Next Article in Journal
Forest Fire Effects on Landscape Snow Albedo Recovery and Decay
Next Article in Special Issue
CRAAS: A European Cloud Regime dAtAset Based on the CLAAS-2.1 Climate Data Record
Previous Article in Journal
Using PRISMA Hyperspectral Satellite Imagery and GIS Approaches for Soil Fertility Mapping (FertiMap) in Northern Morocco
Previous Article in Special Issue
Using GOES-R ABI Full-Disk Reflectance as a Calibration Source for the GOES Imager Visible Channels
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Remote Sensing
Volume 14
Issue 16
10.3390/rs14164077
remotesensing-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
Correction published on 2 April 2024, see Remote Sens. 2024, 16(7), 1257.
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Warming Trend and Cloud Responses over the Indochina Peninsula during Monsoon Transition

by
Mien-Tze Kueh
* and
Chuan-Yao Lin
Research Center for Environmental Changes, Academia Sinica, Taipei 11529, Taiwan
*
Author to whom correspondence should be addressed.
Remote Sens. 2022, 14(16), 4077; https://doi.org/10.3390/rs14164077
Submission received: 12 July 2022 / Revised: 6 August 2022 / Accepted: 16 August 2022 / Published: 20 August 2022 / Corrected: 2 April 2024
(This article belongs to the Special Issue Satellite-Based Cloud Climatologies)
Browse Figures
Review Reports Versions Notes

Abstract

An exacerbated precipitation–temperature relationship can lead to compound extremes. The role of clouds in such a relationship is relatively uncertain. Here, we investigate the cloud–precipitation–temperature relationships over the Indochina Peninsula during the summer monsoon transition. The negative correlation between cloudiness/precipitation and surface maximum temperature is valid on seasonal and interannual timescales. The near-surface temperature exhibits interdecadal variability and a long-term warming trend. The warming trend has accelerated in the past two decades. In the anomalous warm years, the remarkably strong western Pacific subtropical high inhibits the development of clouds, especially the middle and high cloud-top regimes, leading to the suppression of deep convection and precipitation. There are more optically thin (moderate to thick) clouds with smaller (larger) effective radii in the high cloud-top regime for the warm (cold) years. The dominance of shallow cumulus is a distinct feature in the warm years. The daytime heating of enhanced surface insolation due to decreased cloudiness is worsened by the dry condition of the precipitation deficit. The water vapor warming effect can prevent an efficient drop in nighttime temperature, thereby exacerbating the warm condition under the warming trend. The cloud–precipitation–temperature relationships coupling with the monsoon development can be used to diagnose the regional scale cloud–climate interactions in climate models.

1. Introduction

Near-surface air temperature (SAT) is a measure of the thermal state above the ground, which is a dynamic consequence of various processes in the energy and hydrologic cycles of the climate system, thereby providing an instrumental record of climate change and variability. In recent decades, numerous studies have investigated the correlation between SAT and a hydrological variable, such as atmospheric water vapor, precipitation, and clouds [1,2,3,4,5,6,7,8,9]. These studies generally interpreted the correlation of interest in terms of local thermodynamical processes and associates the correlation with dynamically driven factors at various spatial and temporal scales. A common message delivered by these studies is the implication of the correlation between SAT and a hydrological variable for the representativeness of climate change and variability over specific regions and for the use of such joint behavior in evaluating the underlying mechanisms in numerical models.
A robust positive relationship between near-surface specific humidity (q) and SAT has long been observed over the globe [1,2,3,4], reflecting the temperature dependence of saturated vapor pressure as described by the Clausius–Clapeyron relation. Significant increasing trends in SAT and q have been reported over the globe, with larger trends in the tropics [2,3,4], as well as larger trends at night than during the day over land [2]. The water vapor positive feedback, through the greenhouse warming effect, potentially contributes to the increasing occurrences of humid-heat events over humid and warm lands, as revealed by historical records and future projections [10,11,12,13,14]. The amplifying water vapor feedback could be modulated by precipitation [1,14]. In the humid and warm tropics, there are abundant cloudiness and precipitation. Positive (negative) associations are generally found between precipitation and SAT over tropical oceans (lands) [5,6,7]. On the interannual timescale, the El Niño-Southern Oscillation (ENSO) acts to amplify these tropical precipitation–SAT associations [5,6,15,16]. ENSO also exerts considerable influence on the global-scale q-Ta association, with warmer and moister (cooler and drier) conditions globally during El Niño (La Niña) events [4] and with considerable responses over the tropical Pacific and Indian Oceans [2]. Over the lands of Southeast Asia, the El Niño-related extreme temperature conditions have been characterized as humid-heat [13,14] and dry-hot [15,16] patterns, respectively, in the content of q-SAT and precipitation–SAT associations. The coexistence of the humid-heat and dry-hot can be interpreted as a consequence of amplifying water vapor feedback alongside the El Niño-related large-scale subsidence, which can lead to clear skies, enhanced surface insolation, tropospheric drying and precipitation deficit. Although soil moisture has been recognized as a vital factor in the negative precipitation–SAT relationship over lands where limited soil moisture can amplify the extreme heat condition [1,7,17,18], Southeast Asia is a moisture-rich region where greenhouse warming effects of water vapor and clouds are relatively crucial [19]. Previous studies using cloud cover reports from ground-based observations have revealed that in the daytime, a negative cloud–SAT relationship is generally observed over warm lands [1,8,9] and, specifically, clouds with low bases can effectively reflect the sunlight so as to reduce the rise of SAT [1]. In the nighttime, however, high humidity conditions can hinder the direct interaction between clouds and SAT [1,8,9].
The Indochina Peninsula (ICP) is the continental portion of Southeast Asia and is bordered by the Bay of Bengal (BoB) to the west and the South China Sea (SCS) to the east. Climatologically, the ICP experiences its seasonal peak temperature in April [20,21,22]. The temperature drops with increasing precipitation and convective activity in May, signifying the transitional stage of the ICP summer monsoon [23,24]. During the ICP monsoon transition, there is a strong negative precipitation–SAT relationship on both seasonal [22] and interannual scales [21,22]. On a broader scale, May is the transitional period of summer monsoons over the BoB–ICP–SCS sector, where the onset of monsoonal convection and heavy rainfall is accompanied by the eastward retreat of the western Pacific subtropical high (WPSH) and the establishment of a monsoon trough (MT), with the climatological monsoon onsets occurring in mid-May [23,24,25,26]. There are interannual and interdecadal variabilities in the timings of the monsoon onsets over this region, especially for the SCS monsoon [23,27,28,29,30,31,32]. During the monsoon transition, anomalous hot and dry (cold and wet) conditions over the ICP and late (early) SCS monsoon onsets tend to be associated with El Niño (La Niña) events [20,21,23,27,28,29,30,31,32,33,34].
An exacerbated negative precipitation–SAT relationship over warm lands can lead to the concurrence of compound extremes events [14,15,16,20,21,22]. However, the role of clouds in the negative precipitation–SAT relationship is relatively uncertain. Hence the purpose of this study is to explore the co-variability in SAT and cloud properties over the ICP on both seasonal and interannual timescales. We focus on the summer monsoon transition during which the negative precipitation–SAT relationship is strong, and the monsoon development exhibits substantial interannual variability. In this context, we can investigate the cloud–precipitation–SAT relationships coupling with the summer monsoon in Southeast Asia. The paper is organized as follows: Section 2 introduces the data and methods. Section 3 presents the analyses of variability and warming trend of SAT alongside the changes in precipitation and cloud properties. Section 4 gives the discussions on the warming trend and atmospheric moisture supply. Section 5 describes the main conclusions.

2. Materials and Methods

The daily maximum and minimum temperatures and daily precipitation data over the ICP are taken from the Climate Prediction Center (CPC) Global Temperature dataset and Global Unified Precipitation dataset [35], respectively. Both datasets are gridded analyses over global land with a 0.5° × 0.5° horizontal resolution and are over the 43-year period from 1979 to 2021. Meteorological parameters are obtained from the NCEP Climate Forecast System Reanalysis (CFSR) with a 0.5° × 0.5° resolution, and from 1979 to 2021 [36,37]. The seasonal Oceanic Niño Index (ONI), a 3-month running mean of anomalous SST anomalies in the Niño 3.4 region, is provided by NOAA/CPC. We analyze the smoothed ONI in this study, which is the 5-month running mean of the original ONI. The smoothed ONI values that exceed the thresholds of +0.5 and −0.5 °C are identified as El Niño and La Niña, respectively, which meet the NOAA’s operational definition of five consecutive 3-month running means for a warm/cold period of ENSO [38]. The smoothed ONI values of December-January-February (DJF) and February-March-April (FMA) are used for the preceding winter and spring ONI of the summer monsoon transition and termed as winter ONI and spring ONI, respectively.
Cloud properties are obtained from the Level-3 Moderate-Resolution Imaging Spectroradiometer (MODIS) Daily Global Product (Collection 061) at a horizontal resolution of 1° × 1° [39,40]. The Level-3 products contain data collected from the Terra platform (MOD08_D3) and Aqua platform (MYD08_D3). In this study, the retrievals of cloud mask cloud fraction (CF), cloud top pressure (CTP), cloud optical thickness (COT), and cloud effective radius (CER) from the two platforms that cover the timespan from 2003 to 2021 are used. In the Level-3 products, there are various categories of aggregation for every cloud product. We used three categories (daytime, nighttime, and daily) of the cloud top properties (CF, CTP) and the daytime COT and CER. Both COT and CER are daytime-only products. The joint histograms of CF, COT or CER binned against CTP, and COT binned against CER are analyzed to reveal the statistics of various cloud properties. These joint histograms are statistical summaries that are stored as pixel counts for each 1° grid cell. For every day, we collected the joint histogram of the grid cells located within the ICP region (shown in Figure 1a) to derive a newly joint histogram by summing the corresponding pixel counts within each histogram bin box. From these daily joint histograms for the ICP region, we can then derive multiday and multiyear joint histograms for statistical analysis. These joint histograms can be used to yield the corresponding joint probability specified for the ICP region and for a particular analysis time period. We adopted the concept of optical cloud type definition in terms of CTP and COT used in the ISCCP D-series datasets [41]. The details will be described in the next section.
To address the meteorological history of the airflows that reach the ICP region, we analyzed the backward air parcel trajectories generated using the Lagrangian particle dispersion model FLEXPART v9.2 [42]. FLEXPART takes into account the grid-scale advection as well as the turbulent and convective transport of particles (the air parcels in this study) for the calculation of trajectories either backward or forward in time [43]. Numerous studies have been conducted using FLEXPART to evaluate the atmospheric moisture sources and sinks as well as the air mass pathways for a wide range of meteorological applications [44,45,46]. We employed the CFSR reanalysis dataset to calculate an ensemble of backward trajectories whose initial positions are distributed within a thin layer, from 1000 to 2000 m above mean sea level and over the ICP region, as denoted by the box in Figure 1. The backward trajectories are initialized on a 1° × 1° grid with 81 grid cells within the ICP region, where 50 parcels are released in every grid cell at 0000 UTC for each individual day in May of every year from 2003 to 2021. For each day, a total of 4050 parcels are released and tracked backward in time for 120 h (5 days) or until they either touch down or leave the calculation domain (10°S–25°N, 70°–130°E). Output variables indicating the position (latitude, longitude, and altitude) and meteorological parameters, including air temperature and specific humidity, of each parcel were recorded every 6 h.
The analysis period is from late April to early June; May is the monsoon transitional period. The seasonal variations are examined on the basis of the daily and 15-day time periods. For each year, the timespan from 1 April to 29 June was divided into six segments of the 15-day time period. The interannual variations are revealed by analyzing a collection of warm and cold years from 2003 to 2021. The definition of the warm and cold years will be detailed in the next section.

