Cloud climatology of northwestern Mexico based on MODIS data

ABSTRACT

The geographical regions of northwestern Mexico consisting of the Pacific Ocean, the Baja California Peninsula with its mountain range along it, the Gulf of California, and the coastal zone with its Western Sierra Madre Mountain range, configure an alternation of water, land, water, land, all interacting with the atmosphere. It suggests investigating the cloud patterns, what clouds are most relevant in controlling radiation and the climate, and what type of cloud is associated with observed precipitation patterns in the region. The principal aim was to carry out a climatology of five types of clouds: cumulonimbus, cumulus, altostratus, stratocumulus and nimbostratus. The data set was obtained from the MODIS sensor, placed on the Aqua and Terra satellites, covering the period from 2001 to 2020. The results revealed that the precipitation distribution predominantly relates to the deep convective cloud pattern. A minor fraction of rain is associated with the nimbostratus cloud pattern. The pattern of the fraction of total coverage of the five types of clouds coincides very well with lower radiation values, demonstrating the regulatory role of clouds in the area’s climatology.

KEYWORDS:

• MODIS data
• clouds climatology
• precipitation
• radiation
• NW Mexico

Introduction

The Northwest region of Mexico (NWM) presents interesting atmospheric dynamics throughout the year; like the incidence of hurricanes in the southern part, the northern region undergoes phenomena like Santa Ana winds and incoming cold fronts from the north. The study region encompasses an alternating configuration of the Ocean, the Baja California peninsula with a mountain range, the Gulf of California, and continental coastal planes with the impressive mountain range denominated Western Sierra Madre (WSM). It is interesting to address the geographical configuration of water-land-water-land, the presence of the two mountain ranges, and its relation to cloud formation, emphasizing the coastal and mountainous areas that cover this area of Mexico, causing spatiotemporal patterns of cloud cover. The balance of radiation between the atmosphere, the land surface and the ocean largely determines the climate of a region. Regarding budget aspects, the net radiation absorbed by the earth’s surface is the result of the difference between the incoming shortwave (SWR) and longwave (LWR) radiation minus the sum of the reflected SWR plus the emitted LWR (Bilgiç & Mert, 2021). Clouds play a decisive role in this radiation balance because they are the best regulator of the land-ocean-atmosphere system, being the main link between the energy exchange as well as the exchange of water (ice, liquid, vapor) Garatuza-Payan, Pinker, Shuttleworth, & Watts (2001). The behavior of the physical characteristics on the surface, the different patterns of atmospheric dynamics and different climatic regimes at various scales determine the type and quantity of clouds, their properties, and vertical distribution, which directly modify the global energy balance affecting the climate (Quante, 2004). All these meteorological patterns at different scales make the northwestern region of Mexico ideal for studying complex atmospheric processes related to the spatial distribution, seasonal variability and diurnal cycle of clouds in a climatic framework by using remote sensing methods

The contrast between the waters of the Pacific Ocean and the Gulf of California and the complex topography of two mountain ranges suggest being agents of spatiotemporal variability of humid atmospheric processes. Deserts, semiarid zones, forests, rivers and large agricultural areas configure the region. Investigating the relation of these climatic regions with cloud formation, cloud cover and precipitation is relevant. Its influence on the modulation of these regions’ moisture transport and the organization of cloud systems that generate precipitation should be explained to understand the different rainfall regimes observed in the area (Giovannettone & Barros, 2009). The mountains along the coast cause significant land-ocean-atmosphere interactions, characterizing the region mainly with a maritime-orographic regime where precipitation is associated with forced convection in the high mountain ranges, with the Gulf of California as an essential source of moisture. The relief in the basin of the Gulf of California causes interactions between mesoscale circulations such as influences of the intertropical convergence zone (ITCZ) flow, eastern Pacific warm water pools, easterly waves, the North American monsoon system (NAMS) and the Southern Oscillation (ENSO), in addition to local circulations that determine the spatiotemporal organization of clouds, the distribution of precipitation and the weather patterns in the NWM (Carbone et al., 2002; Mendez-Barroso, 2009; Vera et al., 2006).

Flows propagating eastward into the Eastern Pacific interact with orographic barriers producing strong vertical coupling due to adiabatic heating and generating high cloud activity in the WSM portion. Moreover, the North American monsoon is a relevant source of humidity for the region, causing humidity flows over the WSM, the Gulf of California and the Baja California peninsula, which generates a marked cloud cover during the summer and increases sudden changes in precipitation (Castro et al., 2007). During ENSO, convective processes dominated regions linked to the sea surface temperature variability (SST). Further, the increase in sea temperature strengthens the Hadley and Walker circulations and increases the cloudiness (Bedacht et al., 2007). In a study by Valdés-Manzanilla (2015), the summer mesoscale convective systems (MCS) that developed in northwestern Mexico during the intense ENSO episodes of 1997 to 1999 were analyzed using images from Geostationary Operational Environmental Satellites (GOES 8). The data analysis indicated that the highest number of mesoscale convective systems was associated with the El Niño episode of 1997 and had a longer active period. Analysis of the La Niña episode of 1999 revealed a lower number of mesoscale convective systems, which developed during a shorter active period. Garatuza-Payan, Pinker, and Shuttleworth (2001), and Garatuza-Payan, Pinker, Shuttleworth, & Watts (2001) analyzed the total cloud cover over the same region in an annual cycle (1993–1994), dividing the study area into six regions with different characteristics of climate, land use and topography. They determined that the average cloud cover for the study area for that period was 0.25 (where 1 indicates total cloud cover), with significant interannual variation and a very spatial distribution contrast. Above the WSM, the most significant amount of cloudiness appeared with a seasonal pattern directly related to the rainy seasons of winter and summer, with maximum cloud cover during the afternoons. In the Gulf of California, the large detected number of clouds during the winter exposes an evident daily cycle with maximums in the morning and a negative trend of cloud cover during the rest of the day in the monsoon season. The desert regions showed few clouds, mainly in the spring.

