. Inhomogeneity of the spatial resolution has been taken into account by remapping all the geostationary data to a common 0.04° equal angle grid. To avoid an oversegmentation of the MCS detected by TOOCAN, the temporal resolution is aligned at 30 min regardless of the geostationary platform. The observed brightness temperatures are intercalibrated, spectrally normalized, and limb-darkening-corrected against the 2012–16 observations of the ScaRaB IR channel, yielding to a residual uncertainty on the order of 2 K (Fiolleau et al. 2020). The TOOCAN algorithm has then been applied to this 5-yr IR homogenized database, giving access to the integrated morphological parameters of each identified tropical MCS (lifetime duration, propagated distance, etc.), as well as the morphological parameters at each stage of their life cycles (cold cloud surface and average brightness temperature).
Illustration of the composite process for cloud systems observed at 0400 UTC 12 Oct 2015 over the Pacific Ocean. (a) Brightness temperature observed by Himawari identified by the TOOCAN algorithm (one color per object). (b) OLR (color level) observed by ScaRaB-3 on board Megha-Tropiques. (c) As in (b), but for TOA albedo. Two zooms within this sample are also shown [location depicted by the two white-outlined rectangles in (a) and repeated in black in (b) and (c)] to illustrate particular sampling questions (see the text). Black contours in (b) and (c) show the edges of the identified MCS.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
3. Methods
a. Building life cycle-resolved composite of radiation properties
A simple classification of MCS based on duration is used (Fiolleau and Roca 2013b). Class 1 consists of MCS lasting less than 5 h, and class 2 consists of longer MCS. A further separation is then applied among the class-2 systems, which are then split according to the complexity of the evolution of their cloud shield. Class-1 MCS only explain 2.5% of the cold cloudiness even though they are very numerous. Class-2a MCS correspond to systems that only have one growth and decay of their cloud shield surface. They explain 84% of the cold cloudiness. The class-2b MCS have a more complex evolution of their cloud shield area and they explain about 13.5% of the cold cloudiness at the tropical scale (Roca et al. 2017).
The life cycle of each individual class-2a system is then divided in 10 life stages in the following and the complete life cycle can be documented by aggregating numerous ScaRaB observations of MCS at the same life stage. This process has already been applied to the macrophysical, microphysical and radiative properties as measured by the A-Train for MCS sampled in three different geographical regions (Bouniol et al. 2016), and for radiative and microphysical properties derived from Himawari observations (Wall et al. 2018).
In the present study, the pixels of the TOOCAN segmented images at a 0.04° spatial resolution are collocated spatially and temporally with the ScaRaB observations at a 0.5° spatial resolution. This process is illustrated in Figs. 1b and 1c for the OLR and the albedo (defined as the ratio of OSR to the incoming SW flux) with the MCS edges superimposed. Lower OLR values and higher albedo values are systematically found within MCS contours. Figures 1b and 1c also highlight two other important aspects: a given 0.5° grid point can be partially or totally filled by one or several MCS. A “filling” fraction Ff is introduced and corresponds to the fraction of a 0.5°ScaRaB grid point occupied by an MCS identified by TOOCAN. This filling fraction is determined by the ratio of the number of MCS pixels at a 0.04° spatial resolution collocated within a 0.5° grid point and the number of geostationary 0.04° pixels located within a 0.5° grid point. The second aspect is that the finite swath width of the ScaRaB instrument partially samples MCS located at the swath edge. The impacts of both effects on the MCS population contributing to the life cycle composite are investigated in the following.
b. Merging data with different spatial resolution
The potential impact of the different resolutions between ScaRaB and TOOCAN output is illustrated in Fig. 1c where only two 0.5°ScaRaB grid points (the brightest) are nearly completely enclosed within the MCS contours. These two ScaRaB grid points have high values of Ff. A relatively high albedo is found for this MCS if only high Ff are considered to compute the averaged radiative properties at the scale of the MCS. In this case, the averaged albedo computed is not representative of the MCS albedo as the edges of the MCS are not considered. If any ScaRaB grid point that hits the MCS (hence any grid point with Ff > 0) is included in the mean computation (dotted contour), a relatively lower value of albedo is found as grid points with a low Ff include contribution from the MCS neighborhood, for example, other cloud types or even clear sky. In this case the averaged value at the MCS scale is not representative of the SW flux reflected to space by this particular object. The same effect exists for OLR. To overcome this issue, a trade-off between these two pictures is needed and an Ff threshold above which 0.5° ScaRaB grid points should be included in MCS averaged radiative properties needs to be identified.
For that purpose, we focus on MCS observed by the imager on board MSG-2 for the 2012–16 period. First, for each identified MCS at each life stage, the “reference” brightness temperature (Tbgeo) is computed as the average of the radiance (at the 0.04° IR image resolution) enclosed within the MCS contour and then converted to brightness temperature. The same radiances are averaged and converted to brightness temperature in the 0.5 × 0.5°ScaRaB grid and the corresponding Ff is determined. From these lower resolution datasets the average brightness temperature is recomputed at the scale of the MCS by considering only ScaRaB grid points above a given value of Ff, noted Tbgrid. Comparing the two mean brightness temperatures makes it possible to determine the more suitable filling threshold that keeps Tbgrid as close as possible to Tbgeo. Figure 2a shows the evolution of the difference between these two mean brightness temperatures as a function of the MCS area for different threshold in Ff (the different colors). As expected, the lower filling thresholds induce a bias that can reach 26 K for MCS of 2 × 103 km2, which means that a large fraction of the atmosphere has either clear sky or lower clouds that contribute to Tbgrid. This difference decreases as the MCS area increases, even if it remains higher than 10 K, because in these situations the fraction of high-Ff grid points increases at the expense of low-Ff grid points. The opposite effect is observed if Tbgrid is computed using only high-Ff grid points (dark red). In this case a negative bias is obtained, which means that only the coldest parts of the MCS are included in the computation of Tbgrid and a large part of the MCS edges are not included anymore.