3. Results

3.1. Precipitation–Temperature Relationship and Warming Trend

In early summer (May and June), the seasonal variation in large-scale atmospheric patterns over the BoB–ICP–SCS sector is characterized by the eastward retreat of WPSH and the establishment of MT [23,24,25,26]. Here, we used the 5870-gpm (1490-gpm) contours to indicate the western flank (southern edge) of the climatological mean WPSH (MT) (Figure 1a). The western flank of WPSH retreats from the ICP in May and out of SCS in June. The developing and fully established stages of MT are revealed by the southward progression and further expansion of the 1490-gpm contours, respectively, in May and June. Meanwhile, the regional conditions over the ICP are undergoing a process of transition. From March to April, both the daily maximum and minimum SAT (Tmax and Tmin) over the ICP region rose and reached 34 and 24 °C, respectively, in late April (Figure 1b). From May to June, Tmax gradually decreases, whereas Tmin levels off at the higher level. The discrepancy in their changing rates results in a gradual reduction in the diurnal temperature range (DTR) (Figure 1c). May is the crucial period with an apparent increase in precipitation amount and the increase in low-level westerlies, signifying the development of the summer monsoon. Although these variables exhibit apparent interannual variabilities, their patterns in seasonality remain clear. Notably, there is a negative association between the decreasing Tmax and the increased precipitation. Figure 2 reveals substantial interannual variabilities in the Tmax and Tmin on a daily basis. Nonetheless, the climatological patterns of seasonal advance, namely their warming stages prior to May and the decreasing (leveling off) stage for Tmax (Tmin) in May, remain visible. We focus on May to reveal the interannual variability in the monsoon transition. The 43-year records of Tmax exhibit an interdecadal variation; there are two relatively warm periods with a colder period in between, spanning approximately from 1999 to 2009. The second warm period appears to be slightly warmer than the first one, implying a long-term warming trend in the data. Tmin exhibits a relatively clear long-term warming trend, indicating an increase in overnight temperatures. The time series and linear trends of monthly Tmax and Tmin of May are given in Appendix A. The short-term interannual variability in Tmin is somewhat ambiguous, as revealed in Figure 2b, although a certain degree of temporal association between Tmax and Tmin can be observed during the period from mid-April to May. Focusing on May, we calculated the standardized monthly anomalies of Tmax and Tmin by dividing the monthly anomaly (the departures from normal) by their respective standard deviation of the monthly means. The standardization could eliminate significant differences in magnitude between Tmax and Tmin, and the resulting time series reveals highly similar variations (Figure not shown). The correction coefficient between the monthly averaged Tmax and Tmin of May is 0.89, suggesting a co-variability on the interannual scale. This also implies that there is a climatic driver, on top of seasonal and diurnal forcings, for this co-variability.
Figure 3a shows the standardized monthly anomalies of Tmax, precipitation, cloud fraction, and zonal wind component for May of 1979–2021. The correlation coefficients between Tmax and precipitation, cloud fraction, and zonal wind are −0.64, −0.79, and −0.2, respectively. On the interannual scale, Tmax is negatively associated with both precipitation and cloudiness. Tmax (precipitation) is positively (negatively) associated with the preceding winter ONI; this is consistent with earlier studies showing that the ICP tends to have wet and cold (dry and hot) conditions following a winter La Niña (El Niño) state [20,21,23,31,32,33,34]. The distribution of the years of Tmax above and below normal provide useful clues for the warming trend. There are 21 positive and 22 negative values of Tmax rather evenly distributed across the 43-year period. Considering that a value of the standardized anomaly exceeds +1 or below −1 as a warm or cold year, respectively, there are 11 warm years and 6 cold years in the period. For the warm years, 8 out of 11 are found in the 21-year period after the year 2000. The cold years, however, are evenly distributed prior to and after the year 2000. The result confirms the long-term warming trend in Tmax. We further modified the threshold of a cold year to be −0.5 and identified 15 cold years from the data; this allows us to have comparable sample sizes to compare the difference between the warm and cold years. The resulting warm and cold years are denoted as red and blue filled rectangles in Figure 3a. A similar result is found from the standardized monthly Tmin anomalies. Considering the data period of the MODIS dataset, we selected eight warm years and eight cold years after the year 2003 (Table 1). The warm and cold years are loosely associated with the ONI of the preceding seasons. We will address this in the discussion section. The comparison between warm and cold years will be carried out by using the composites of warm and cold years. Hereafter we term the composites as warm years and cold years for brevity.
Figure 3b–e show the daily time series of various atmospheric variables composited with respect to the warm and cold years. The seasonal trends of these variables found in the warm and cold years are generally similar, but the exact magnitudes are different. Both the Tmax and Tmin in the warm years are apparently warmer than that in the cold years for the period from April to June, with larger differences found in May. In addition, the Tmax and Tmin in the warm years are also warmer than their respective climatology, as shown in Figure 1b, whereas that of the cold years are comparable to the climatology. In the cold years, larger decreasing rates are found in the first half of May as compared to the warm years. This results in larger decreases in the diurnal temperature range in May of the cold years. The differences in SAT (Tmax and Tmin) between the warm and cold years suggest a delayed or moderate monsoon development in the warm years. The monsoon transition is also the beginning of the rainy season of ICP; given the negative precipitation–temperature relationship, the reductions in precipitation and cloudiness found in May of the warm years are supportive evidence for this argument. We will further address the differences in monsoon development in the discussion section. However, no apparent difference can be observed in the time series of the zonal wind component, which is a key element of the summer monsoonal flow over the BoB–ICP–SCS sector [23,24,25,26]. We will address this in the discussion section. In short, the composites can demonstrate the differences between the warm and cold years in terms of the Tmax, Tmin, cloudiness, and precipitation, as well as the negative cloudiness/precipitation–Tmax relationship. We also assess the statistical significance of the differences between the warm and cold composites for each calendar day (Supplementary Material Table S1). The results reveal that the calendar days that are statistically significant are generally consistent with the days in their region, and the standard errors of the sample means are well apart from each other in Figure 3. In the second half of the 43-year period, the warming trend is evident. During this latter period, seven out of eight cold years are found before 2012, and six out of eight warm years are found after 2014. The differences between the warm and cold years are thus indicative of the changes with respect to the warming trend.
The negative cloudiness/precipitation–Tmax relationship is valid on both seasonal and interannual scales, as well as under the warming trend. A negative relationship is also valid for Tmin on an interannual scale but not on a seasonal scale. Taken in combination, we can have a straightforward interpretation of the negative cloudiness/precipitation–Tmax relationship. In March and April, the ICP region is mostly under the control of WPSH, where large-scale subsidence conditions and a clear sky dominate. With such conditions, cloudiness and precipitation are commonly suppressed, thereby leading to high SAT in the daytime (thus an increased likelihood of high Tmax) owing to enhanced surface insolation. On the other hand, the retreat of WPSH and progression of MT starting in May are favorable for intensive cloudiness and precipitation, where a considerable reduction in daytime solar heating and supplemental soil moisture and raindrop-related evaporative cooling can lead to a substantial drop in Tmax. Nighttime cloudiness–SAT association operates differently, where cloudy nights tend to be comparatively warmer than cloudless nights because of the insulating effect of clouds on outgoing infrared emissions.