Giovannettone and Barros (2009) used microwave and infrared data from the TRMM, operated by the National Aeronautics and Space Administration (NASA) to identify diurnal distribution patterns of cloudiness. They showed that cloud formation and associated diurnal precipitation responds to mesoscale moisture transport primarily associated with the passage of the monsoon that occurs during the summer. Clouds with extensive vertical development, such as cumulonimbus clouds (Cb), are called convective clouds due to the central role of updrafts in their development and structure. Deep convective clouds (DCC) are characterized mainly by their immediate response to diurnal forcing and rapid injection of air (they can reach upward velocities of several tens of meters per second) transporting moisture, heat and momentum to the upper troposphere (Hong et al., 2006). Diurnal changes in deep convection clouds affect the incidence and reflection of SWR and LWR, directly modifying the Earth’s radiation budget (Hong et al., 2006). DCC can be dangerous, and severe weather conditions are often accompanied, such as flash floods, tornadoes and hail that cause significant damage (Morel & Senesi, 2002a). Deep convective clouds occur mainly near coasts, on the lee side of mountain ranges, and near areas where long-wave radiation gradients occur (Morel & Senesi, 2002b). It is possible to characterize clouds through remote sensing in complex terrain, as in NWM. Previous studies on cloud formation based on satellite observations have examined their physical and radiative properties (albedo, emittance, absorption, cloud optical depth, optical thickness and others) as well as local and seasonal variations and their behavior on broader spatial and temporal scales (Bergman & Salby, 1996; Doelling et al., 2004; Garatuza-Payan, Pinker, and Shuttleworth 2001, and Garatuza-Payan, Pinker, Shuttleworth, & Watts 2001; Henken et al., 2011; León-Cruz et al., 2021; Meerkötter & Zinner, 2007; Meskhidze et al., 2009; Pfeifroth et al., 2012; Yuan et al., 2010; Zuidema et al., 2007). The determining factor in climate studies is the spatiotemporal distribution of the different types of clouds.

These satellite measurements of clouds serve as a complement to surface observations, generating information on larger spatial and temporal scales of dynamic patterns and creating a cloud climatology based on their detection and characterization according to their morphological and radiative properties. This work’s primary purpose is to research cloud climatology in northwestern Mexico. In this region, atmospheric processes interact with an interesting ocean-land-sea-land configuration of the eastern Pacific Ocean, Baja California Peninsula, Gulf of California and mainland Mexico, and with two mountain ranges, one along the peninsula and the Western Sierra Madre.

All these meteorological patterns at different scales make the northwestern region of Mexico ideal for studying complex atmospheric processes related to the spatial distribution, seasonal variability and diurnal cycle of clouds in a climatic framework by using remote sensing methods. Therefore, this study conducts a cloud climatological analysis of 20 years (2001–2020) on the detection, quantification, distribution and classification of different types of clouds (cumulonimbus, cumulus, altostratus, stratocumulus and nimbostratus) with satellite data and methods of remote sensing finding trends, patterns and relationships to understand the atmospheric consequences of these clouds.

Study area

The Northwestern part of Mexico configures an interesting ocean-land combination that includes the semi-enclosed Gulf of California basin (Figure 1). An elevated topography surrounds the gulf, except for the southern boundary. To the west are the mountains of the Baja California peninsula that have an average height ranging between 700 and 1,000 m above sea level and function as a natural barrier causing different atmospheric and oceanic conditions between the Gulf of California and the Pacific Ocean (Badan‐Dangon et al., 1991). To the east and north, the WSM elevations exceed 3,000 m above sea level with steep slopes, especially in the southern part, where the mountains almost reach the coast Garatuza-Payan, Pinker, Shuttleworth, & Watts (2001). This orographic barrier crosses the Mexican states of Chihuahua, Sonora, Sinaloa and Nayarit. The study area of this work extends between latitudes 17° and 34°N and longitudes −100° and −121°W. In this region, the atmospheric circulation responds mainly to local interactions such as the thermal contrast between the Pacific Ocean and the Gulf of California, the land surface temperature, and the complex orography (Brito‐Castillo et al., 2022). The synoptic factors like the high-pressure belt in the northern hemisphere associated with Bermuda’s high system imply an easterly airflow and subsidence resulting in relatively sparse cloud cover and, therefore, little rainfall throughout the year, especially in late spring when the center of the high-pressure belt locates at the same latitude as the Gulf of California Garatuza-Payan, Pinker, and Shuttleworth (2001); Morales-Acuña et al., 2019.