(a) Difference in mean brightness temperature at the scale of MCS (K) as a function of MCS area (km2) computed from the full geostationary spatial resolution (Tbgeo) and from the ScaRaB resolution (Tbgrid) considering different Ff (colors). The numbers to the right of (a) show the number of MCS considered in the comparison. (b) Population of MCS as function of the MCS area for the different filling thresholds.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
Computing Tbgrid with Ff ≥ 70% grid points appears as a reasonable trade-off between not considering the MCS edges and not introducing a bias in the average values of Tbgrid to the colder characteristics as the differences between the two temperatures remain smaller than 2 K (in absolute value), which corresponds to the uncertainty attached to the IR dataset of Fiolleau et al. (2020) for MCS larger than 6 × 103 km2. Therefore, MCS of smaller areas cannot be accurately documented using the 0.5°ScaRaB observations. This critical area of 6 × 103 km2 corresponds to an equivalent radius of 43 km, which is a typical scale used to define MCS in Xu et al. (2005) or in Del Genio and Kovari (2002) where an MCS is defined by contiguous 85 GHz TMI rain pixels having an equivalent radius of 25 km. The 235-K envelope is largely enclosed in this raining part and the nonprecipitating anvil can exceed twice this raining region. Imposing this 70% filling threshold reduces by 44% the number of class-2a MCS that can be used to document the life cycle with the ScaRaB data. This ratio is the same if the full tropical belt is considered. Including the critical area threshold in order to keep the difference lower than 2 K leads to the removal of an additional 6% of the available sample. This means that 50% of the class-2a detected MCS can be considered in the present study.
The Fig. 2b shows the population of MCS as a function of their area and the different filling thresholds. As expected from the earlier considerations the more numerous MCS are the ones of smallest area (lower than 100 km2) and considering MCS that have Ff at least at 70% points toward the largest MCS. MCS are known to be smaller at the beginning and at the end of their life cycle (Roca et al. 2017). This implies that these life stages will be less sampled or sampled with a larger uncertainty in the present study, as documented in the next section.
c. Impact of a finite swath width, the MT orbit, and ScaRaB data resolution on the MCS sample
Figure 3a shows the area distribution of class-2a MCS detected by TOOCAN within the five years of data without (solid lines) and with including the minimum area and filling threshold defined in the previous section (dashed line). The total distribution of class-2a MCS (bold solid black line) has its mode value about 3.2 × 103 km2 with a majority of MCS having a surface between 103 and 5 × 104 km2. Imposing a minimum Ff of 70% and a 6 × 103 km2 minimum area automatically shifts the distribution toward wider cluster (dashed line), with a modal value that becomes 1.5 × 104 km2.
(a) MCS area distribution independent of Ff and with a minimum Ff of 70% and a minimum area of 6 × 103 km2 (dashed lines). The black line shows the distribution obtained from the TOOCAN database at the scale of the whole tropics; the colored lines show the distributions considering the ScaRaB orbit and swath assuming different thresholds of MCS partial sampling. The numbers in the upper left are the total number of MCS used to compute the distributions for the different swath–MCS intersections. The second number for a 50% intersection corresponds to the number of MCS that can be considered for composite building assuming a 70% Ff and the minimum sampling area. (b) Number of MCS tracked by TOOCAN at the scale of the whole tropics (solid line), sampled by ScaRaB, even partially (dashed line), and at least 50% sampled by ScaRaB, with an MCS grid filling of 70% and a minimum area of 6 × 103 km2 (dotted line).
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
MT intercept 25% of class-2a MCS at the scale of the tropical belt (corresponding in this study to all latitudes lower than 30°N and 30°S). However, this number does not consider that a given fraction of these MCS may be partially sampled by the MT wide swath. The requirement of a minimum overlap between the cluster surface and the ScaRaB swath (color lines) tends to narrow the distribution around the modal value. However, in all cases the area distribution obtained using the MT orbit remains close to the distribution obtained directly from the TOOCAN database. Based on these observations we only consider class-2a MCS that are at least 50% enclosed in ScaRaB swath. This leads to a reduction of about 10% of the MCS sample. Only considering clusters entirely sampled (dark red curve) by ScaRaB reduces the number of MCS by 25%. In the present case we are far from the TRMM PR swath (215 km) impact as documented by Nesbitt et al. (2000) for which more than 54% of the mesoscale storm and 82% of more intense storm detected using the same definition as Del Genio and Kovari (2002) are partially sampled. The number of clusters is approximately divided by two when the filling/area threshold is considered (second number for the green 50% intercept line on Fig. 3a).
The solid line in Fig. 3b shows the number of class-2a MCS detected by TOOCAN. The MT sampling reduces the number of class-2a MCS detected by TOOCAN to the number shown by the dashed line in Fig. 3b. Applying the filling/area constraint leads to the dotted line and has the effect of undersampling the beginning and the end of the life cycle with the number of MCS reduced by an order of magnitude with respect to the dashed line. The results for these two bins are included in the following (generally shown with dashed lines) but the observational limitations should be kept in mind for analysis of the MCS life cycle.