3.2. Cloud Responses

From mid-April to mid-June, the cloudiness over the ICP and vicinity increased substantially (Figure 4). The largest increasing rate of CF occurs between the first half and the second half of May. Although similar patterns of seasonal development are found among different composites, more and less CF is observed in cold and warm years, respectively, as compared to the climatological CF. The longitudinal distributions also reveal a substantial reduction of CF in warm years. The nighttime CF is generally larger than the daytime CF in both composites. The magnitudes of the differences between the warm and cold years are apparently larger than that of the diurnal range, namely the differences between daytime and nighttime from either warm or cold years prior to the second half of May. Although increased cloudiness can generally suppress nighttime cooling of the SAT [1,8,9], it is not the case in our study. We have a reduction in cloudiness and suppression of nighttime cooling in the warm years as compared to the cold years.
The vertical distribution of clouds is further examined using the joint histograms of cloud optical properties COT and CER, respectively, against the CTP. Here we followed the COT/CTP-based cloud classification defined in the ISCCP [41] with marginal modification. According to the ISCCP cloud classification, clouds are grouped into three cloud-top regimes: high clouds with CTP < 440 hPa, middle clouds with 440 hPa ≤ CTP< 680 hPa, and low clouds with CTP ≥ 680 hPa. Every cloud-top regime is further classified into three categories of thin, moderate, and thick optical thicknesses, respectively, using the COT thresholds of below 3.6, between 3.6 and 23, and above 23. The ISCCP cloud classification is used for classifying nine optical cloud types with respect to the morphological cloud types. From low to middle and high cloud regimes, those of thin optical thicknesses are cumulus, altocumulus, and cirrus, those of moderate optical thicknesses are stratocumulus, altostratus, and cirrostratus, and those of thick optical thicknesses are stratus, nimbostratus, and deep convection. The joint histogram bins of COT against CTP in the MODIS dataset are slightly different from that in the ISCCP. We followed the CTP and COT settings of our previous study [47], where the resulting optical cloud types have been compared with ground-based cloud observations and proven to be appropriate for revealing the dependence of springtime clouds on synoptic weather patterns and near-surface conditions. Accordingly, the ranges of CTP are 50–450, 450–700, and 700–1100 hPa for high, middle, and low cloud regimes. The ranges of COT are 0–4, 4–20, and 20–150 for thin, moderate, and thick optical thicknesses. For ice phase clouds, the results of comparison between the warm and cold years are similar to that of the high regime in liquid water clouds. Therefore, only the results of liquid water clouds are shown.
The joint probability of COT versus CTP of four 15-day statistical periods analyzed for the period from 2003 to 2021 is displayed with three types of histogram plots in Figure 5. We will take advantage of each histogram display to interpret the data distribution. First, we examine the total counts of these 15-day periods (the numbers in the square brackets). All the total counts are more than 2.5 million and are used as the denominators for computing the joint probability (relative frequency) of the respective 15-day periods. The total counts increase with time, indicating that the number of detected water clouds increases as the season advances, where the largest increment of 11.9% emerges between the first half and second half of May. Given that the numbers of the denominator for computing the joint probability of every 15-day period are different, we shall concentrate on the changes in the proportions of various cloud regimes or cloud types across the analyzed period.
The two-dimensional density histogram depicts the detailed distributions of COT against CTP in the context of optical cloud types (Figure 5a). The bimodal pattern with respect to CTP shown in all 15-day periods reveals that the whole analyzed period is dominated by larger proportions of both high and low cloud regimes. On the other hand, the unimodal pattern with respect to COT, which is generally found during the whole period, reveals the dominance of thin to moderate optical thicknesses. The whole period is mainly characterized by a mixture of cirriform clouds (high regime) and cumulus and stratocumulus (low regime), where the majority of cirrostratus and stratocumulus are optically thin within the COT range of the moderate category. The monsoon transition is featured by the changes in the proportions of the cloud regimes/types. Briefly, the proportion of the high regime is increasing whereas the proportion of the low regime is declining during the period from 16 April to 14 June. The change in the middle regime is marginal. Although the dominant cloud types remain to be cirrus-cirrostratus and cumulus-stratocumulus across the whole period, the increasing and decreasing proportions of deep convection and stratus, respectively, over time are of interest because they signify the fundamental change in the monsoon transition. The increase in the occurrence of deep convection is consistent with the substantial increase in the rainfall amount from late April to May (Figure 1c).
We use the marginal probability of CTP with respect to the three COT categories designed for optical cloud classification to summarize the distributions and changes demonstrated by the two-dimensional density histograms. The relative frequency, computed by dividing the count in each bin by the total count, is shown for the marginal probabilities. While this can lead to some visual distortion due to the uneven intervals of the CTP histogram bins, the bimodal distribution is still valid owing to the well-separated double-peak pattern. The bimodal pattern showing the dominance of high and low regimes and their increasing and decreasing proportions during the statistical period, respectively, are clearly presented in the stacked histogram display (Figure 5b). The generally comparable proportions of thin and moderate COT categories, as well as the substantially small proportions of thick COT, for every cloud regime also exhibit the major features depicted by the two-dimensional display. The marginal probability values in each CTP bin can be roughly inspected from the chart. For example, taking the combined contributions of all COT categories, the summed probability value of the top bin for the period of 16 April to 30 April is approximately 0.165, denoting a proportion of 16.5%. A close inspection reveals that the proportions of the high regime (50–450 hPa) increase approximately from 30% to 50% during the period from 16 April to 30 May, whereas that of the low regime (700–1100 hPa) decreases approximately from 50% to 35%. Because satellites detect the cloud layer from space, lower clouds can be hidden by higher clouds when multi-deck clouds exist. Thus, a portion of the proportional decline in the low regime may be due to the increase in the high regime.
The marginal probabilities of CTP can provide further information on the vertical structure of every cloud-top regime as compared to the nine cloud types identified by the ISCCP cloud classification because the original CTP histogram bin intervals have been retained. Although the three sets of the marginal probability of CTP are bimodally distributed, their shapes are obviously different (Figure 5c). During the transition, all COT categories in the high regime are characterized by increases in proportions, with their largest increments in the top bin where the moderate COT (cirrostratus) increases the most. Both the thin and moderate COT categories in the low regime decrease with time. The thin COT category has the largest decline in the CTP bin of 900–1000 hPa, with the upward decreasing pattern remaining similar, showing a prevalence of shallow cumulus. The moderate COT category exhibits substantial declines in the CTP bins of 800–1000 hPa, along with a structural change from approximately uniform to the upward increasing pattern in the vertical, suggesting a rising trend in the cloud top of stratocumulus. Overall, the combined contribution from all COT categories leads to a structural change of the low regime, namely from an upward decreasing pattern to a relatively uniform pattern (Figure 5b). Finally, the increase (decrease) in the thick COT in the CTP bin of 50–250 hPa (700–800 hPa) clearly depicts the development (decline) of deep convection (stratus) during the transition.
For the warm and cold years, the dominant patterns of cloud regimes and cloud types and the major features of seasonal evolution as displayed in two-dimensional histograms are visually similar to the long-term probability distribution. We can proceed to make the cloud-type comparison between the warm and cold years using the simplified histogram displays of the joint probability (Figure 6). Because our focus is May, and the difference between the second half of May and the first half of June is marginal, the results for June will be omitted. Similar to the long-term statistical distributions, the total counts in both warm and cold years increase as the season advances. However, the total counts in the warm years are smaller than that in the same periods of cold years, suggesting that the amount of detected water clouds are lower in the warm years. The vertical structures of the low regime are distinct in May of warm and cold years, whereas their decreasing patterns of seasonal evolution are both generally similar to the long-term statistical distribution (Figure 6a,b). The shallow cumulus is more prevalent in warm years, thereby leading to the distinct vertical structures of the low regime. The thin COT category in the high regime has a larger increment in the warm years, whereas the moderate to thick COT categories in the middle–high regimes have a higher amount in cold years. Because the total counts (denominators) are larger in cold years, therefore, their larger proportions in specific histogram bins can ensure a higher cloud amount. We further examine the vertical distributions of a cloud optical radius using the marginal probability of CTP with respect to CER, where the CER histogram bins are divided into three categories: small, intermediate, and large sizes (Figure 6c,d). Visually, the proportions of small CER in May of warm years are relatively larger than that of cold years. Overall, for the low cloud regimes in both warm and cold years, the intermediate CER has the largest proportion, whereas the small CER has the least. In warm years, notably for the shallow clouds, the intermediate CER shows a substantial decline, and the small CER has a larger proportion compared to the cold years. In addition, the high regime in the warm years exhibits larger increments of small to intermediate CER, whereas the middle to high regimes in the cold years tends to have more intermediate to large CER.
Figure 7 shows the statistical association between the COT and CER using the marginal probability of CER with respect to the three COT categories. Note that the bimodal distribution of the CER shown is a visual distortion due to the uneven intervals of the CER histogram bins. May of warm (cold) years tends to have a larger proportion of small (large) CER. For the thin COT category, there are increases (decreases) in the smaller CER in the warm (cold) years and very few changes in the larger CER for both. The moderate to thick COT categories show similar time evolution, both with increases in larger CER and decreases in smaller CER. Although the warm and cold years show comparable increments during the transition, the cold years have higher amounts of cloud with larger CER. Considering the results from Figure 6, for the moderate to thick COT categories, the increases of larger CER are more likely to occur in the high regime, whereas the decreases of smaller CER are likely to occur in the low regime. Regarding the changes of smaller CER in the thin COT category, the declines in cold years and increases in warm years are likely to occur in low and high regimes, respectively. Taken in combination, the middle and high cloud regimes are suppressed in the warm years. Even though the proportion of high regime increases during the transition, a large proportion of optically thin clouds with smaller cloud drops are observed, but fewer deep convections occur. The relatively less amount of high clouds in the warm years could favor low cloud detection. Nonetheless, the dominance of shallow cumulus is a distinct feature in the warm years.

4. Discussion

4.1. Drivers of the Warming Trend

The WPSH in May of the warm and cold years are distinct. The WPSH in May of warm years is remarkably strong, such that most areas of ICP and SCS are enclosed by the 5880-gpm contours (Figure 8a). The WPSH in May of cold years is relatively weak, as denoted by the 5870-gpm contours, and shows a sequence of eastward retreats from the ICP and SCS. In the warm years, the MT exhibits a weak or delayed development with a generally belated southward advance of the 1490-gpm contours (Figure 8b), even though the low-level winds over the ICP are indistinguishable from the cold years (Figure 3c). The remarkably strong WPSH can lead to the suppression of convective clouds as well as precipitation, which, in turn, contributes to a weak or delayed MT development. The result reveals the cloud changes over the ICP involving large-scale monsoonal development. We further examined the cloud changes over the BoB and SCS in May (the regions denoted in Figure 8b).
Figure 9 shows the marginal probabilities of CTP for warm and cold years, as in Figure 6a but with a further reduced dimension of CTP. The first impression is the higher amounts of cloud pixels over the land (ICP) than over the oceans (BoB and SCS), as indicated by the total counts, consistent with Figure 4. The percentage reductions of cloud amounts in warm years with respect to cold years are 8.5%, 13.9%, and 43.2%, respectively, for BoB, ICP, and SCS. The suppression of clouds in SCS is stronger than that in ICP. There is a difference in the combinations of cloud regimes over the ICP and the adjacent waters. For the oceanic regions, there are always more low clouds detected, irrespective of warm or cold years. The cloud response in SCS is similar to ICP, namely the decreased (increased) proportion of the high (low) regime, as well as less deep convection and more cumulus in the warm years. However, the BoB exhibits an increase in high clouds and a decrease in low clouds in the warm years. Enhanced high cloud regimes over the BoB are expected. For the BoB, the development of deep convection is more active in ICP warm years, in consistent with previous studies for late SCS monsoon onset years in the past decade [28,30]. We also analyzed the convective available potential energy (CAPE) over the BoB, ICP, and SCS (Figure 10). There are land–sea contrasts between the ICP and BoB and SCS, specifically the large day–night differences in the ICP, but the differences between the warm and cold years are thermodynamically consistent with the previous results. Over the ICP and SCS, CAPE is larger in warm years. The pattern is reversed in BoB. In warm years, near-surface hot and humid conditions would result in high CAPE, but the strong WPSH can dynamically suppress the development of deep convection. The smaller CAPE over the BoB in warm years can be interpreted as a negative relationship between CAPE and convection and precipitation, where near-surface cooling due to cold convective wake regions in the boundary layer can reduce the CAPE values [48]. In short, the CAPE analysis is supportive of our results.
The remarkably strong WPSH in May of the warm years is indicative of a late SCS monsoon onset. Earlier studies have revealed the El Niño dependence of the precipitation deficit and anomalous warm conditions over the ICP and the late SCS monsoon onsets and ascribed the enhanced WPSH to the anomalous subsidence over the Maritime Continent generated by the El Niño conditions in the preceding winters [20,21,23,27,28,29,30,31,32,33,34]. The suppressed cloudiness over the ICP and SCS in the warm years (Figure 9) is supportive of the enhanced WPSH [27,28,29,30,31,32,33,34]. However, the warm and cold years in our study are loosely associated with the ENSO states (Table 1). On an interdecadal scale, teleconnection patterns of ENSO could be modulated by the tropical sea surface temperature anomalies over the Pacific and Indian Oceans [27,28,29,30,31,32,33,34]. Previous studies have indicated a trend of dry and warm anomalies over the ICP as well as late SCS monsoon onsets during the past two decades, where their El Niño dependences have been weakened [28,29,30]. Here, we also find weaker ENSO dependences of the SAT and 500-hPa geopotential height over the ICP in May of the second half of the 43-year period (Table 2). The ENSO dependences of precipitation and cloud fraction are weaker than that of Tmax and Tmin, and there is no interdecadal difference in the precipitation. Overall, the warming trend in the SAT over the ICP is a combination of interdecadal variation and long-term trends.