Figure 1. Topography of northwestern Mexico in meters (m). Son indicates the state of Sonora, sin indicates the state of Sinaloa and nay means the state of Nayarit. the black line indicates national borders.

In this mesoscale condition, the relative humidity is lowest, and the temperature reaches its maximum point just before the start of the North American monsoon Garatuza-Payan, Pinker, and Shuttleworth (2001). All these meteorological conditions determine that the Northwestern part of Mexico is dominantly a semi-arid region with an average annual rainfall of 350 mm, with a bimodal pattern characterized by peaks in the subtropical summer and mid-latitude winter (Morales-Acuña et al., 2019). In the summer, the SST is uniformly high, about 29°C, and most of the summer rainfall is convective and tends to be intense. It occurs due to the increase in cloudiness caused by tropical disturbances, either in the form of cyclones or mesoscale circulations and complex convective events present along the west coast of Mexico, and the effect of the North American monsoon system (NAMS) (Barlow et al., 1998; Garatuza-Payan, Pinker, and Shuttleworth (2001). Precipitation from July to September represents 60% to 80% of the total annual precipitation (Berbery, 2001). During winter, the region suffers the intermittent influence of polar disturbances that cause a consistent increase in cloudiness throughout the area, generating rain events Garatuza-Payan, Pinker, and Shuttleworth (2001). A weak trough on the peninsula of Baja California at 700 hPa causes a peak winter rainfall. SST at this time of year averages 18°C near the northern limit and up to 25°C near the gulf’s entrance (Badan‐Dangon et al., 1991). This study conducts a cloud climatological analysis of 20 years (2001–2020) on the detection, quantification, distribution and classification of different types of clouds (cumulonimbus, cumulus, altostratus, stratocumulus and nimbostratus) with satellite data and methods of remote sensing finding trends, patterns and relationships to understand the atmospheric consequences of these clouds.

Data and methodology

MODIS satellite data

The cloud product is a relevant atmospheric content produced by the MODIS (Moderate Resolution Imaging Spectroradiometer) sensor on the polar-orbiting Terra and Aqua satellites. These datasets are technically known as Collection 6 Level-2 MOD06 (Terra) and MYD06 (Aqua) (Menzel et al., 2015). Both have cloud-top properties (pressure, temperature and height) for both day and night and cloud optical properties (optical thickness, effective particle radius and water path for thermodynamic phases) only during the day (Platnick et al., 2015). The Tracking and Data Relay Satellite System (TDRSS) transfers MODIS data to ground stations. They are sent to the EOS Data and Operations System (EDOS) at the Goddard Space Flight Center. The MODIS Adaptive Processing System (MODAPS) produces the cloud and atmosphere products, then parceled out among three DAACs for distribution (Platnick et al., 2016). The data are stored in a Hierarchical Data Format (HDF) version 4, and the information collection period is every five minutes along the path of the orbit. The spatial resolution of the Level-2/C6 data is 1 × 1 km (nadir) for all cloud properties, and the radiometric resolution ranges from 0.405 to 14,385 µm (Platnick et al., 2015). Data of different parameters were collected for 20 years (2001–2020), such as cloud top pressure (CTP), cloud top temperature (CTT), cloud water path (CWP), cloud optical thickness (COT) and cloud top height (CTH), for the study area. The temporal coverage of the MODIS observations was from 03:00 to 22:00 UTC. The products derived from the MODIS observations describe the characteristics of the land, the oceans and the atmosphere. They can be used for studies of processes on a global to local-scale (León-Cruz et al., 2021).

ECMWF v5 reanalysis data (ERA5)

The fifth-generation global atmospheric reanalysis (ERA5) was created by the Copernicus Climate Change Service (C3S) of the ECMWF (European Center for Medium-range Weather Forecasts) (Copernicus Climate Change Service C3S, 2017; Hersbach, 2017). Its main objective is to obtain the best possible description of the atmosphere and the Earth’s surface within a coherent macro (Castro et al., 2007). This database hourly provides many atmospheric, land and oceanic climate variables. It has a significant advantage in that it has long historical records combining observations from weather stations and satellite observations in conjunction with global estimates, using meteorology modeling forecasts in CY41R2 of the Integrated Forecast System (IFS) with a fixed dynamic core, as well as a 4D-Var data assimilation system. This assimilation uses 12-h time windows, from 09:00 to 21:00 UTC and from 21:00 to 09:00 UTC to the next day. ERA5 has a 30 km grid coverage of the Earth and solves the atmospheric equations using 137 sigma/pressure hybrid levels in the vertical, reaching up to 80 km in height. Atmospheric data are available at these levels but are also interpolated to 37 pressure levels, 16 potential temperature levels, and one potential vorticity level (Hersbach et al., 2020). Data for 20 years (2001–2020) of atmospheric and oceanic variables were collected at three pressure levels (1000, 650, and 450 hPa): the temperature at 2 m, sea surface temperature, specific humidity at 2 m, relative humidity at 2 m, atmospheric pressure, precipitation, net solar radiation, energy transfer fluxes, wind speed at 10 m and monthly averages were calculated. Model-level parameters are archived in GRIB2 format, and all other parameters are in GRIB1 format.