As a summary, composite life cycles are built using MCS that are intercepted by more than 50% of the ScaRaB swath, with a horizontal extent larger than 6 × 103 km2. Their averaged radiative properties are constructed by averaging ScaRaB grid points filled by more than 70% of an MCS.
4. Analysis of radiative properties over the MCS life cycle
In this section the evolution of radiative property over the MCS life cycle is analyzed. The evolution of the radiative property is investigated for MCS over the whole tropical belt, and in a second step for different major convective areas.
a. Analysis at the tropical scale and for different environments
A tropical mean composite of OLR and albedo as a function of the life stage is computed and shown in Figs. 4a and 4b. A well-defined minimum in OLR and a maximum in albedo is observed once in the life cycle, which is a similar behavior to that found by Futyan and Del Genio (2007). The present albedo strongly differs with results from Bouniol et al. (2016) despite the use of the same method. In this earlier study, albedo values were examined separately for the different parts of the MCS, but the albedo at the scale of the whole MCS can be approximated to the albedo of the cirriform part, which has the larger areal contribution to MCS anvil. In most cases, these values remain around 0.5. However, the 0.6 value found at the scale of the tropical belt is rather close to the values found in stratiform parts in Bouniol et al. (2016). For OLR the black bold line in Fig. 4a may be interpreted as an area-weighted average (in particular at the beginning of the life cycle) of the OLR values found for the various parts of the MCS.
Mean composite (thick black line) of (a) OLR (W m−2) and (b) albedo as a function of the life stage at the scale of the whole tropical belt for MCS tracked between 2012 and 2016 by TOOCAN. Different colors show the mean composites according to different local solar times. (c),(d) The fraction of MCS sampled at each LST for each life stage, where each column is normalized by the number of MCS observed in a given life stage.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
The specificity of the Bouniol et al. (2016) composites is that they were built from the A-Train data located on a sun-synchronous orbit with an equatorial crossing time of 0130 local time (LT). The precessing orbit of the MT satellite is able to sample all the local times for each life stage and hence to further analyze the composite shown in Fig. 4 (thick black line) according to the local solar time (LST) of observation by MT. These LST composites are shown as color lines in Figs. 4a and 4b. As expected, OLR shows a relatively low sensitivity to the LST. The variability found in Fig. 4a is mainly related to the variation in temperature of the underlying surface, as documented later. The spread of albedo as a function of LST is wider, with lower albedos for 1200 LST than for sunset and sunrise. These higher values at dusk or dawn are due to the apparent increase in cloud fraction with zenith angle and the higher Rayleigh scattering of radiation with increasing zenith angle (Liou 2002; Rutan et al. 2014). For a given life stage, the spread in albedo due to solar zenith angle (which is about 0.2) is larger than the variations found within the life cycle (which are between 0.05 and 0.1 for a given solar time). The green lines in Fig. 4, close to the 1330 LT A-Train sampling, show the same order of magnitude as in Bouniol et al. (2016). Particular attention should hence be exerted when building composites of albedo using precessing satellite observations since the resulting composite contains a signal from the MCS life cycle (changes observed along a color line in Fig. 4) and microphysical property evolution on one hand and a signal from the diurnal cycle (spread among the different color lines in Fig. 4) on the other hand.
Figures 4c and 4d show the distribution of MCS as a function of LST for each life stage. Even at the scale of the full tropical belt a preferred time of triggering of MCS is observed in the middle of the afternoon and MCS are the largest between 1700 and 1800 LST (middle of the life cycle) when the sun is down, as shown in both Fig. 4c for OLR and Fig. 4d for albedo measurements. The impact of this heterogeneous sampling will be further analyzed in the following sections.
Figure 5 shows oceanic and continental MCS statistics. An oceanic MCS is defined as a cluster that is observed over the oceanic surface but never passed over land surface. This definition therefore excludes all MCS initiated over islands that export over the oceanic surface, which may artificially create a diurnal cycle of oceanic convection (Yang and Slingo 2001). Figures 5e–h show the contrasted diurnal cycle of convection between land and sea with a preferential time of triggering between 1300 and 1400 LST (and nearly no MCS triggering between 0500 and 1000 LST), which imprints a very preferential LST of observation for a given life stage (Figs. 5g and 5h). The occurrence of MCS remains rather homogeneous for the oceanic surfaces (Figs. 5e and 5f), even if small preferential daytime convective life cycle seems to exist.
As in Fig. 4, but for (a),(b),(e),(f) oceanic MCS and (c),(d),(g),(h) continental MCS.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
The effect of continental surface is clearly seen in OLR composites (Figs. 5a and 5c) with a reduced spread for composite life cycle as a function of LST for oceanic MCS. Both spreads have the same order of magnitude (between 10 and 20 W m−2 and between 0 and 10 W m−2, respectively) as the root-mean-square of the first principal component of OLR calculated by Smith and Rutan (2003) attributed in their study to solar heating of the surface–atmosphere system. For continental MCS, the OLR evolutions according to LST are split in two families: the night times, with lower values in OLR in particular at the beginning of the life cycle, and the day times, with higher values in OLR. The resulting composite (bold black lines in the first row in Fig. 5) corresponds to the average of the colored line weighted by the fraction of MCS that occurs at each LST within a given life stage. This weighted average leads to very similar shape in OLR evolution over the MCS life cycle for both land and ocean. A well-defined minimum is observed for both about life stage 4, with just slightly lower values for continental MCS. As triggering time is relatively homogeneous over the day for oceanic areas, a part of the spread of the evolution of the radiative properties over MCS life cycle may be attributed to different evolution according to the triggering time as documented by Sokol and Hartmann (2020). More quantitatively, Wall et al. (2020), using a similar composite approach, found a difference in OLR of about 10 W m−2 for aged anvil cloud over the western Pacific. Higher values of OLR are found for MCS triggering and developing during night but are aged during daytime.