4.2. Role of Moisture Content in Nighttime Warming

Over humid and warm lands, the greenhouse warming effect of atmospheric water vapor on nighttime SAT can hinder the cloud effect [1,8,9]. Here we analyze the Lagrangian history of the air parcels bound for the ICP to reveal the major moisture pathways. The first impression is the wide spread of trajectories over the majority of the domain as backward in time (Figure 11). A comparison of the numbers of trajectories remaining in the domain (numbers in the square brackets) reveals that more trajectories are trapped within the domain in the warm years, even at the backward time of 24 h, suggesting that air parcels are relatively slower in reaching their destination. The seasonal advance is characterized by the major air mass pathways from the east and west, respectively, during the first half and second half of May. However, the trajectories in the warm years are more concentrated in the adjacent regions of ICP, with a higher proportion of air parcels from the east. In the warm years, there are more trajectories from above the layer of destination (1 to 2 km), revealing the ambient subsidence under the control of the remarkably strong WPSH (Figure 11e,f). For the cold years, more trajectories are found below 2 km at the backward time of 24 h, suggesting a relatively convective condition over the ICP.
Because a high proportion of the trajectories travel across the BoB or SCS before reaching their destinations, the differences in moisture supplies bound for the ICP regions between the warm and cold years are of interest. The probability distributions of the specific humidity carried by the backward trajectories for the first half and second half of May are presented in Figure 12 (left panels). The specific humidity and temperature generally decrease with height. Therefore, we further stratify the specific humidity distribution into three marginal probabilities with respect to the temperature of trajectories (central and right panels). The distribution at the backward time of 72 h shows a dry shift as compared to that of 24 h, which is presumably arising from the higher trajectory altitudes at 72 h (left and central panels). The distributions of warm years show modest dry shifts in general but with relatively higher proportions of warmer (≥25 °C) and moister (≥17 g/kg) air parcels. There is no evidence of water vapor deficit in the warm years, suggesting that the reduction in cloudiness and precipitation primarily arises from the subsidence of the remarkably strong WPSH. Given the high humidity in the region, the radiative cooling owing to the suppression of cloudiness could be compensated by the greenhouse warming effect of water vapor, thereby preventing an efficient drop in nighttime temperature. Notably, the high Tmin in warm years peaks around early May and plateaus through the transitional period (Figure 3b), signifying that the anomalous warm condition could be worsened by the nighttime water vapor warming effect. On the seasonal timescale, this water vapor warming effect can also contribute to the leveling off of Tmin in May and June, along with the cooling effects of increased precipitation and cloudiness on the Tmax, leading to reduced DTR. Therefore, the high similarity of the interannual variabilities in Tmax and Tmin is presumably a consequence of the reduced DTR during the transition.
The high Tmin in warm years is indicative of exacerbated warm conditions under the warming trend. In recent decades, there has been an increasing trend of humid-heat events over the ICP [12,13,14]. While humid-heat events are generally accompanied by persistent subsidence conditions that lead to tropospheric drying and precipitation deficit, the near-surface humidity remains high [11,14]. Under climate warming, extreme heat events would be amplified by cloud and water vapor feedbacks [10,11,12,14]. The compound extreme conditions exacerbated by local thermodynamical processes can have an impact on remote regions, for example, the springtime high temperature alongside dryer soil moisture over the ICP has been related to heavy summer rainfall in East Asia [17,18]. The study of the underlying mechanisms of the joint behaviors of SAT and various hydrological variables, including humidity, precipitation, and clouds, will further our understanding of the weather and climate extremes. The proper representation of clouds in climate models remains a challenge. The use of the cloud regimes based on ISCCP COT-CTP joint histograms [49,50], as well as the associations of cloud cover with SAT [51] and precipitation [52], for the diagnosis of the clouds in climate models, can provide crucial information for model improvements. In this context, the cloud regimes based on COT-CTP coupling with various dynamical and thermodynamical conditions, in our case, the monsoon development and negative precipitation-Tmax relationships, respectively, can be used to diagnose the regional scale cloud–climate interactions in climate models.

5. Conclusions

In this study, we investigate the cloud–precipitation–SAT relationships coupling with the summer monsoon over the ICP on seasonal and interannual scales. We focus on May, the period of summer monsoon transition.
Over the ICP, Tmax reaches its seasonal peak in late April. From May to June, a decrease in Tmax is associated with an increase in precipitation and cloudiness. The study period is mainly characterized by optically thin cirrus-cirrostratus and cumulus-stratocumulus. The monsoon transition is featured by increasing high cloud-tip regimes, including optically thick deep convection.
During the 43-year period from 1979 to 2021, Tmax exhibits interannual and interdecadal variabilities, along with a long-term warming trend. The negative cloudiness/precipitation–Tmax relationship is valid on both seasonal and interannual timescales. There is an ENSO dependence of Tmax and a negative relationship, but the dependences have been weakened in the past two decades, during which the warming trend has accelerated. During the period from 2003–2021, strong (weak) WPSH dominates the anomalous warm (cold) years. The remarkably strong WPSH inhibits the development of clouds, especially the middle and high cloud-top regimes, leading to the suppression of deep convection and precipitation. There are more optically thin (moderate to thick) clouds with smaller (larger) effective radii in the high cloud-top regime in the warm (cold) years. The relatively fewer high clouds in the warm years could favor low cloud detection. Nonetheless, the dominance of shallow cumulus is a distinct feature in warm years.
Over a humid region such as ICP where the negative precipitation–Tmax relationship is strong, daytime heating of the enhanced surface insolation is due to decreased cloudiness is worsened by the dry condition of precipitation deficit, whereas reduced surface insolation due to increased cloudiness is conducive to the cooling effect of precipitation. Lastly, the negative cloudiness/precipitation–Tmin relationship is only valid on the interannual scale. This is because, in a humid atmosphere, the water vapor warming effect can prevent an efficient drop in nighttime temperature, resulting in small DTR and high similarity of the interannual variabilities in Tmax and Tmin. The high Tmin in warm years is indicative of exacerbated warm conditions under the warming trend.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/2072-4292/14/16/4077/s1, Table S1: The statistical significance of the differences between the means of warm and cold years.

Author Contributions

Conceptualization, methodology, software, validation, formal analysis, and investigation, M.-T.K.; resources, M.-T.K. and C.-Y.L.; data curation, M.-T.K.; writing—original draft preparation, M.-T.K.; writing—review and editing, M.-T.K. and C.-Y.L.; visualization, M.-T.K.; supervision, M.-T.K. and C.-Y.L.; project administration, C.-Y.L.; funding acquisition, C.-Y.L. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Ministry of Science and Technology, Taiwan, grant numbers 109-2111-M-001-004 and 110-2111-M-001-013.

Data Availability Statement

The CPC Global Temperature dataset and Global Unified Precipitation dataset are provided by the NOAA/OAR/ESRL PSL, Boulder, Colorado, USA, and can be accessed from their Websites at https://psl.noaa.gov/data/gridded/data.cpc.globaltemp.html (accessed on 10 April 2022) for the temperature dataset and at https://psl.noaa.gov/data/gridded/data.cpc.globalprecip.html (accessed on 10 April 2022) for the precipitation dataset. The NOAA/CPC seasonal ONI values since 1950 are publicly available at https://www.cpc.ncep.noaa.gov/data/indices/oni.ascii.txt (accessed on 1 June 2022). The CFSR reanalysis datasets can be accessed from the Research Data Archive (RDA), managed by the National Center for Atmospheric Research at https://rda.ucar.edu/ (accessed on 12 April 2022) [53,54]. The Collection 061 Level-3 MODIS Atmosphere Daily Global Product can be accessed at https://ladsweb.modaps.eosdis.nasa.gov/missions-and-measurements/science-domain/l3-atmosphere/ (accessed on 26 March 2022).

Acknowledgments

The first author thanks Kun-Cheng Lee for all the assistance on the FLEXPART Lagrangian simulations. We appreciate the anonymous reviewers for their valuable comments and suggestions.

Conflicts of Interest

The authors declare no conflict of interest.