CHIRPS reanalysis data

The US Geological Survey Earth Resources Observation and Science Center (EROS) developed quasi-global precipitation data known as CHIRPS (Climate Hazards Group InfraRed Precipitation with Station data) that incorporates information from three main components from an interpolation algorithm (Hazards Center, 2015). These data include a global climatological database known as CHPclim (Climate Hazards Group Precipitation Climatology), infrared satellite observations (TRMM 7), and in situ station data spanning national and regional meteorological services, generating rainfall time helpful series for trend analysis of this variable, the monitoring of seasonal droughts and the study of recent climatic extremes (Funk et al., 2015). The CHIRPS data contain information only from the continental part at different temporal resolutions: daily, pental and monthly available for each point of the grid and with a high spatial resolution of 0.05°, which generates a marked difference concerning other databases (Katsanos et al., 2016). For this work, monthly data for 2001–2020 in the continental part of the study area were used. León-Cruz et al. (2021) highlight that CHIRPS’s advantages are the significant geographic coverage, temporal uniformity, validation processes and updates in almost real-time. Additionally, the temporal and spatial coverage of these data helps understand the contribution of clouds to the precipitation regime.

Cloud classification and quantification

The study of clouds requires remote sensing methods because they are challenging to observe directly from the earth’s surface. Large-scale organized cloud systems such as fronts (cold or tropical) and cyclones are detected from space because weather satellites identify cloud structures based on their vertical and horizontal extensions and through wind and pressure patterns (Davis et al., 2013). According to Houze (2014), there are three scales of movement: the synoptic scale, where phenomena are observed above 2000 km of longitudinal extension; the mesoscale, which includes phenomena between 20 and 2000 km; and the convective scale, which includes processes between 0.2 and 20 km. The internationally recognized cloud classification based on the method of visual identification is described in the International Cloud Atlas of the World Meteorological Organization (WMO) and is summarized in Table 1. However, for this work, the cloud classification described by the International Satellite Cloud Climatology Project (ISCCP) is based on the optical properties of clouds. The main differences between cumuliform clouds (cumulonimbus and cumulus) and stratiform clouds (stratocumulus and nimbostratus) are their formation and appearance. In terms of shape, cumuliforms have the appearance of absorbent cotton, bubbles, or domed mountains. At the same time, stratiforms occur more in layers of large horizontal extent. If a cumulus cloud develops and enters a growth phase, it usually becomes a cumulonimbus cloud. Cumulonimbus are clouds of great vertical extension that can exceed the troposphere (~12 km high), although their base occurs in the lower atmosphere. The general structure of this type of cloud is in the form of a very tall mountain or tower, and at the top, they usually have a flattened shape known as an anvil or plume, which is because the cloud reaches its maximum height. It fails to penetrate the next layer of the atmosphere (stratosphere). However, cumulonimbus clouds are associated with severe precipitation events and hail falls. A meteorological satellite such as MODIS, quickly identifies a cumulonimbus by the anvil that extends horizontally and by the brilliant white color that these clouds present (Houze, 2014). Satellite images show that the stratocumulus can cover regions of about 1000 km on the horizontal scale. In coastal regions, stratocumulus clouds are organized in cloud streets due to strong surface winds blowing from the continent toward the sea’s warm waters. Nimbostratus is a dense layer of gray or dark clouds thick enough to block the sun and is associated with heavy precipitation. Low, torn clouds often appear below the base of a nimbostratus, with which it may or may not merge. Deep nimbostratus clouds, like cumulonimbus clouds, can extend beyond the troposphere, even though their base is at the lower or middle levels. This type of cloud is closely related to cumulonimbus clouds because of their vertical characteristics and because both can be part of a precipitation-generating cloud complex. The area of precipitation due to a nimbostratus is extensive and restricts horizontal visibility and is known as stratiform precipitation.

Table 1. Classification of clouds according to the International cloud Atlas of the world meteorological organization (WMO) (taken from Houze, 2014).

		Height of Cloud Base
Genus	Group	Polar Regions	Temperate Regions	Tropical Regions	Cloud depth
Cu	Low	Below 2 km			1.5 km
Cb (DCC)					6–12 km
Sc					0.05–0.6 km
Ns					0.5–3 km
As	Middle	2 to 4 km	2 to 7 km	2 to 8 km

Note: (Cumulus-Cu, Cumulonimbus-Cb/Deep convective clouds-DCC, Stratocumulus-Sc, Nimbostratus-Ns and Altostratus-As).