The mean composites of albedo (Figs. 5b and 5d) are different among the oceanic and the land environments with a well-defined maximum for continental MCS and a constant value followed by a decrease for oceanic MCS. This difference in shape in albedo evolution over the MCS life cycle is mainly due to the diurnal cycle of continental MCS, with an average weighted by the high sunset albedos at about the middle of the life cycle. However, after the middle of the life cycle, the albedos of oceanic MCS are systematically lower than albedos of continental MCS regardless of the LST (or the SZA). This means that some differences in microphysical properties such as higher ice water content or/and cloud optical depth should exist among continental and oceanic MCS during the second phase of their life cycle.
b. MCS life cycle at regional scale
Figures 6b and 6c show the comparison of mean composites for three oceanic domains of Fig. 6a. The evolution of albedo is on average remarkably similar within the three oceanic regions. However, albedo values are systematically higher for MCS sampled in the Atlantic Ocean (solid lines) than for MCS sampled in the Pacific Ocean (dotted lines). They are themselves brighter than MCS sampled in the Indian Ocean (dashed lines), when split according to SZA (for life stages 2, 5, and 8 corresponding to growing, mature, and decaying phase, respectively). This holds in particular when the sun is close to zenith (about 1200 LST) corresponding to the lower values of albedo. The similar resulting averages (Fig. 6c) are a consequence of the sampling within the diurnal cycle, despite very similar patterns (Figs. 6d–f).
Comparison of OLR and albedo over the MCS life cycle for (a) three oceanic domains. (b) Mean OLR as a function of life stage. (c) Mean albedo as a function of life stage and evolution of albedo with the solar zenith angle for three life stages 2, 5, and 8. Distribution of MCS as a function of LST for each life stage for the (d) Atlantic, (e) Indian, and (f) Pacific domains.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
A systematic difference exists for the Atlantic Ocean where MCS have an OLR about 5 W m−2 higher than those in the Indian and Pacific Oceans. Bouniol et al. (2016) found a lower cloud-top altitude, as observed by the CALIPSO lidar, for Atlantic MCS with respect to the Indian Ocean MCS by about 1 km. A similar cloud-top altitude is found in the CloudSat observations, which implies a deeper layer of thin clouds at the top of Indian Ocean anvil clouds. This difference in cloud-top altitude as detected by the lidar is about 500 m at stage 4. Vaillant de Guélis et al. (2017) demonstrated that the OLR is directly linked to the altitude at which the lidar signal vanishes (Topaque) and to the cloud top detected by the lidar. They showed that OLR decreases by 2 W m−2 when Topaque decreases by 1 K. In the present case, Topaque can be approximated to the cloud top detected by the radar and the 5 W m−2 difference can be attributed to the 500-m-lower lidar cloud-top altitude in the Atlantic Ocean. These observations demonstrate that this layer of thin clouds at the top of deep clouds has an impact on cloud emissivity.
Figure 7 shows the evolution of radiative properties for continental MCS. Very contrasted continental regions were selected: the semiarid region of West Africa, the continental South America (including the Amazonia region) and the “Maritime Continent.” The latter is not properly a continental region, but rather a mix between island and surrounding ocean, and all MCS are included in the composites without any distinction. The distribution of MCS for the Maritime Continent is very similar to oceanic MCS. Interestingly a shift of about two hours in MCS triggering is observed between West Africa and South America and this shift in preferential time of occurrence persists for the rest of the life cycle.
As in Fig. 6, but for continental MCS sampled in West Africa, South America, and the Maritime Continent.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
Radiative properties have different behaviors depending on the continental area being considered (Figs. 7b and 7c). Values in OLR differ up to stage 5, with the lowest values found in West Africa and marked deepening for MCS of South America with a decrease of 10 W m−2 in the first half of the life cycle. Indeed, West Africa MCS reach low OLR values after the first stages of their life cycle. The evolution of MCS radiative properties over the life cycle in Maritime Continent shows the same shape and values as the other oceanic regions examined previously. After the middle of the life cycle, OLR shows a similar behavior for all the regions, with an increase of about 20 W m−2 in the decaying phase.
The life cycle–resolved albedo for Maritime Continent MCS follows that of oceanic MCS. For low SZA, West African MCS are the brightest, with high albedos observed at the beginning of their life cycles. But albedo of South America MCS becomes higher at half of the life cycle to reach the values of West Africa about stage 5, followed for both regions, by a slow decrease, with the same slope, but still higher by about 0.05 than for oceanic MCS. Higher values for both continental regions with respect to ocean MCS confirm that the intense convection over land surfaces generates denser hydrometeors that are more efficient for scattering and reflecting solar flux in agreement with Del Genio and Kovari (2002) or Sohn et al. (2015). When examined as a function of SZA, the differences at the beginning of the life time of MCS between South America and West Africa results from different albedo properties. Values are lower for South America (dashed lines in Fig. 7) than for West Africa (solid line in Fig. 7) regardless of the SZA. Sohn et al. (2015) found lower effective radius and ice water content for the whole vertical profile for South America’s MCS with respect to Africa.