Appendix A

The variabilities in monthly Tmax and Tmin of May are similar, both exhibit long-term warming trend. The warming trends are significant during the second half of the 43-year period.
Remotesensing 14 04077 g0a1
Figure A1. The linear trends of monthly Tmax and Tmin of May for the periods (left) 1979–2021 and (right) 1979–2000 and 2000–2021.
Figure A1. The linear trends of monthly Tmax and Tmin of May for the periods (left) 1979–2021 and (right) 1979–2000 and 2000–2021.
Remotesensing 14 04077 g0a1

References

  1. Dai, A.; Trenberth, K.E.; Karl, T.R. Effects of clouds, soil moisture, precipitation, and water vapor on diurnal temperature range. J. Clim. 1999, 12, 2451–2473. [Google Scholar] [CrossRef] [Scilit]
  2. Dai, A. Recent climatology, variability, and trends in global surface humidity. J. Clim. 2006, 19, 3589–3606. [Google Scholar] [CrossRef] [Scilit]
  3. Willett, K.M.; Dunn, R.J.H.; Thorne, P.W.; Bell, S.; de Podesta, M.; Parker, D.E.; Jones, P.D.; Williams, C.N., Jr. HadISDH land surface multi-variable humidity and temperature record for climate monitoring. Clim. Past 2014, 10, 1983–2006. [Google Scholar] [CrossRef] [Scilit]
  4. Allan, R.P.; Willett, K.M.; John, V.O.; Trent, T. Global changes in water vapor 1979–2020. J. Geophys. Res. 2022, 127, e2022JD036728. [Google Scholar] [CrossRef] [Scilit]
  5. Trenberth, K.E.; Shea, D.J. Relationships between precipitation and surface temperature. Geophys. Res. Lett. 2005, 32, L14703. [Google Scholar] [CrossRef] [Scilit]
  6. Adler, R.F.; Gu, G.; Wang, J.J.; Huffman, G.J.; Curtis, S.; Bolvin, D. Relationships between global precipitation and surface temperature on interannual and longer timescales (1979–2006). J. Geophys. Res. 2008, 113, D22104. [Google Scholar] [CrossRef] [Scilit]
  7. Berg, A.; Lintner, B.R.; Findell, K.; Seneviratne, S.I.; van den Hurk, B.; Ducharne, A.; Chéruy, F.; Hagemann, S.; Lawrence, D.M.; Malyshev, S.; et al. Interannual coupling between summertime surface temperature and precipitation over land: Processes and implications for climate change. J. Clim. 2015, 28, 1308–1328. [Google Scholar] [CrossRef] [Scilit]
  8. Groisman, P.Y.; Bradley, R.S.; Sun, B. The relationship of cloud cover to near-surface temperature and humidity: Comparison of GCM simulations with empirical data. J. Clim. 2000, 13, 1858–1878. [Google Scholar] [CrossRef] [Scilit]
  9. Sun, B.; Groisman, P.Y.; Bradley, R.S.; Keiming, F. Temporal changes in the observed relationship between cloud cover and surface air temperature. J. Clim. 2000, 13, 4341–4357. [Google Scholar] [CrossRef] [Scilit]
  10. Fischer, E.; Knutti, R. Robust projections of combined humidity and temperature extremes. Nat. Clim. Chang. 2013, 3, 126–130. [Google Scholar] [CrossRef] [Scilit]
  11. Kueh, M.T.; Lin, C.Y.; Chuang, Y.J.; Sheng, Y.F.; Chien, Y.Y. Climate variability of heat waves and their associated diurnal temperature range variations in Taiwan. Environ. Res. Lett. 2017, 12, 074017. [Google Scholar] [CrossRef] [Scilit]
  12. Russo, S.; Sillmann, J.; Sterl, A. Humid heat waves at different warming levels. Sci. Rep. 2017, 70, 7477. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  13. Li, X.X. Heat wave trends in Southeast Asia during 1979–2018: The impact of humidity. Sci. Total Environ. 2020, 721, 137664. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  14. Ha, K.J.; Seo, Y.W.; Yeo, J.H.; Timmermann, A.; Chung, E.S.; Franzke, C.L.E.; Chan, J.C.L.; Yeh, S.W.; Ting, M. Dynamics and characteristics of dry and moist heatwaves over East Asia. npj Clim. Atmos. Sci. 2022, 5, 49. [Google Scholar] [CrossRef] [Scilit]
  15. Hao, Z.; Zhang, X.; Singh, V.P.; Hao, F. Joint modeling of precipitation and temperature under influences of El Niño Southern Oscillation for compound event evaluation and prediction. Atmos. Res. 2020, 245, 105090. [Google Scholar] [CrossRef] [Scilit]
  16. Feng, S.; Hao, Z. Quantitative contribution of ENSO to precipitation-temperature dependence and associated compound dry and hot events. Atmos. Res. 2021, 260, 105695. [Google Scholar] [CrossRef] [Scilit]
  17. Gao, C.; Li, G.; Xu, B.; Li, X. Effect of spring soil moisture over the Indo-China Peninsula on the following summer extreme precipitation events over the Yangtze River basin. Clim. Dyn. 2020, 54, 3845–3861. [Google Scholar] [CrossRef] [Scilit]
  18. Gao, C.; Chen, H.; Chen, H.; Yan, H. Interdecadal change in the effect of spring soil moisture over the Indo-China Peninsula on the following summer precipitation over the Yangtze River basin. J. Clim. 2020, 33, 7063–7082. [Google Scholar] [CrossRef] [Scilit]
  19. Pinault, J.L. The Moist Adiabat, Key of the Climate Response to Anthropogenic Forcing. Climate 2020, 8, 45. [Google Scholar] [CrossRef] [Scilit]
  20. Thirumalai, K.; DiNezio, P.; Okumura, Y.; Deser, C. Extreme temperatures in Southeast Asia caused by El Niño and worsened by global warming. Nat. Commun. 2017, 8, 15531. [Google Scholar] [CrossRef] [Scilit]
  21. Lin, L.; Chen, C.; Luo, M. Impacts of El Niño–Southern Oscillation on heat waves in the Indochina peninsula. Atmos. Sci. Lett. 2018, 19, e856. [Google Scholar] [CrossRef] [Scilit]
  22. Luo, M.; Lau, N.C. Synoptic characteristics, atmospheric controls, and long-term changes of heat waves over the Indochina Peninsula. Clim. Dyn. 2018, 51, 2707–2723. [Google Scholar] [CrossRef] [Scilit]
  23. Zhang, Y.; Li, T.; Wang, B.; Wu, G. Onset of the summer monsoon over the indochina peninsula: Climatology and interannual variations. J. Clim. 2002, 15, 3206–3221. [Google Scholar] [CrossRef] [Scilit]
  24. Zhang, Z.; Chan, J.C.; Ding, Y. Characteristics, evolution and mechanisms of the summer monsoon onset over Southeast Asia. Int. J. Climatol. 2004, 24, 1461–1482. [Google Scholar] [CrossRef] [Scilit]
  25. Wang, B.; Lin, H.O. Rainy season of the Asian–Pacific summer monsoon. J. Clim. 2002, 15, 386–398. [Google Scholar] [CrossRef] [Scilit]
  26. Kueh, M.T.; Lin, S.C. A climatological study on the role of the South China Sea monsoon onset in the development of the East Asian summer monsoon. Theor. Appl. Climatol. 2010, 99, 163–186. [Google Scholar] [CrossRef] [Scilit]
  27. Kajikawa, Y.; Wang, B. Interdecadal change of the South China Sea summer monsoon onset. J. Clim. 2012, 25, 3207–3218. [Google Scholar] [CrossRef] [Scilit]
  28. Jiang, N.; Zhu, C. Seasonal forecast of South China Sea summer monsoon onset disturbed by cold tongue La Nina in the past decade. Adv. Atmos. Sci. 2021, 38, 147–155. [Google Scholar] [CrossRef] [Scilit]
  29. Xu, M.; Xu, H.; Ma, J.; Deng, J. Impact of Pacific Decadal Oscillation on interannual relationship between El Niño and South China Sea summer monsoon onset. Int. J. Climatol. 2022, 42, 2739–2753. [Google Scholar] [CrossRef] [Scilit]
  30. Ai, Y.; Jiang, N.; Qian, W.H.; Leung, J.C.H.; Chen, Y.Y. Strengthened Regulation of the Onset of the South China Sea Summer Monsoon by the Northwest Indian Ocean Warming in the Past Decade. Adv. Atmos. Sci. 2022, 39, 943–952. [Google Scholar] [CrossRef] [Scilit]
  31. Hu, P.; Chen, W.; Li, Z.; Chen, S.; Wang, L.; Liu, Y. Close Linkage of the South China Sea Summer Monsoon Onset and Extreme Rainfall in May over Southeast Asia: Role of the Synoptic-Scale Systems. J. Clim. 2022, 35, 4347–4362. [Google Scholar] [CrossRef] [Scilit]
  32. Ge, F.; Zhu, S.; Sielmann, F.; Fraedrich, K.F.; Zhu, X.; Zhang, L.; Zhi, X.; Wang, H. Precipitation over Indochina during the monsoon transition: Modulation by Indian Ocean and ENSO regimes. Clim. Dyn. 2021, 57, 2491–2504. [Google Scholar] [CrossRef] [Scilit]
  33. Li, G.; Gao, C.; Lu, B.; Chen, H. Inter-annual variability of spring precipitation over the Indo-China Peninsula and its asymmetric relationship with El Niño-Southern Oscillation. Clim. Dyn. 2021, 56, 2651–2665. [Google Scholar] [CrossRef] [Scilit]
  34. Li, G.; Gao, C.; Xu, B.; Lu, B.; Chen, H.; Ma, H.; Li, X. Strengthening Influence of El Niño on the Following Spring Precipitation over the Indochina Peninsula. J. Clim. 2021, 34, 5971–5984. [Google Scholar]
  35. Chen, M.; Shi, W.; Xie, P.; Silva, V.B.S.; Kousky, V.E.; Higgins, R.W.; Janowiak, J.E. Assessing objective techniques for gauge-based analyses of global daily precipitation. J. Geophys. Res. 2008, 113, D04110. [Google Scholar] [CrossRef] [Scilit]
  36. Saha, S.; Moorthi, S.; Pan, H.; Wu, X.; Wang, J.; Nadiga, S.; Tripp, P.; Kistler, R.; Woollen, J.; Behringer, D.; et al. The NCEP Climate Forecast System Reanalysis. Bull. Am. Meteor. Soc. 2010, 91, 1015–1105. [Google Scholar] [CrossRef] [Scilit]
  37. Suranjana, S.; Shrinivas, M.; Michael, E.; Jesse, M.; Yang, R.; Peña, M.M.; Huug, V.D.D.; Zhang, Q.; Wang, W.; Chen, M.; et al. The NCEP Climate Forecast System Version 2. J. Clim. 2014, 27, 2185–2208. [Google Scholar] [CrossRef] [Scilit]
  38. NOAA (National Oceanic and Atmospheric Administration). Cold and Warm Episodes by Season. Available online: https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php (accessed on 10 June 2022).
  39. King, M.D.; Platnick, S.; Menzel, W.P.; Ackerman, S.A.; Hubanks, P.A. Spatial and temporal distribution of clouds observed by MODIS onboard the Terra and Aqua satellites. IEEE Trans. Geosci. Remote Sens. 2013, 51, 3826–3852. [Google Scholar] [CrossRef] [Scilit]
  40. Hubanks, P.A.; King, M.D.; Platnick, S.; Ridgway, B. MODIS Atmosphere L3 Gridded Product Algorithm Theoretical Basis Document (ATBD) & Users Guide; MODIS Algorithm Theoretical Basis Document No. ATBD-MOD-30; NASA Goddard Space Flight Center: Greenbelt, MD, USA, 2020. Available online: https://atmosphere-imager.gsfc.nasa.gov/sites/default/files/ModAtmo/documents/L3_ATBD_C6_C61_2020_08_06.pdf (accessed on 22 March 2022).
  41. Rossow, W.B.; Schiffer, R.A. Advances in understanding clouds from ISCCP. Bull. Am. Meteor. Soc. 1999, 80, 2261–2288. [Google Scholar] [CrossRef] [Scilit]
  42. Stohl, A.; Forster, C.; Frank, A.; Seibert, P.; Wotawa, G. Technical note: The Lagrangian particle dispersion model FLEXPART version 6.2. Atmos. Chem. Phys. 2005, 5, 2461–2474. [Google Scholar] [CrossRef] [Scilit]
  43. Seibert, P.; Frank, A. Source-receptor matrix calculation with a Lagrangian particle dispersion model in backward mode. Atmos. Chem. Phys. 2004, 4, 51–63. [Google Scholar] [CrossRef] [Scilit]
  44. Sodemann, H.; Schwierz, C.; Wernli, H. Interannual variability of Greenland winter precipitation sources: Lagrangian moisture diagnostic and North Atlantic Oscillation influence. J. Geophys. Res. 2008, 113, D03107. [Google Scholar] [CrossRef] [Scilit]
  45. Gimeno, L.; Drumond, A.; Nieto, R.; Trigo, R.M.; Stohl, A. On the origin of continental precipitation. Geophys. Res. Lett. 2010, 37, L13804. [Google Scholar] [CrossRef] [Scilit]