For detecting and classifying different cloud types from passive satellite data, the methodology described by León-Cruz et al. (2021) was followed using the cloud categorization described by the International Satellite Cloud Climatology Project (ISCCP). The cloud climatology presented by the ISCCP is elaborated from a combined set of atmospheric products with a temporal resolution of 3 h and a spatial resolution of 10 km from geostationary and polar-orbiting satellites belonging to the National Oceanic and Atmospheric Administration (NOAA). With satellite data, it is possible to recognize the physical characteristics of clouds to distinguish between one type of cloud and another by obtaining the values of cloud top temperature (CTT) and cloud top pressure (CTP) and by the reflectance of a cloud according to its liquid or ice content, the cloud optical thickness (COT) (Rossow & Schiffer, 1999) (Figure 2). The COT can also be considered as additional information about the vertical development of clouds, estimating that this parameter is largely independent of cloud height. Thus, a satellite pixel covers a deep convective cloud (DCC) when the CTP is equal to or less than 440 hPa and the COT recorded equals or exceeds 23. A nimbostratus (Ns) cloud is detected with a CTP between 680 and 440 hPa (thus classified as a medium cloud) and a COT greater than 23. Cumulus (Cu), stratocumulus (Sc) and altostratus (As) clouds are detected when the CTP is greater than 680 hPa, and the COT is between 0 and 3.6, from 3.6 to 23, from 23 onwards, respectively. The choice of these five cloud types was based on the direct relationship that their presence in the atmosphere has with hydrometeorological events (Cohn, 2017). León-Cruz et al. (2021) explain that once the different types of clouds have been detected, it is possible to calculate the frequency with which they occur and their relationship with local characteristics such as topography. The cloud frequency per grid cell was obtained from the mean values of CTP and COT, considering that only cloudy pixels with successful CTP and COT retrievals were used without explicitly counting the cloud phase. Monthly values of the total number of observed DCC, Ns, Cu, Sc and As, or their frequencies, were obtained by summing the daily counts of each cloud type. For the analysis in this work, monthly, seasonal, annual and inter-annual averages of the 20 years of available satellite data were calculated, considering mainly the total cloud cover, the cover of each type of cloud, its frequency and its spatial distribution.

Figure 2. Cloud classification according to the International satellite cloud climatology Project (ISCCP), uses the parameters of optical thickness and pressure at the top of the cloud.

Results

Total cloud cover

Figure 3 shows the fraction of the total cloud cover of the MODIS data at a horizontal resolution of 0.1º (approximately 12 km). Each plot represents the monthly average coverage, including the five cloud types. The cloud cover in the northwest region of Mexico shows a seasonal behavior influenced by summer monsoon activity, mainly from June to September. At that time of the year, the cloud cover is located mainly in the south of the Gulf of California, Sonora and Sinaloa on the west coast of Mexico. At the beginning of summer in June, an increase in cloud cover occurs over the geographical location of the WSM associated with the flux moisture from the Eastern Pacific Ocean (Dominguez & Kumar, 2005). The flux of moisture inland increases the forced ascents of water vapor and convection activity at the mountains region of the WSM (Farfán et al., 2021). In July, clouds cover large areas of the continental coastal territory, the southern parts of the Gulf of California, and the Baja California Peninsula; the maximums values of cover occur on the coasts of Jalisco (Jal), Nayarit (Nay) and Sinaloa (Sin). Cloud cover maintains high values over WSM and the coast in August and September. At the end of the monsoon, the cloud cover decreases on the eastern Pacific coasts and along the northern part of WSM, characterizing the end of the rainy season and tropical cyclones in the western Pacific basins (Farfán et al., 2021). From November to March, cloud cover is observed in the northeastern Pacific region during winter when clouds associated with the North Pacific High drive the cool and wet air masses from the north and the central Pacific (Douglas et al., 1982). This dynamic in winter is associated with the intensification of the Jetstream, the transport of humidity toward the northeastern Pacific, which generates high cloud cover and rain in southern California and the Baja California peninsula. It is interesting to observe the transition months, April and May, there is a weakening of the jets and their movement towards high latitudes, showing a movement of cloud covers over all eastern coasts of the Pacific. This period also coincides with the intensification of the ITCZ northward, denoting the beginning of the hurricane season (Cavazos et al., 2020; Gutzler, 2004). Likewise, there is a relevant cloud cover in the northeast region of Mexico. The fractions of clouds that seasonally move during the monsoon also develop more significant coverage in the center and northeast of Mexico.

Figure 3. Monthly average of cloud cover fraction for 20 years (2001–2020), including five considered cloud species detected in MODIS.

Fraction by cloud type

Cloud amount refers to the fraction of the sky covered by clouds of a particular type or combination. Figure 4 describes the cloud amount for each type of cloud (DCC, Ns, As, Sc and Cu). In the total fraction of could cover that includes all genera, it is impossible to determine between factors such as the effect of top cloud height. The deep convective clouds are the most prevailing cloud type, with a frequency average of 0.23, followed by the cumulus (0.08). The presence of the other three cloud types (Ns, As and Sc) is lower, with mean cover values of only 0.03 for Ns and Sc clouds and 0.02 for As. Each cloud genera is associated with different dynamics and is vertically distributed differently in the atmosphere. The Gulf of California region presents a high percentage of deep clouds, mainly in summer, due to more convective activity due to the warming of the sea surface, a region with a monsoon-type circulation. Considering this fact, the mainly dominant clouds of high vertical development are expected (Comrie & Glenn, 1998). The formation of this type of cloud responds to daytime dynamics, with more significant storm activity associated with cumulus clouds and cumulonimbus clouds forming in the afternoon, around 4 to 5 pm (Figure S1). In winter, in January and February, the total cloud cover of Ns, Sc and As types reaches relatively high values, of the same order like in summer (Figure 4b). The interannual variability of the total cloud cover is related to the SST patterns, which show a behavior associated with these anomalies in the eastern Pacific basins, for example, the El Niño 3.4 region. The possible relationship of cloud cover with climatic indices such as ENSO is discussed later.