As a summary, regional differences in shape and magnitude of the composite life cycle–resolved radiative properties are found. They are related to the convective controlling factor. For the albedo, the major influence appears to be the phasing within the diurnal cycle.
5. Discussion
In the previous section, it was demonstrated that OLR and albedo are shaped by different parameters over the life cycle of MCS. For the albedo, differences in magnitude (even split with SZA) are observed, in particular between continental and oceanic areas, as well as on the regional scale. However, the SZA (or the LST) parameter seems to have the strongest influence on albedo. To quantify how these different parameters recombine at the scale of the life cycle, the evolution of the albedo of a “typical” MCS of a given lifetime duration for a usual triggering time can be simulated from the different composites. The different color lines in Fig. 8 show the evolution of albedo over the life cycle for the different LST. Depending on the triggering time and the life duration, albedo of a typical MCS jumps from one color line to the other as depicted by the black crosses. Over continental areas, MCS are mainly triggered during the afternoon and have therefore low albedos that increase rapidly as the sun sets (Figs. 6 and 7). These continental MCS reach their maximum albedo values at the middle of their life cycles, which is also the time stage when MCS reach their largest area (Roca et al. 2017). In the particular case of a system lasting about 10 h, it is also a time when sun is down (Fig. 8). The sensitivity to the shift in triggering time is also observed (Figs. 8a and 8b) where the earlier triggering in South America offers a longer exposure to the sun light. Oceanic convection (Fig. 8c) does not have a strong diurnal cycle (Fig. 6) but is regularly observed to be triggered in the morning (Yang and Slingo 2001). Wall et al. (2020) also show that anvil clouds are more laterally expansive and more persistent when heated by the sun and hence can largely contribute to the radiative budget. This morning triggering corresponds to high values of albedos (because of large SZA) but small area. These albedo values then decrease to lower values as the sun rises. Oceanic MCS also reach their lower values in albedo after the middle of their life cycle. Despite a relatively well distributed triggering of the convection within the diurnal cycle, the shape of the evolution of the simulated life cycles with LST in Fig. 8c for the longest MCS is very similar to the shape of the mean albedo of marine convective region as a function of LST of Rutan et al. (2014) (see their Fig. 1). In short, the diurnal cycle of convection tends to amplify the albedo properties of continental areas and to moderate the one of oceanic areas, even if microphysical properties are identical.
Simulated life cycle for MCS of different life time durations: 10 (solid black line) and 20 (dashed black line) h for typical times of triggering (a) 1500 LST in West Africa, (b) 1300 LST in South America, and (c) 0700 LST in the Pacific Ocean. Colored lines show the composites of albedo for all observed MCS according to the LST (color scale).
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
Distribution of anomalies in CRE resulting from the omission of the evolution of radiative properties with the life cycle as a function of the gridpoint filling (the distribution being normalized for each Ff bin) for (a) continental Africa, (b) South America, and (c) Pacific Ocean. For each region the anomalies in (top) LW, (middle) SW, and (bottom) net radiation are shown. Dashed lines show the mean anomaly in CRE as a function of Ff.
Citation: Journal of Applied Meteorology and Climatology 60, 8; 10.1175/JAMC-D-20-0244.1
Omission of the evolution of radiative properties within the life cycle leads to an underestimation of CRE in the LW and an overestimation in the SW because this anomaly is a function of Ff. The obtained values are higher for larger Ff and can be larger than ±−10 W m−2. Distributions in the SW are relatively similar in the considered regions, with positive anomalies that can be higher than 10 W m−2. The LW contributions have more differences, with more spread over continental Africa for large Ff with respect to Pacific. The net CRE anomaly for continental Africa is dominated by the anomaly in the LW because of the late triggering of MCS as illustrated in Fig. 8. In the South America region where MCS are triggered two to three hours earlier, both anomalies tend to compensate. This compensation effect is observed even stronger in the Pacific Ocean. However, the compensation effect may also be fortuitous as both errors have effects that compensate. It would not have occurred if different assumptions were made for this sensitivity test and points toward compensating errors that may induce small errors on radiative properties. In summary this sensitivity experiment has shown that neglecting the evolution of radiative properties with the life cycle led to uncertainty in radiative properties that may exceed the accuracy of flux estimates especially when the MCS have a wide area (and therefore occupy several grid points with high Ff). The occurrence within the diurnal cycle is of particular importance for balancing the effect in the SW and in the LW: error on MCS radiative properties are more easy to balance between SW and LW in oceanic regions where MCS spent more time in daylight as compared with continental regions where MCS radiative properties are dominated by effect in the LW.
6. Conclusions
This paper explores the evolution of radiative properties throughout the MCS life cycle using TOA observations from the ScaRaB-3 radiometer on board the Megha-Tropiques satellite. The data analysis is carried at the whole tropics scale using five years of collocated geostationary and ScaRaB-3 data. The MCS life cycle is documented using a tracking algorithm applied to geostationary data. Each time ScaRaB swath intercepts an MCS the life stage is stored as well as the TOA radiative properties measured at the scale of the MCS. This approach is used to build a composite view of the evolution of radiative properties over the MCS life cycle using the tracking as a framework. This study extends former ones (Futyan and Del Genio 2007; Feng et al. 2011; Bouniol et al. 2016; Wall et al. 2018) focused on limited season and geographical areas.