  46. Drumond, A.; Nieto, R.; Gimeno, L. Sources of moisture for China and their variations during drier and wetter conditions in 2000–2004: A Lagrangian approach. Clim. Res. 2011, 50, 215–225. [Google Scholar] [CrossRef] [Scilit]
  47. Kueh, M.T.; Lin, P.L. Springtime cloud properties in the Taiwan Strait: Synoptic controls and local processes. Theor. Appl. Climatol. 2014, 116, 463–480. [Google Scholar] [CrossRef] [Scilit]
  48. McBride, J.; Frank, W. Relationships between stability and monsoon convection. J. Atmos. Sci. 1999, 56, 24–36. [Google Scholar] [CrossRef] [Scilit]
  49. Williams, K.D.; Webb, M.J. A quantitative performance assessment of cloud regimes in climate models. Clim. Dyn. 2009, 33, 141–157. [Google Scholar] [CrossRef] [Scilit]
  50. Tsushima, Y.; Ringer, M.A.; Webb, M.J.; Williams, K.D. Quantitative evaluation of the seasonal variations in climate model cloud regimes. Clim. Dyn. 2013, 41, 2679–2696. [Google Scholar] [CrossRef] [Scilit]
  51. Lobell, D.B.; Bonfils, C.; Duffy, P.B. Climate change uncertainty for daily minimum and maximum temperatures: A model inter-comparison. Geophys. Res. Lett. 2007, 34, L05715. [Google Scholar] [CrossRef] [Scilit]
  52. Hill, P.G.; Allan, R.P.; Chiu, J.C.; Stein, T.H.M. A multisatellite climatology of clouds, radiation, and precipitation in southern West Africa and comparison to climate models. J. Geophys. Res. 2016, 121, 10857–10879. [Google Scholar] [CrossRef] [Scilit]
  53. Saha, S.; Moorthi, S.; Pan, H.L.; Wu, X.; Wang, J.; Nadiga, S.; Tripp, P.; Kistler, R.; Woollen, J.; Behringer, D.; et al. NCEP Climate Forecast System Reanalysis (CFSR) 6-hourly Products, January 1979 to December 2010. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. 2010. Available online: https://data.ucar.edu/dataset/ncep-climate-forecast-system-reanalysis-cfsr-6-hourly-products-january-1979-to-december-2010 (accessed on 7 February 2022). [CrossRef] [Scilit]
  54. Saha, S.; Moorthi, S.; Wu, X.; Wang, J.; Nadiga, S.; Tripp, P.; Behringer, D.; Hou, Y.-T.; Chuang, H.-Y.; Iredell, M.; et al. Updated Daily. NCEP Climate Forecast System Version 2 (CFSv2) 6-hourly Products. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. 2011. Available online: https://search.datacite.org/works/10.5065/d61c1txf (accessed on 7 February 2022). [CrossRef] [Scilit]
Remotesensing 14 04077 g001
Figure 1. (a) The climatological 5870-gpm (green) and 1490-gpm (orange) contours derived from the 500 and 850 hPa geopotential height fields, respectively, for April, May, and June. The topography is shown in meters. The cyan box denotes the ICP region used in this study. The climatological time series of the area-averaged (b) daily maximum and minimum surface air temperatures (Tmax and Tmin) and (c) diurnal temperature range (DTR), precipitation (Precip), and zonal wind component at 850 hPa (U850) over the ICP region. The solid lines represent the means of daily climatology, while the shaded envelopes enclose the interannual spreads denoted by their respective ± one standard deviation from the mean. The monthly (daily) climatology is computed as a time mean for each calendar month (day) during the period 1979 to 2021. The daily standard deviation is estimated for each calendar day from this 43-year period. The horizontal gray lines in panel (b) indicate 24, 25, 33, and 34 °C. The vertical gray lines indicate 1 May and 31 May, respectively.
Figure 1. (a) The climatological 5870-gpm (green) and 1490-gpm (orange) contours derived from the 500 and 850 hPa geopotential height fields, respectively, for April, May, and June. The topography is shown in meters. The cyan box denotes the ICP region used in this study. The climatological time series of the area-averaged (b) daily maximum and minimum surface air temperatures (Tmax and Tmin) and (c) diurnal temperature range (DTR), precipitation (Precip), and zonal wind component at 850 hPa (U850) over the ICP region. The solid lines represent the means of daily climatology, while the shaded envelopes enclose the interannual spreads denoted by their respective ± one standard deviation from the mean. The monthly (daily) climatology is computed as a time mean for each calendar month (day) during the period 1979 to 2021. The daily standard deviation is estimated for each calendar day from this 43-year period. The horizontal gray lines in panel (b) indicate 24, 25, 33, and 34 °C. The vertical gray lines indicate 1 May and 31 May, respectively.
Remotesensing 14 04077 g001
Remotesensing 14 04077 g002
Figure 2. The temporal variations of daily (a) maximum and (b) minimum surface temperatures averaged over the ICP region, as shown in Figure 1a. The vertical axis is the 43-year period from 1979 to 2021, and the horizontal axis is the 90-day span of calendar days from 17 March to 14 June, with the color representing the temperature value for a day of the year. The green color indicates a missing record. Vertical gray lines indicate 1 May and 31 May, respectively.
Figure 2. The temporal variations of daily (a) maximum and (b) minimum surface temperatures averaged over the ICP region, as shown in Figure 1a. The vertical axis is the 43-year period from 1979 to 2021, and the horizontal axis is the 90-day span of calendar days from 17 March to 14 June, with the color representing the temperature value for a day of the year. The green color indicates a missing record. Vertical gray lines indicate 1 May and 31 May, respectively.
Remotesensing 14 04077 g002
Remotesensing 14 04077 g003
Figure 3. (a) Time series of the standardized monthly anomalies of daily maximum temperature (Tmax), precipitation (Precip), zonal wind component at 850 hPa (U850), and cloud fraction (CF). The analysis period is from 1979 to 2021 for all the variables except CF, for which a shorter period from 2003 to 2021 is applicable. The bar chart displays the winter ONI with pink (light blue) denoting a positive (negative) anomaly. The ONI is the SST anomaly in °C. The blue (red) filled rectangles along the time axis denote the cold (warm) years of a standardized Tmax anomalous value below 0.5 (exceed 1) standard deviation. See text for details. The daily time series of cold (blue) and warm (red) years consist of (b) daily maximum and minimum temperatures, (c) diurnal temperature range (DTR) and U850, (d) precipitation, and (e) CF. The composites are computed for the 91-day span of calendar days from 1 April to 30 June derived from 8 cold and 8 warm years, respectively, as indicated in panel (a). Only the cold and warm years during the period from 2003 to 2021 are used for the composites. The horizontal gray lines in panel (b) indicate 24, 25, 33, and 34 °C. The vertical gray lines indicate 1 May and 31 May, respectively. All the time series shown are area averages over the ICP region, as indicated in Figure 2a.
Figure 3. (a) Time series of the standardized monthly anomalies of daily maximum temperature (Tmax), precipitation (Precip), zonal wind component at 850 hPa (U850), and cloud fraction (CF). The analysis period is from 1979 to 2021 for all the variables except CF, for which a shorter period from 2003 to 2021 is applicable. The bar chart displays the winter ONI with pink (light blue) denoting a positive (negative) anomaly. The ONI is the SST anomaly in °C. The blue (red) filled rectangles along the time axis denote the cold (warm) years of a standardized Tmax anomalous value below 0.5 (exceed 1) standard deviation. See text for details. The daily time series of cold (blue) and warm (red) years consist of (b) daily maximum and minimum temperatures, (c) diurnal temperature range (DTR) and U850, (d) precipitation, and (e) CF. The composites are computed for the 91-day span of calendar days from 1 April to 30 June derived from 8 cold and 8 warm years, respectively, as indicated in panel (a). Only the cold and warm years during the period from 2003 to 2021 are used for the composites. The horizontal gray lines in panel (b) indicate 24, 25, 33, and 34 °C. The vertical gray lines indicate 1 May and 31 May, respectively. All the time series shown are area averages over the ICP region, as indicated in Figure 2a.
Remotesensing 14 04077 g003
Remotesensing 14 04077 g004
Figure 4. The seasonal variation of cloud fraction (CF) as revealed by averages of four 15-day periods from 16 April to 14 June. (a–d) The spatial distribution of climatological CF (shaded) and composited CF for warm years (red contour) and cold years (blue contour). The CF is a dimensionless quantity and is expressed as a percentage here. The contour interval is 10%; with the first contour at 70%. The plots were derived from daily CF. (e–h) The longitudinal distribution is of composited CF for warm years (red lines) and cold years (blue lines). The daily, daytime, and nighttime CFs are shown. The longitudinal values were obtained by averages from 10.5°N to 18.5°N. The vertical gray lines indicate the western and eastern bounds of the ICP. From left to right panels, the 15-day periods are 16 April to 30 April, 1 May to 15 May, 15 May to 30 May, and 31 May to 14 June, respectively.
Figure 4. The seasonal variation of cloud fraction (CF) as revealed by averages of four 15-day periods from 16 April to 14 June. (a–d) The spatial distribution of climatological CF (shaded) and composited CF for warm years (red contour) and cold years (blue contour). The CF is a dimensionless quantity and is expressed as a percentage here. The contour interval is 10%; with the first contour at 70%. The plots were derived from daily CF. (e–h) The longitudinal distribution is of composited CF for warm years (red lines) and cold years (blue lines). The daily, daytime, and nighttime CFs are shown. The longitudinal values were obtained by averages from 10.5°N to 18.5°N. The vertical gray lines indicate the western and eastern bounds of the ICP. From left to right panels, the 15-day periods are 16 April to 30 April, 1 May to 15 May, 15 May to 30 May, and 31 May to 14 June, respectively.
Remotesensing 14 04077 g004
Remotesensing 14 04077 g005