Figure 4. a) annual average (2001.2020) of the different types of clouds amount (deep convective clouds-DCC, nimbostratus-Ns, altostratus-as, stratocumulus-Sc and cumulus-Cu), and b) monthly averages for each type of cloud.

Deep convective clouds (DCC)

The DCC type prevails in all the years within the study period, with an average coverage of 0.23, representing 76% of the total cloud fraction (Figure 4a). In this sense, the spatial distribution of DCC overwhelms all cloud coverage in the entire NWM region. Figure 5 shows the monthly mean distribution of deep convective clouds averaged from 2001 to 2020. By observing the total cloud cover, it is now possible to notice that the variation in the cloud formation process with deep development occurs mainly in the summer months (JJAS), with a distribution from the south of the Gulf of California towards the continent along the WSM. This cloud formation is associated with increased moisture transport processes during the North American Monsoon (NAM) system (Comrie & Glenn, 1998; Higgins et al., 1997). An extraordinary humidity flow contributes mainly to convective activity during the establishment of the NAM. This deep convective development over northwestern Mexico rapidly propagates along the western slope of the WSM in early July (Gebremichael et al., 2007). Vertical moisture transport by convection increases over the northwest region of Mexico and in the channeled region of the Gulf of California. In addition, significant moisture convergence over the WSM and the eastern Pacific represents a source of moisture during the monsoon. It triggers cloud cover, including in the desert regions of northwestern Mexico and the southwestern United States (Dominguez & Kumar, 2005; Giovannettone & Barros, 2008). During October and November (transition months), the moisture flux exchange between the ocean-atmosphere decreases, inhibiting deep convection and less fraction of DCC (less than 49% of total cloud cover). The reduction of DCC occurs in almost all of Northwestern Mexico except the coasts of Jalisco and Nayarit.

Figure 5. Monthly average distribution of the deep convective clouds (DCC) of 20 years (2001–2020) where 1 indicates complete cover.

Cumulus clouds (cu)

Figure 6 shows the monthly average of the spatial distribution of Cumulus (Cu) for the period 2001–2020. Cumulus clouds cover the adjacent eastern Pacific coastal area. The fraction cover of Cumulus gradually increases on the Pacific Ocean and extends from south to north parallel to the Baja California peninsula. The maximum value of the Cu cloud fraction occurred during June, with an average value of 0.41. By considering coverage by clouds Cu and DCC, the cloudiness during the summer months (JJAS) is mainly the result of clouds with more vertical development, 96% of total cloud cover, while low clouds (As, Ns, Sc) had a contribution of less than 5%.

Figure 6. Monthly average distributions of the cumulus clouds (Cu) of 20 years (2001–2020), where 1 indicates complete cover.

Discussion

Incoming solar shortwave radiation

Clouds interact with many atmospheric parameters, primarily with shortwave radiation from the sun, thus playing a central role in climate behavior. Cloud climatology regulates seasonal variations of relevant climatic parameters of the system, like longwave radiation, topography and wind. The Changes produced in clouds feedback temperature, pressure, humidity and wind which in turn modify the clouds non-linearly. The magnitudes and possible climate changes associated with clouds have yet to be well known, but the long historical climate record can help to identify them (Warren et al., 1988). The shortwave solar radiation (SWR) at the surface is one of the parameters immediately affected by the presence or absence of clouds. The cover and the type of clouds in a given region control the process of the dispersion and absorption of solar radiation in the atmosphere. Therefore, more clouds covering the atmosphere causes low radiation on the surface, and most of this effect is determined by the thickness of clouds, mainly those with high vertical development (Cess et al., 1993). However, it is necessary to consider that the availability of solar energy also depends on geographical conditions and seasonal variations (Letu et al., 2020).

Figure 7 shows the spatial distribution of shortwave radiation from ERA5 reanalysis data at the surface in monthly averages for the entire study period (2001–2020). The seasonal behavior of solar radiation revealed maximum values from April to August and minimum values during winter. By analyzing the spatial distribution of the cloud cover, it is possible to determine its effect on the shortwave radiation. During the winter, the extensive cloud cover is located over the Pacific Ocean, causing a minimum of solar radiation, reaching 120 $w/m^2$ in that region. In the absence of clouds during May, mainly on the continent, maximum values exceeded 270 $w/m^2$. In spring and early summer (AMJJ), there is a high incidence of solar radiation. In contrast, the maximum presence of DCC-type clouds causes a minimum of radiation with significant differences compared with other regions of northwestern Mexico, like the WSM and the southern Gulf of California, that reach 100$w/m^2$. The spatial patterns of the summer radiation minima coincide with the maxima cloud cover of high vertical development (DCC and Cu) in the monsoon season. The dynamic due to the thermal contrast between the ocean-continent transition is a determining factor in the development of the Monsoon towards the northwest of Mexico and the southwest of the USA (Comrie & Glenn, 1998). Values of shortwave radiation exhibit that the maximums occur during the months before the development of the Monsoon, and mainly during July, the contrast decreases, forcing the decline of the Monsoon towards August and September.

Figure 7. Monthly average distribution of downward shortwave radiation ($w/m^2$) of 20 years (2001–2020) from ERA5 data.

Surface temperature

The oceans, the land and the atmosphere partially absorb the incident solar radiation; consequently, the variations in the radiation directly cause changes in the surface temperature (Martínez‐Díaz‐de‐León et al., 2006). The temperature at the surface is determined by the amount of energy absorbed by the oceans, land and atmosphere. Clouds play an important role in balancing the total radiation reaching the surface, causing unequal heating on land and oceans, and through feedback from the surface temperature, cloudiness varies too. The temperature anomalies show a seasonal behavior based on the annual variation of cloudiness (Figure 8). The Surface Temperature anomaly (STA) on land is negative (colder) in winter (from October to March) and positive (warmer) in summer (from April to September). The temperature anomalies in the Gulf of California and the eastern Pacific are not in phase with each other or continental land. It indicates that the ocean has its reactions to incident radiation. The displacement to the north of the positive anomaly also determines the beginning of the rainy season since warmer waters favor evaporation and cloud formation in this region.

Figure 8. Monthly average distribution of anomalies for land surface temperature (°C) and sea surface temperature (°C) from ERA5 data (2001–2020). The scale on the left refers to sea surface temperature, and the one on the right to land.

In July, August and September, the negative temperature anomaly along the WSM reflects the presence of DCC (Figure 5) in the same region that damped the shortwave radiation. In the Baja California peninsula and the northwest region, the solar radiation reaches up to 260 $w/m^2$, while along the WSM, the radiation (185 $w/m^2$) and temperature (anomaly of −6°C) decrease. In August and September, these conditions prevail. From April to September, the arid and semiarid parts of central and north Mexico exhibit high values in radiation (Figure 8), positive temperature anomalies, and low cloud cover (Figure 3). The thermal contrast between the ocean and the continent is maximum in April and May. In summer, the high cloud cover causes negative anomaly of temperature over the WSM due to the development of deep clouds (DCC and Cu) and mesoscale convective systems (MCS) (Farfán et al., 2021). The warmer waters traveling to the north and towards the Gulf of California determine the monsoon circulation, intervening in the balance of flux moisture from the ocean into the continent. The temperature anomalies correspond with this dynamic, causing an increase over the ocean and continent as a primary driver of the monsoon dynamic. There are also substantial anomalies on the western coast of the Baja California peninsula that decrease as warm water moves into the central Pacific. Soto-Mardones et al. (1999) discuss how these anomalies moderate surface ocean dynamic processes correlated with changes in the upper layer related to upwelling. Bedacht et al. (2007) explain that the eastern Pacific region is dominated by convective processes linked to SST variability, which drives cloud formation towards the northwestern part of Mexico. When SST increases, the circulation as Pacific Walker strengthens, leading to more cloud cover, as seen throughout the year.

Total and convective precipitation

We examined the spatial distribution of average monthly cumulative total precipitation (Figure 9 a, b, c) and convective precipitation (Figure 9 d, e, f) for the summer months (JAS) from the CHIRPS database between 2001 and 2020. The patterns defined by the DCC coverage correspond to the distribution of deep convective clouds for the summer months. The deep convective precipitation is known as the monsoonal type, where a circulation of moist and warm air flows toward the land, causing a rapid ascent and formation of deep convective clouds (Farfán et al., 2021; Valdés-Manzanilla, 2015). This rain is typical of the monsoon and the generated severe storms in that region (Pineda-Martinez et al., 2020). Interestingly, the region with deep convection clouds covers much of western and northwestern Mexico. It includes, in the continental part to the south of the domain, the Trans-Mexican Volcanic Belt and practically the entire region in which the WSM extends. Figure 9 reveals the importance of this distribution of deep convective clouds in climate regulation through precipitation and by the absorption of shortwave radiation (Figure 7) that regulates the temperature distribution (Figure 8).

Figure 9. Monthly comparison of summer (JAS) between total precipitation (a, b, c) in mm, convective precipitation fraction (d, e, f) in mm, and the deep convective cloud cover (g, h, i) in mm. All graphs represent monthly averages for the period of 2001–2020.

From the distributions of monthly averages of cloud cover fraction for 2001–2020 (Figure 3), from the distribution of the monthly average of deep convection clouds for the same period, and the distribution of the monthly average of cumulus clouds for the mentioned period, it follows that there are cloud covers that are not explained by deep convection clouds or cumulus clouds. In addition, observing the monthly averages (July, August, September) of total precipitation for the period 2001–2020 and the monthly averages of precipitation associated with deep convection, it can be deduced that other types of clouds must contribute to precipitation in north-western Mexico. Figure 10 reveals the monthly average of nimbostratus cloud cover in winter (a) (January) and in summer (b) (September) for the period 2001 – 2020. For the same months, Figure 10c, d show the associated precipitation to nimbostratus clouds. In winter (January), precipitation is significant on the eastern side of the WSM in Durango and in areas of the open Pacific Ocean. In summer, precipitation by nimbostratus clouds occurs mainly on the eastern side of Sonora, in the northern part of Sinaloa, south of the domain in the Trans-Mexican Volcanic Belt and in the coastal zone of the Eastern Pacific Ocean.

Figure 10. Nimbostratus cloud cover for January (a) and September (b) and total rain minus deep convective rain (mm) in January (c) and September (d). Averages from 2001 to 2020.

Climatic indices and convective activity

There is a relation between increased air temperature at different levels, more deep convective cloudiness, and precipitation over the central Pacific region with the positive phases of ENSO (El Niño South Oscillation). El Niño is a meteorological and climatic phenomenon that influences on a planetary scale. Northwestern Mexico is not exempt from being affected by the different phases of this phenomenon. Furthermore, the different phases of the El Niño phenomenon must be reflected in the massive process of cloud formation by the many forms of atmospheric circulation (convergence, convection, forced convection, instability, turbulence). We compare the ENSO phases (La Niña, El Niño and neutral phase) with the anomalies of deep convective clouds in annual averages for 2001–2020 (Figure 11). Data on ENSO phases were taken from the NOAA National Weather Service Climate Prediction Center. We found a correlation for the years with the lowest DCC values within the negative ENSO phases (La Niña). In contrast, the years with higher DCC numbers were associated with the positive ENSO phase. The years with the highest total cloud cover values and relevant deep convective activity were 2006, 2015, 2018 and 2019. Specifically, 2015 was described as an extreme year with the highest DCC data within 20 years, during the strong El Niño signal recorded in these two decades of study. These results agree with other authors (Dommenget & Yu, 2016); they show that the increase in SST in the eastern Pacific causes more formation of deep clouds, correlated with the positive SST anomalies. The years with less cloud cover and DCC were 2010, 2011 and 2017. These years correspond to the negative periods of the ENSO, especially in 2010 during an intense La Niña event.

Figure 11. Annual comparison between the deep convective cloud cover anomalies and the ENSO phases from 2001 to 2020. All years are classified into SST anomalies in agreement with de ENSO phase.

As expected, the annual average of shortwave radiation over the entire study domain and the annual average of the total coverage of the five cloud types reveal a dominantly negative correlation (Figure 12). Both show trend curves, but the total cloud cover trend line tends to increase, while the shortwave radiation trend line tends to decrease. In the case of the 2015 El Niño, the annual average of shortwave radiation was reduced by approximately 6 $w/m^2$. The maximum range of variation of shortwave radiation in the period 2001–2020 was between 190 $w/m^2$ and 203 $w/m^2$. In the same period, the total coverage of the considered clouds varied between 0.34 and 0.46.

Figure 12. Comparison of annual averages in the entire domain of shortwave radiation (broken line $w/m^2$, right side) and annual cloud cover averages also in the whole domain (continuous line, left side).

Conclusions

In an alternating land-water configuration in northwestern Mexico, a detailed estimate of the cloud cover distribution was carried out, considering the most relevant ones; deep convection clouds called Cumulonimbus (Cb), cumulus (Cu), nimbostratus (Nb), altostratus (As) and stratocumulus (Sc). Using MODIS sensor data, placed on the Aqua and Terra satellites, for 2001 to 2020, average monthly and annual cloud cover were calculated individually for each cloud type. It allowed having a first approximation to cloud climatology in this geographically and orographically variable area. The deep convection clouds (Cumulonimbus) seasonally dominate over extensive regions of NWM, and their distribution is closely related to the rainfall patterns. The patterns of monthly averages revealed basic information about their influence on climate regulation. The combined analysis of cloud distribution and short-wave radiation patterns explains the differences between areas where there is cloud cover and abundant vegetation and desert areas where clouds are scarce. The data analysis revealed that the rain associated with nimbostratus clouds contributes to a large extent to the precipitation observed in the northern part of the WSM, in the eastern part of Sonora, and in the western part of Chihuahua. Regarding the fraction of total cloud cover, the part of the Pacific Ocean to the west of Baja California is mainly due to the formation of cumulus, while deep convection clouds (cumulonimbus) prevail in the southern part of the Gulf of California and throughout the WSM dominating the southern part of the study domain. Annual deep convective cloud cover anomalies allowed a relationship with the different phases of the El Niño phenomenon (La Niña, Neutral, El Niño). The values of cloud coverage showed that in 2010, the deep convection cloud coverage anomaly was highly negative, clearly associated with the La Niña phase. On the contrary, in 2015, the anomaly was highly positive, associated with the El Niño phase. It was shown that deep convection cloud cover considerably mitigates the incidence of shortwave radiation and plays a regulating role in temperature and climate in general. The deep convective clouds are the dominant cloud type, with a frequency average of 0.23, followed by the cumulus (0.08). The mean cover values of the other three cloud types (Ns, As and Sc) are lower, only 0.03 for Ns and Sc clouds and 0.02 for As.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the Instituto Potosino de Investigación Científica y Tecnológica .