Particular attention has been paid to the limitations resulting from sampling ability of the ScaRaB-3 radiometer. This step ensures that the isolated signal is well attributable to MCS anvil cloud. To do so we showed that only ScaRaB grid points filled at least at 70% by a convective cloud and belonging to MCS larger than 6 × 103 km2 should be considered. The wide swath of the ScaRaB only reduces the considered sample, without introducing a bias toward a particular population.
Composites are built at the tropical scale, and then separately for continent and ocean or according to particular geographical areas. All composites share common features, including a unique maximum in albedo and a unique minimum in OLR. However, amplitudes, minimum and maximum values of albedo and OLR and amplitude of the life cycle differ depending on the region. In particular continental MCS are brighter (in terms of albedo) and colder (in terms of OLR) in agreement with previous studies. The precessing orbit allows us to distinguish the albedo values for each LST (or SZA) and if the shape of the evolution with the life cycle is kept unchanged for a given LST the values are higher when SZA is small. The variations in albedo with SZA dominate the variations of albedo throughout the life cycle in response to the evolution of the microphysical properties within the anvils. Evolution in microphysical and macrophysical properties are directly responsible for changes in OLR. A first conclusion that can be drawn from this study is the absolute necessity of considering the SZA when assessing cloud radiative properties in the SW because normalization by the incoming solar radiation does not remove this dependency, as highlighted by Rutan et al. (2014). In addition, Wall et al. (2020) raise the facts that the diurnal variations of radiative heating affect the anvil life cycle and that MCS triggering and developing during nighttime have different macrophysical and radiative properties than does the one triggering and developing during daytime because of a different balance between shortwave and longwave heating.
These results highlight the subtle combination at play among occurrence, cloud area, radiative properties, and phasing with diurnal cycle in the building of radiation budget at regional scale and further confirm the importance of this combination at the whole tropical scale. Our study indicates that uncertainty in the representation of life cycle related radiative properties may lead to anomalies in CRE larger than 10 W m−2 in absolute value, in particular when the systems are the largest (i.e., about the middle of their life cycle). These CRE anomalies may exceed the magnitude of the uncertainty in the current flux estimations, stressing the importance of accounting for the life cycle–resolved systems properties in interpreting the radiative budget. The results also show significant regional variability that prompts cautionary use of local microphysical properties such as those obtained during fields campaigns on a more global scale.
Acknowledgments
This study benefited from the IPSL mesocenter ESPRI facility, which is supported by CNRS, UPMC, Labex L-IPSL, CNES, and Ecole Polytechnique. The authors acknowledge the CNES and CNRS support under the Megha-Tropiques program.
Data availability statement
The output dataset from the TOOCAN algorithm for the 2012 and 2016 period is publicly available (https://doi.org/10.14768/20191112001.1). More information can be found online (http://toocan.ipsl.fr). The ScaRaB-3 data can be downloaded. The ScaRab L2B are available in the ICARE data and service center with an anonymous file transfer protocol (
REFERENCES
Bony, S., and Coauthors, 2015: Clouds, circulation and climate sensitivity. Nat. Geosci., 8, 261–268, https://doi.org/10.1038/ngeo2398.
Bony, S., B. Stevens, D. Coppin, T. Becker, K. A. Reed, A. Voigt, and B. Medeiros, 2016: Thermodynamic control of anvil cloud amount. Proc. Natl. Acad. Sci. USA, 113, 8927–8932, https://doi.org/10.1073/pnas.1601472113.
Bouniol, D., J. Delanoë, C. Duroure, A. Protat, V. Giraud, and G. Penide, 2010: Microphysical characterisation of West African MCS anvils. Quart. J. Roy. Meteor. Soc., 136, 323–344, https://doi.org/10.1002/qj.557.
Bouniol, D., R. Roca, T. Fiolleau, and E. Poan, 2016: Macrophysical, microphysical, and radiative properties of tropical mesoscale convective systems along their life cycle. J. Climate, 29, 3353–3371, https://doi.org/10.1175/JCLI-D-15-0551.1.
Del Genio, A. D., and W. Kovari, 2002: Climatic properties of tropical precipitating convection under varying environmental conditions. J. Climate, 15, 2597–2615, https://doi.org/10.1175/1520-0442(2002)015<2597:CPOTPC>2.0.CO;2.
Del Genio, A. D., W. Kovari, M. S. Yao, and J. Jonas, 2005: Cumulus microphysics and climate sensitivity. J. Climate, 18, 2376–2387, https://doi.org/10.1175/JCLI3413.1.
Del Genio, A. D., J. Wu, and Y. Chen, 2012: Characteristics of mesoscale organization in WRF simulations of convection during TWP-ICE. J. Climate, 25, 5666–5688, https://doi.org/10.1175/JCLI-D-11-00422.1.
Donner, L. J., C. J. Seman, and R. S. Hemler, 2001: A cumulus parameterization including mass fluxes, convective vertical velocities, and mesoscale effects: Thermodynamic and hydrological aspects in a general circulation model. J. Climate, 14, 3444–3463, https://doi.org/10.1175/1520-0442(2001)014<3444:ACPIMF>2.0.CO;2.
Duvel, J. P. M., P. Viollier, R. Raberanto, M. Kandel, L. A. Haeffelin, V. A. Pakhomov, J. Golovko, and R. Mueller, 2001: The ScaRaB-Resurs Earth Radiation Budget dataset and first results. Bull. Amer. Meteor. Soc., 82, 1397–1408, https://doi.org/10.1175/1520-0477(2001)082<1397:TSRERB>2.3.CO;2.
Eitzen, Z. A., K.-M. Xu, and T. Wong, 2009: Cloud and radiative characteristics of tropical deep convective systems in extended cloud objects from CERES observations. J. Climate, 22, 5983–6000, https://doi.org/10.1175/2009JCLI3038.1.
Elsaesser, G. S., A. D. Del Genio, J. H. Jiang, and M. van Lier-Walqui, 2017: 2016: An improved convective ice parameterization for the NASA GISS global climate model and impacts on cloud ice simulation. J. Climate, 30, 317–336, https://doi.org/10.1175/JCLI-D-16-0346.1.
Feng, Z., X. Dong, B. Xi, C. Schumacher, P. Minnis, and M. Khaiyer, 2011: Top-of-atmosphere radiation budget of convective core/stratiform rain and anvil clouds from deep convective systems. J. Geophys. Res., 116, D23202, https://doi.org/10.1029/2011JD016451.
Fiolleau, T., and R. Roca, 2013a: An algorithm for the detection and tracking of tropical mesoscale convective systems using infrared images from geostationary satellite. IEEE Trans. Geosci. Remote Sens., 51, 4302–4315, https://doi.org/10.1109/TGRS.2012.2227762.
Fiolleau, T., and R. Roca, 2013b: Composite life cycle of tropical mesoscale convective systems from geostationary and low Earth orbit satellite observations: method and sampling considerations. Quart. J. Roy. Meteor. Soc., 139, 941–953, https://doi.org/10.1002/qj.2174.
Fiolleau, T., R. Roca, S. Cloché, D. Bouniol, and P. Raberanto, 2020: Homogenization of geostationary infrared imager channels for cold cloud studies using Megha-Tropiques/ScaRaB. IEEE Trans. Geosci. Remote Sens., 58, 6609–6622, https://doi.org/10.1109/TGRS.2020.2978171.
Futyan, J. M., and A. D. Del Genio, 2007: Deep convective system evolution over Africa and the tropical Atlantic. J. Climate, 20, 5041–5060, https://doi.org/10.1175/JCLI4297.1.
Gif, N., O. Chomette, and P. Raberanto, 2011: Co-location algorithms: Geophysical data projection using pixel point spread function. Megha-Tropiques Tech. Memo. 2, 28 pp., https://meghatropiques.ipsl.fr/download/megha-tropiques-technical-memorandum-n2/#.
Hartmann, D. L., 2002: An important constraint on tropical cloud–climate feedback. Geophys. Res. Lett., 29, 1951, https://doi.org/10.1029/2002GL015835.
Hartmann, D. L., B. Gasparini, S. E. Berry, and P. N. Blossey, 2018: The life cycle and net radiative effect of tropical anvil clouds. J. Adv. Model. Earth Syst., 10, 3012–3029, https://doi.org/10.1029/2018MS001484.
Hohenegger, C., and B. Stevens, 2016: Coupled radiative convective equilibrium simulations with explicit and parameterized convection. J. Adv. Model. Earth Syst., 8, 1468–1482, https://doi.org/10.1002/2016MS000666.
Houze, R. A., Jr., 2004: Mesoscale convective systems. Rev. Geophys., 42, RG4003, https://doi.org/10.1029/2004RG000150.
Kandel, R., J.-L. Monge, M. Viollier, L. A. Pakhomov, V. I. Adasco, R. G. Reitenbach, R. G. Raschke, and R. Stuhlmann, 1994: The ScaRaB project: Earth radiation budget observations from the meteor satellites. Adv. Space Res., 14, 47–54, https://doi.org/10.1016/0273-1177(94)90346-8.
Kooperman, G. J., M. S. Pritchard, M. A. Burt, M. D. Branson, and D. A. Randall, 2016: Robust effects of cloud superparameterization on simulated daily rainfall intensity statistics across multiple versions of the Community Earth System Model. J. Adv. Model. Earth Syst., 8, 140–165, https://doi.org/10.1002/2015MS000574.
Liou, K. N., 2002: An Introduction to Atmospheric Radiation. 2nd ed. International Geophysics Series, Vol. 84, Academic Press, 583 pp.
Loeb, N. G., S. Kato, K. Loukachine, N. Manalo-Smith, and D. R. Doelling, 2007: Angular distribution models for top-of-atmosphere radiative flux estimation from the Clouds and the Earth’s Radiant Energy System instrument on the Terra satellite. Part II: validation. J. Atmos. Oceanic Technol., 24, 564–584, https://doi.org/10.1175/JTECH1983.1.
Nesbitt, S. W., E. J. Zipser, and D. J. Cecil, 2000: A census of precipitation features in the tropics using TRMM: Radar, ice scattering, and lightning observations. J. Climate, 13, 4087–4106, https://doi.org/10.1175/1520-0442(2000)013<4087:ACOPFI>2.0.CO;2.
Palmer, T., and B. Stevens, 2019: The scientific challenge of understanding and estimating climate change. Proc. Natl. Acad. Sci. USA, 116, 242390–242395, https://doi.org/10.1073/pnas.1906691116.
Ramanathan, V., 1987: The role of Earth radiation budget studies in climate and general circulation research. J. Geophys. Res., 92, 4075–4095, https://doi.org/10.1029/JD092iD04p04075.
Ramanathan, V., and W. Collins, 1991: Thermodynamic regulation of ocean warming by cirrus clouds deduced from observations of the 1987 El Niño. Nature, 351, 27–32, https://doi.org/10.1038/351027a0.
Roca, R., and V. Ramanathan, 2000: Scale dependence of monsoonal convective systems over the Indian Ocean. J. Climate, 13, 1286–1298, https://doi.org/10.1175/1520-0442(2000)013<1286:SDOMCS>2.0.CO;2.
Roca, R., M. Viollier, L. Picon, and M. Desbois, 2002: A multisatellite analysis of deep convection and its moist environment over the Indian Ocean during the winter monsoon. J. Geophys. Res., 107, 8012, https://doi.org/10.1029/2000JD000040.
Roca, R., S. Louvet, L. Picon, and M. Desbois, 2005: A study of convective systems, water vapor and top of the atmosphere cloud radiative forcing over the Indian Ocean using INSAT-1B and ERBE data. Meteor. Atmos. Phys., 90, 49–65, https://doi.org/10.1007/s00703-004-0098-3.
Roca, R., and Coauthors, 2015: The Megha-Tropiques mission: A review after three years in orbit. Front. Earth Sci., 3, 17, https://doi.org/10.3389/feart.2015.00017.
Roca, R., T. Fiolleau, and D. Bouniol, 2017: A simple model of the life cycle of mesoscale convective systems cloud shield in the tropics. J. Climate, 30, 4283–4298, https://doi.org/10.1175/JCLI-D-16-0556.1.
Rutan, D., G. L. Smith, and T. Wong, 2014: Diurnal variations of albedo retrieved from Earth Radiation Budget Experiment measurements. J. Appl. Meteor. Climatol., 53, 2747–2760, https://doi.org/10.1175/JAMC-D-13-0119.1.
Smith, G. L., and D. A. Rutan, 2003: The diurnal cycle of outgoing longwave radiation from Earth Radiation Budget Experiment measurements. J. Atmos. Sci., 60, 1529–1542, https://doi.org/10.1175/2997.1.
Sohn, B.-J., M.-J. Choi, and J. Ryu, 2015: Explaining darker deep convective cloud over the western Pacific than over tropical continental convective regions. Atmos. Meas. Tech., 8, 4573–4583, https://doi.org/10.5194/amt-8-4573-2015.
Sokol, A. B., and D. L. Hartmann, 2020: Tropical anvil clouds: Radiative driving toward a preferred state. J. Geophys. Res. Atmos., 125, e2020JD033107, https://doi.org/10.1029/2020JD033107.
Trémas, T. L., O. Aznay, and O. Chomette, 2015: ScaRaB: First results of absolute and cross calibration. Proc. SPIE, 9643, 964304, https://doi.org/10.1117/12.2194866.
Vaillant de Guélis, T., H. Chepfer, V. Noël, R. Guzman, P. Dubuisson, D. M. Winker, and S. Kato, 2017: The link between outgoing longwave radiation and the altitude at which a spaceborne lidar beam is fully attenuated. Atmos. Meas. Tech., 10, 4659–4685, https://doi.org/10.5194/amt-10-4659-2017.
Viollier, M., C. Standfuss, O. Chomette, and A. Quesney, 2009: Top-of-atmosphere radiance-to-flux conversion in the SW domain for the ScaRaB-3 instrument on Megha-Tropiques. J. Atmos. Oceanic Technol., 26, 2161–2171, https://doi.org/10.1175/2009JTECHA1264.1.
Wall, C. J., D. Hartmann, M. M. Thieman, W. L. Smith Jr., and P. Minnis, 2018: The Life cycle of anvil clouds and the top-of-atmosphere radiation balance over the tropical west Pacific. J. Climate, 31, 10 059–10 080, https://doi.org/10.1175/JCLI-D-18-0154.1.
Wall, C. J., J. R. Norris, B. Gasparini, W. L. Smith Jr., M. M. Thieman, and O. Sourdeval, 2020: Observational evidence that radiative heating modifies the life cycle of tropical anvil clouds. J. Climate, 33, 8621–8640, https://doi.org/10.1175/JCLI-D-20-0204.1.
Webster, P. J., and G. L. Stephens, 1980: Tropical upper-tropospheric extended clouds: inferences from winter MONEX. J. Atmos. Sci., 37, 1521–1541, https://doi.org/10.1175/1520-0469-37.7.1521.
Wing, A. A., 2019: Self-aggregation of deep convection and its implications for climate. Curr. Climate Change Rep., 5, 1–11, https://doi.org/10.1007/s40641-019-00120-3.
Xu, K.-M., T. Wong, B. A. Wiekicki, L. Parker, and Z. A. Eitzen, 2005: Statistical analyses of satellite cloud object data from CERES. Part I: Methodology and preliminary results of the 1998 El Niño/2000 La Niña. J. Climate, 18, 2497–2514, https://doi.org/10.1175/JCLI3418.1.
Yang, G.-Y., and J. Slingo, 2001: The diurnal cycle in the tropics. Mon. Wea. Rev., 129, 784–801, https://doi.org/10.1175/1520-0493(2001)129<0784:TDCITT>2.0.CO;2.
Zelinka, M. D., and D. L. Hartmann, 2010: Why is longwave cloud feedback positive? J. Geophys. Res., 115, D16117, https://doi.org/10.1029/2010JD013817.
Zelinka, M. D., S. A. Klein, and D. L. Hartmann, 2012: Computing and partitioning cloud feedbacks using cloud property histograms. Part II: Attribution to changes in cloud amount, altitude, and optical depth. J. Climate, 25, 3736–3754, https://doi.org/10.1175/JCLI-D-11-00249.1.