Figure 5. The seasonal variation of the joint probability distribution of the COT binned against CTP for liquid water clouds over the ICP. The seasonal variation is revealed by the statistics of four 15-day periods from 16 April to 14 June in the years from 2003 to 2021. From top to bottom panels, the 15-day periods are 16 April to 30 April, 1 May to 15 May, 15 May to 30 May, and 31 May to 14 June, respectively. (a) The normalized joint probability is displayed as two-dimensional histograms, with COT as the x-axis and CTP as the y-axis. The number in square brackets is the total count for each statistical period. For every bin box, the joint probability value is equal to the bin count divided by the total count (relative frequency). The normalized joint probability in each bin box is computed by dividing the corresponding joint probability value by the bin size (the area of that particular bin box), thereby obtaining the probability density value of the bin box. This normalization is intended to take into account the bin sizes so as to eliminate any visual distortions when comparing bins of different sizes in the two-dimensional histogram plot. The probability of any pixel that falls in a joint histogram bin box can be retrieved after multiplying the corresponding probability density value by that bin size. The bin colors represent the probability density values in each bin. The COT histogram bins ranging from 30 to 150 were chopped off for plotting purposes, owing to the large bin range and very low probability density. The marginal probabilities of CTP with respect to three COT categories are displayed as (b) stacked and (c) overlapping histograms, with probability as the x-axis and CTP as the y-axis. The marginal probabilities of CTP are derived from the joint histograms of COT binned against CTP. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, 20–150. Three sets of the marginal probability of CTP are obtained by integrating (summing) the joint probability over the specific ranges of the aggregated bins from the three COT categories, respectively.
Figure 5. The seasonal variation of the joint probability distribution of the COT binned against CTP for liquid water clouds over the ICP. The seasonal variation is revealed by the statistics of four 15-day periods from 16 April to 14 June in the years from 2003 to 2021. From top to bottom panels, the 15-day periods are 16 April to 30 April, 1 May to 15 May, 15 May to 30 May, and 31 May to 14 June, respectively. (a) The normalized joint probability is displayed as two-dimensional histograms, with COT as the x-axis and CTP as the y-axis. The number in square brackets is the total count for each statistical period. For every bin box, the joint probability value is equal to the bin count divided by the total count (relative frequency). The normalized joint probability in each bin box is computed by dividing the corresponding joint probability value by the bin size (the area of that particular bin box), thereby obtaining the probability density value of the bin box. This normalization is intended to take into account the bin sizes so as to eliminate any visual distortions when comparing bins of different sizes in the two-dimensional histogram plot. The probability of any pixel that falls in a joint histogram bin box can be retrieved after multiplying the corresponding probability density value by that bin size. The bin colors represent the probability density values in each bin. The COT histogram bins ranging from 30 to 150 were chopped off for plotting purposes, owing to the large bin range and very low probability density. The marginal probabilities of CTP with respect to three COT categories are displayed as (b) stacked and (c) overlapping histograms, with probability as the x-axis and CTP as the y-axis. The marginal probabilities of CTP are derived from the joint histograms of COT binned against CTP. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, 20–150. Three sets of the marginal probability of CTP are obtained by integrating (summing) the joint probability over the specific ranges of the aggregated bins from the three COT categories, respectively.
Remotesensing 14 04077 g005
Remotesensing 14 04077 g006
Figure 6. The monthly (May) and seasonal variation of the marginal probabilities of CTP with respect to three categories of (a,b) COT and (c,d) CER, respectively, compiled over the ICP as collected from cold (blue tone) and warm (red tone) years. The seasonal variation is revealed by the statistics of three 15-day periods from 16 April to 30 May of the cold and warm years. The numbers in square brackets are the total counts for these statistical periods; the total counts of CER are equal to COT. The COT and CER are for liquid water clouds in the daytime; COT is a dimensionless quantity, and CER is in μm. The marginal probabilities of CTP are derived from the joint histograms of COT or CER binned against CTP. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, and 20–150, whereas the CER are grouped into three categories of newly aggregated bins from the original CER histogram bins: 4–10,10–17.5, and 17.5–30. Three sets of the marginal probabilities of CTP are obtained by integrating the joint probability (relative frequency) over the specific ranges of the aggregated bins from the three corresponding COT or CER categories, respectively. (a,c) The marginal probabilities of May displayed as stacked histograms. (b,d) The marginal probabilities of 15-day periods displayed as overlapping histograms, with each plot panel for each individual category. The x-axis is probability, the y-axis is CTP.
Figure 6. The monthly (May) and seasonal variation of the marginal probabilities of CTP with respect to three categories of (a,b) COT and (c,d) CER, respectively, compiled over the ICP as collected from cold (blue tone) and warm (red tone) years. The seasonal variation is revealed by the statistics of three 15-day periods from 16 April to 30 May of the cold and warm years. The numbers in square brackets are the total counts for these statistical periods; the total counts of CER are equal to COT. The COT and CER are for liquid water clouds in the daytime; COT is a dimensionless quantity, and CER is in μm. The marginal probabilities of CTP are derived from the joint histograms of COT or CER binned against CTP. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, and 20–150, whereas the CER are grouped into three categories of newly aggregated bins from the original CER histogram bins: 4–10,10–17.5, and 17.5–30. Three sets of the marginal probabilities of CTP are obtained by integrating the joint probability (relative frequency) over the specific ranges of the aggregated bins from the three corresponding COT or CER categories, respectively. (a,c) The marginal probabilities of May displayed as stacked histograms. (b,d) The marginal probabilities of 15-day periods displayed as overlapping histograms, with each plot panel for each individual category. The x-axis is probability, the y-axis is CTP.
Remotesensing 14 04077 g006
Remotesensing 14 04077 g007
Figure 7. The monthly (May) and seasonal variation of the marginal probabilities of CER with respect to the three categories of COT compiled over the ICP as collected from cold (blue tone) and warm (red tone) years. The seasonal variation is revealed by the statistics of three 15-day periods from 16 April to 30 May of the cold and warm years. The numbers in square brackets are the total counts for these statistical periods. The COT and CER are for liquid water clouds in the daytime; COT is a dimensionless quantity, and CER is in μm. The marginal probabilities of CER are derived from the joint histograms of COT binned against CER. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, and 20–150. Three sets of the marginal probabilities of CER are obtained by integrating the joint probability (relative frequency) over the specific ranges of the aggregated bins from the three corresponding COT categories, respectively. (a) The marginal probabilities of May displayed as stacked histograms. (b) The marginal probabilities of 15-day periods displayed as overlapping histograms, with each plot panel for each individual category. The x-axis is probability, the y-axis is CER.
Figure 7. The monthly (May) and seasonal variation of the marginal probabilities of CER with respect to the three categories of COT compiled over the ICP as collected from cold (blue tone) and warm (red tone) years. The seasonal variation is revealed by the statistics of three 15-day periods from 16 April to 30 May of the cold and warm years. The numbers in square brackets are the total counts for these statistical periods. The COT and CER are for liquid water clouds in the daytime; COT is a dimensionless quantity, and CER is in μm. The marginal probabilities of CER are derived from the joint histograms of COT binned against CER. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, and 20–150. Three sets of the marginal probabilities of CER are obtained by integrating the joint probability (relative frequency) over the specific ranges of the aggregated bins from the three corresponding COT categories, respectively. (a) The marginal probabilities of May displayed as stacked histograms. (b) The marginal probabilities of 15-day periods displayed as overlapping histograms, with each plot panel for each individual category. The x-axis is probability, the y-axis is CER.
Remotesensing 14 04077 g007
Remotesensing 14 04077 g008
Figure 8. (a) The 5870-gpm (gray and blue) and 5880-gpm (red) contours derived from the composites of 500 hPa geopotential height for climatology, cold, and warm years, respectively. The 5870-gpm is shown for cold years because all the gridded values in the corresponding 500 hPa geopotential height composites over the plotting domain are below 5875-gpm. (b) The 1490-gpm contours derived from the composites of 850 hPa geopotential height for climatology (gray), cold (blue), and warm (red) years. All the composites were computed for the first (solid) and second (dashed) 15-day period of May for climatology (1979–2021), cold, and warm years. The cyan boxes denote the BoB, ICP, and SCS regions with the same area size.
Figure 8. (a) The 5870-gpm (gray and blue) and 5880-gpm (red) contours derived from the composites of 500 hPa geopotential height for climatology, cold, and warm years, respectively. The 5870-gpm is shown for cold years because all the gridded values in the corresponding 500 hPa geopotential height composites over the plotting domain are below 5875-gpm. (b) The 1490-gpm contours derived from the composites of 850 hPa geopotential height for climatology (gray), cold (blue), and warm (red) years. All the composites were computed for the first (solid) and second (dashed) 15-day period of May for climatology (1979–2021), cold, and warm years. The cyan boxes denote the BoB, ICP, and SCS regions with the same area size.
Remotesensing 14 04077 g008
Remotesensing 14 04077 g009
Figure 9. The marginal probabilities of CTP with respect to three categories of COT compiled over the (a) BoB, (b) ICP, and (c) SCS regions, collected from May of cold (blue tone) and warm (red tone) years. The BoB, ICP, and SCS regions are indicated in Figure 8b. The COT is for liquid water cloud in the daytime and is a dimensionless quantity, CTP is in hPa. The numbers in square brackets are the total counts for these statistical periods. The marginal probabilities of CTP are derived from the joint histograms of COT binned against CTP and are displayed as stacked histograms with probability as the x-axis and CTP as the y-axis. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, and 20–150, whereas the CTP are grouped into three categories of newly aggregated bins from the original CTP histogram bins: 50–450, 440–700, and 700–1100 hPa. For each plot panel, three sets of the marginal probabilities of CTP are obtained by integrating the joint probability (relative frequency) over the specific ranges of the aggregated bins from the three corresponding COT categories, respectively.
Figure 9. The marginal probabilities of CTP with respect to three categories of COT compiled over the (a) BoB, (b) ICP, and (c) SCS regions, collected from May of cold (blue tone) and warm (red tone) years. The BoB, ICP, and SCS regions are indicated in Figure 8b. The COT is for liquid water cloud in the daytime and is a dimensionless quantity, CTP is in hPa. The numbers in square brackets are the total counts for these statistical periods. The marginal probabilities of CTP are derived from the joint histograms of COT binned against CTP and are displayed as stacked histograms with probability as the x-axis and CTP as the y-axis. The COT is grouped into three categories of newly aggregated bins from the original COT histogram bins: 0–4, 4–20, and 20–150, whereas the CTP are grouped into three categories of newly aggregated bins from the original CTP histogram bins: 50–450, 440–700, and 700–1100 hPa. For each plot panel, three sets of the marginal probabilities of CTP are obtained by integrating the joint probability (relative frequency) over the specific ranges of the aggregated bins from the three corresponding COT categories, respectively.
Remotesensing 14 04077 g009
Remotesensing 14 04077 g010
Figure 10. The box-and-whisker plots for the daily CAPE in May for BoB, ICP, and SCS. The daily CAPE is averaged over the corresponding regions. Blue and red for cold and warm years, respectively. Outliers are excluded by 1.5 interquartile ranges.
Figure 10. The box-and-whisker plots for the daily CAPE in May for BoB, ICP, and SCS. The daily CAPE is averaged over the corresponding regions. Blue and red for cold and warm years, respectively. Outliers are excluded by 1.5 interquartile ranges.
Remotesensing 14 04077 g010
Remotesensing 14 04077 g011
Figure 11. The density distributions of the backward trajectories from the ICP region started within a layer bounded with heights between 1000 and 2000 m above mean sea level. The horizontal distributions of the trajectory density for the 15-day periods of 1–15 May from the (a) cold and (b) warm years and of 16–30 May from the (c) cold and (d) warm years. For each 15-day period, there are 486,000 initial air parcels released and tracked, and their six-hourly horizontal positions are then binned onto a 1° × 1° latitude–longitude grid, from which the histograms are calculated. Using the histograms (count in each grid cell), the trajectory density per grid cell is computed as the percentage of the total. The density of the 120, 72, and 24 h backward trajectories are shown from left to right panels, respectively, in (a–d). The numbers in square brackets are the total counts of the trajectory remaining in the domain at the corresponding backward times. The vertical distributions of the trajectory density were collected in the 15-day periods of (e) 1–15 May and (f) 16–30 May. The six-hourly trajectory histograms of the horizontal positions are further binned with respect to the trajectory altitudes using 1 km increments, and the domain-wide trajectory density per 1 km increment is computed as the percentage of the total. Trajectory densities from warm and cold years are superimposed for comparison.
Figure 11. The density distributions of the backward trajectories from the ICP region started within a layer bounded with heights between 1000 and 2000 m above mean sea level. The horizontal distributions of the trajectory density for the 15-day periods of 1–15 May from the (a) cold and (b) warm years and of 16–30 May from the (c) cold and (d) warm years. For each 15-day period, there are 486,000 initial air parcels released and tracked, and their six-hourly horizontal positions are then binned onto a 1° × 1° latitude–longitude grid, from which the histograms are calculated. Using the histograms (count in each grid cell), the trajectory density per grid cell is computed as the percentage of the total. The density of the 120, 72, and 24 h backward trajectories are shown from left to right panels, respectively, in (a–d). The numbers in square brackets are the total counts of the trajectory remaining in the domain at the corresponding backward times. The vertical distributions of the trajectory density were collected in the 15-day periods of (e) 1–15 May and (f) 16–30 May. The six-hourly trajectory histograms of the horizontal positions are further binned with respect to the trajectory altitudes using 1 km increments, and the domain-wide trajectory density per 1 km increment is computed as the percentage of the total. Trajectory densities from warm and cold years are superimposed for comparison.
Remotesensing 14 04077 g011
Remotesensing 14 04077 g012
Figure 12. The probability distributions of the specific humidity carried by the backward trajectories at the backward times of 72 and 24 h are presented for the 15-day periods of (a) 1–15 May and (b) 16–30 May. The probability is computed as a relative frequency; that is, the counts in each humidity bin divided by the total counts of the trajectory remaining within a subdomain (5°–18°N, 80°–120°E) at the corresponding backward times. Similar patterns of distributions are obtained by using the calculation domain (10°S–25°N, 70°–130°E) for backward trajectory. The left panels are for the specific humidity from all trajectories; the central and right panels are the marginal probability of specific humidity with respect to the temperature values. The numbers in square brackets (left panels) are the total counts of the trajectory remaining in the subdomain at the corresponding backward times.
Figure 12. The probability distributions of the specific humidity carried by the backward trajectories at the backward times of 72 and 24 h are presented for the 15-day periods of (a) 1–15 May and (b) 16–30 May. The probability is computed as a relative frequency; that is, the counts in each humidity bin divided by the total counts of the trajectory remaining within a subdomain (5°–18°N, 80°–120°E) at the corresponding backward times. Similar patterns of distributions are obtained by using the calculation domain (10°S–25°N, 70°–130°E) for backward trajectory. The left panels are for the specific humidity from all trajectories; the central and right panels are the marginal probability of specific humidity with respect to the temperature values. The numbers in square brackets (left panels) are the total counts of the trajectory remaining in the subdomain at the corresponding backward times.
Remotesensing 14 04077 g012
Table 1. The ICP cold and warm years and the corresponding seasonal ENSO conditions as revealed by the preceding winter and spring ONI. The cold, warm, and neutral conditions are denoted as characters C, W, and N, respectively, in the square brackets. The first (second) character is for winter (spring) ONI. Note that only seven cold years can be identified after the year 2003; the year 2004, with an anomalous value of −0.34, was selected to be a cold year to ensure the same sample size for the warm-year and cold-year composites.
Table 1. The ICP cold and warm years and the corresponding seasonal ENSO conditions as revealed by the preceding winter and spring ONI. The cold, warm, and neutral conditions are denoted as characters C, W, and N, respectively, in the square brackets. The first (second) character is for winter (spring) ONI. Note that only seven cold years can be identified after the year 2003; the year 2004, with an anomalous value of −0.34, was selected to be a cold year to ensure the same sample size for the warm-year and cold-year composites.
ENSO-RelatedNon-ENSO Factors
Cold years2006 [CC], 2008 [CC], 2009 [CC],
2011 [CC], 2012 [CC]		2004 [NN], 2017 [NN]
2007 [WN]
Warm years2005 [WN], 2010 [WW], 2015 [WW],
2016 [WW], 2019 [WW]		2014 [NN], 2020 [NN]
2021 [CC]
Table 2. The correlation coefficients between the preceding winter (spring) ONI and some monthly atmospheric variables over the ICP region of May for three different periods: 1979–2000, 2000–2021, and 1979–2021. The correlation coefficients calculated against spring ONI are given in parentheses. Correlation coefficient values that exceed 0.424 (df = 20) and 0.3 (df = 41) are statistically significant at the 95% level (p < 0.05). The bold numbers indicate that the differences between the correlation coefficients for 1979–2000 and 2000–2021 are statistically significant at the 95% level (p < 0.05).
Table 2. The correlation coefficients between the preceding winter (spring) ONI and some monthly atmospheric variables over the ICP region of May for three different periods: 1979–2000, 2000–2021, and 1979–2021. The correlation coefficients calculated against spring ONI are given in parentheses. Correlation coefficient values that exceed 0.424 (df = 20) and 0.3 (df = 41) are statistically significant at the 95% level (p < 0.05). The bold numbers indicate that the differences between the correlation coefficients for 1979–2000 and 2000–2021 are statistically significant at the 95% level (p < 0.05).
TmaxTminH500H850U850PrecipCF
1979–20000.80 (0.87)0.92 (0.89)0.68 (0.63)0.58 (0.61)−0.07 (−0.06)−0.41 (−0.52)
2000–20210.66 (0.69)0.69 (0.73)0.65 (0.70)0.40 (0.45)−0.04 (−0.05)−0.41 (−0.41)−0.40 (−0.38)
1979–20210.69 (0.73)0.68 (0.67)0.50 (0.47)0.45 (0.48)−0.02 (−0.02)−0.37 (−0.42)
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Kueh, M.-T.; Lin, C.-Y. Warming Trend and Cloud Responses over the Indochina Peninsula during Monsoon Transition. Remote Sens. 2022, 14, 4077. https://doi.org/10.3390/rs14164077

AMA Style

Kueh M-T, Lin C-Y. Warming Trend and Cloud Responses over the Indochina Peninsula during Monsoon Transition. Remote Sensing. 2022; 14(16):4077. https://doi.org/10.3390/rs14164077

Chicago/Turabian Style

Kueh, Mien-Tze, and Chuan-Yao Lin. 2022. "Warming Trend and Cloud Responses over the Indochina Peninsula during Monsoon Transition" Remote Sensing 14, no. 16: 4077. https://doi.org/10.3390/rs14164077

APA Style

Kueh, M.-T., & Lin, C.-Y. (2022). Warming Trend and Cloud Responses over the Indochina Peninsula during Monsoon Transition. Remote Sensing, 14(16), 4077. https://doi.org/10.3390/rs14164077

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop