Introduction

In recent years, increased attention has been paid to the large amounts of airborne dust derived from deserts (Liu et al. 2003; Goudie and Middelton 2006). Desert dust responds to three phases (input, transport, and output) dependent on wind erosion, subject to atmospheric conditions and soil properties, characteristics, and use (Shao 2008). Suspended and airborne particulate matter plays an essential role in climate (Shao 2008; Wang et al. 2017; Kok et al. 2017) and can affect human health (Chen et al. 2004; Sprigg et al. 2014). When deposited in the ocean it has implications on marine food webs (Mahowald et al. 2014; Wang et al. 2015; Muñoz-Barbosa et al. 2017) and Sea Surface Temperature (SST) (Evan et al. 2009; Martínez-Avellaneda 2010).

Studies on the dust cycle in the northwestern region of Mexico have been detected in satellite images, plumes of this material, coming from the BCP and heading towards the GC and the Pacific Ocean (PO) (Álvarez 2014; Morales-Acuña 2015). In addition, simulations in numerical models have quantified deposition cups of 0.25 \({\times 10}^{-5}\) y 6.64 \({\times10}^{-5}\mathrm{Tg}\;\mathrm{km}^{-2}\mathrm{yr}^{-1}\) for North America (Zender et al. 2003; Ginoux et al. 2004; Tanaka and Chiba 2006; Ginoux et al. 2012), and parameterization schemes for dust emissions yielded values of 23.84 and 169.93 \(\mathrm{mg }{\mathrm{m}}^{-2} {\mathrm{d}}^{-1}\), for the San Felipe and Vizcaíno deserts, respectively (Morales-Acuña et al. 2019a, b).

In the northern region of the BCP, meteorological events have not been studied; these are known in popular slang as "El Torito" and take place during the summer seasons. "El Torito" generally causes dust storms, rainfall, and increases in wind intensity, thereby affecting fishing activities, causing accidents and tragedies, such as the one reported in various portals and news media in the country on July 3, 2011 (see, https://www.uniradioinforma.com).

Due to the lack of studies documenting the dust storms provoked by "El Torito" and the climatic factors that give rise to them, through the analysis of an event that occurred on June 14, 2016, the present work aims to understand if the occurrence of these dust storms responds to fluctuations of hydrometeorological variables and the displacement of pressure systems. It becomes necessary to have a quantifiable approach that allows establishing strategies to improve air quality and ecosystems found in this region. In this work, the atmospheric dispersion of dust from the BCP to the GC is analyzed by characterizing dust events from satellite products, reanalysis data, and simulations in a model for the trajectory of particles in the atmosphere.

Data and methods

Study area

The northern region of the BCP is in northwestern Mexico between the PO and the GC (Fig. 1). The classification of the approximate distribution of bioclimatic floors elaborated by Peinado et al. (1993) establishes that for the coast adjacent to the PO, there are infra-Mediterranean, thermo-Mediterranean, and mesotrophic environments, while for the coasts adjacent to the GC, this classification establishes supra-Mediterranean and mesotrophic environments. A distinguishing factor of this region based on its climatic persistence and predominant vegetation is the desert zone or San Felipe ecoregion (González-Abraham et al. 2010). The San Felipe Desert is a low mountainous area dominated by extensive alluvial bajadas, gravel plains, sand, and a dune system (González-Abraham et al. 2010); coupled with these characteristics, the wind features (intensity and direction) that favor the emission of desert dust, the type of soil and the predominant vegetation establish this desert region as a source of dust (Morales-Acuña 2015). San Felipe is one of the hottest and driest deserts in North America; summer temperature exceeds 45 °C, and average precipitation reaches 14 mm yr−1.

Fig. 1
figure 1

Study area. The region's topography is shown in color. Red dot is the San Felipe desert region

Full size image

HYSPLT model description

The HYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) model uses a particle or puff approach to calculate trajectories, dispersion, and deposition (Stein et al. 2015). A calculation method is a hybrid approach that combines the eulerian (uses a fixed three-dimensional grid a reference frame to calculate air pollutant concentrations that are concentrations calculated at each grid cell using the integration of fluxes due to advection and diffusion) and lagrangian ( =) approaches. More details can be found at http://www.ready.noaa.gov/HYSPLIT.php (NOAA Air Resources Laboratories). In the present work, the HYSPLIT model is used to analyze the possible horizontal and vertical trajectories of particles emitted during the June 14, 2016, dust event from 0.3° resolution meteorological data.

Because HYSPLIT calculations are sensitive to input parameters (Gebhart et al. 2005; Hernández-Ceballos et al. 2014; Su et al. 2015; Khalidy et al. 2019), meteorological inputs from the North American Regional Reanalysis (NARR) data of the National Center for Environmental Prediction (NCEP) whose spatial resolution is 0.3° and temporal resolution every 3 h were used for our simulations. Considering that dust sources in this region have been previously identified by Morales-Acuña (2015), the forward trajectory option with 12 maximum trajectories and a matrix of 8 coordinate positions 114°W—29°N in the lower-left corner (see Fig. 1) and 113°W—30°N in the upper left corner (see Fig. 1) was used to determine the dust particle trajectories. Furthermore, the vertical velocity model was selected for three height levels (50, 10, 0 m) in the vertical motion. The simulation started at 9 A.M on June 14, 2016, and the total simulation time is 24 h.

Observational data and methods

To describe the meteorological conditions that gave rise to the "El Torito" event of June 14, 2016, data were taken one day before, during, and one day after the event. Then, from daily sea surface temperature (SST) imagery, provided by MODIS Aqua level 3, derived from the 11 and 12 µm thermal (IR) band, equivalent to channels 31 and 32, with a spatial resolution of ~ 4 km and available at the website https://oceancolor.gsfc.nasa.gov/, and daily atmospheric pressure, wind (Vel) and relative humidity (RH) data provided by the North American Regional Reanalysis (NARR), with a spatial resolution of ~ 32 km and available at ftp://ftp.cdc.noaa.gov/Datasets/NARR/, maps were made to represent the spatial and temporal distribution of each variable. Subsequently, to identify the associations between these variables, data were extracted using a longitudinal transect with fixed latitude at 30.5°N and spanning from the PO region to the GC, right at the "El Torito" event (see Fig. 1). Then, following the criteria set forth by Santamaría-del-Angel et al. (2011), a mathematical association analysis (principal component analysis, PCA, in its numerical representation) was performed. First, in general terms, a PCA was applied based on a correlation matrix of all the Z-transformed variables (calculated as the data minus the mean and this difference between the standard deviation), after which a PCA was applied based on a binary coding matrix of the Z transformations for the study variables; in this coding, low values were considered those with Z < 0 and high values those with Z > 0. Finally, considering these coded data, a PCA was applied. The significance criterion for each component was its value (eigenvalue); if the value was > 1.0, the component was considered significant, and if it was < 1.0, it was not significant (Fig. 2).

Fig. 2
figure 2

Schematic diagram of the principal component analysis by numerical resolution proposed by Santamaría-del-Angel et al. (2011). Z is the transformation of the databases to standardized spatial anomalies and EV indicates the eigenvalues

Full size image

To identify the source and Spatio-temporal distribution of dust storms in the northern BCP, daily aerosol optical thickness (AOT) imagery was employed for June 13, 14, and 15, 2016. The AOT is a standard MODIS product at 869 nm, with a spatial resolution of 1 km, available from the website https://oceancolor.gsfc.nasa.gov/l3/.

Results and discussions

Description of "El Torito" weather conditions on June 14, 2016

The daily sea surface temperature images allow identifying a thermal contrast of ~ 8 °C between the PO and the GC (Fig. 3), as reported by Turrent et al. (2014), Morales-Acuña (2015), and Morales-Acuña et al. (2019a, b), during the summer seasons. In addition, the thermal contrast that predominates in the region during this period, it can be identified for the PO there is a decrease in SST of ~ 7 °C on 14 June (Fig. 3b), compared to the other two days, which intensifies the thermal contrast between PO and GC during this day. Longitudinal SST transects for the study period (Fig. 3d, e, and f) allow us to identify SST fluctuations in the PO region. In the GC, the SST tends to remain constant during the study period.

Fig. 3
figure 3

SST spatiotemporal distribution and longitudinal transect (black dashed line), June 13 (a), 14 (b) and 15 (c). Longitudinal transects of SST (solid black line) and BCP region (dashed red line) for June 13 (d), 14 (e) and 15 (f), 2016

Full size image

In addition to the thermal contrast between the PO and the GC, the existence of a longitudinal west–east decrease for sea-level atmospheric pressure (Fig. 4a, b, and c) allows us to identify predominant values of ~ 1017 hPa from the PO to the central region of the BCP and ~ 1014 hPa from the coastal region of the BCP to the state of Sonora during June 13 and 14 (Fig. 4b and c). Longitudinal transects for atmospheric pressure during June 13, 14, and 15 (Fig. 4d, e, and f), allow us to identify the decrease shown above. The slope calculations for this study period present values of \(-3.18\times {10}^{-5}\), \(-2.57\times {10}^{-5}\) and \(-2.60\times {10}^{-5} \frac{\mathrm{hPa}}{\mathrm{m}}\), for June 13, 14 and 15, respectively.

Fig. 4
figure 4

Spatio-temporal distribution of atmospheric pressure and longitudinal transect (black dashed line), June 13 (a), 14 (b), and 15 (c). Longitudinal transects of atmospheric pressure (solid black line) and BCP region (dashed red line) for June 13 (d), 14 (e), and 15 (f), 2016

Full size image

In response to the thermal contrast and the decrease in atmospheric pressure, air masses coming from the North Pacific are redirected towards the BCP and channeled through the mountain lows, where they increase their speed up to 9 m s-1 (Fig. 5a, b, and c), as reported by Morales-Acuña et al. (2019a, b). The predominant wind direction in the BCP region, during the study period, obeys that reported by Turrent et al. (2014), Morales-Acuña (2015) and Morales-Acuña et al. (2019a, b). Furthermore, in contrast to day 13 (Fig. 5a), during 14 and 15 June (Fig. 5b and c), wind intensity is observed to increase (~ 1 \(\mathbf{m}{\mathbf{s}}^{-1})\) in the PO region. The longitudinal transects (Fig. 5d, e, and f) allow us to identify that the wind intensity increases up to ~ 1 \(\mathbf{m}{\mathbf{s}}^{-1}\) during June 14 and 15, except for the coastal region of the BCP adjacent to San Felipe, where a decrease of ~ 1. 3 \(\mathbf{m}{\mathbf{s}}^{-1}\), about this region Morales-Acuña (2015) established that this was characterized as a source of dust for northwestern Mexico, indicating that under appropriate wind conditions (intensity and direction), events such as dust storms can occur.

Fig. 5
figure 5

Spatio-temporal wind distribution and longitudinal transect (black dashed line), June 13 (a), 14 (b) and 15 (c). Longitudinal wind transects (solid black line) and BCP region (dashed red line) for June 13 (d), 14 (e) and 15 (f), 2016

Full size image

It can be observed that the air entering through the BCP loses humidity so that on the western coast of the GC, air trails with relative humidity < 50% can be perceived (Fig. 6). These atmospheric conditions (increased wind intensity and decreased relative humidity) in the BCP optimize the process of erosion and aeolian transport of dust from dust sources to the GC. Therefore, the longitudinal slices of the RH (Fig. 6d, e, and f) do not show relevant fluctuations during the study period.

Fig. 6
figure 6

Spatio-temporal distribution of relative humidity and longitudinal transect (black dashed line), 13 (a), 14 (b) and 15 (c) June. Longitudinal transects of relative humidity (solid black line) and BCP region (dashed red line) for June 13 (d), 14 (e), and 15 (f), 2016

Full size image

Considering that the significance criterion for all the analyses showed that only one of the components had results of ≥ 1 eigenvalue, a component is presented to describe each day (Table 1), this allowed to identify with a reliability of 95%, for June 13 (see column 1, Table 1) that low SSTs are associated with low values of wind intensity, high RH and high atmospheric pressure, while high SST values are associated with high values of wind intensity and low values of RH and atmospheric pressure. For June 14 (see column 2, Table 1). The low SSTs are associated with high values of wind intensity, RH, and atmospheric pressure, while high SST values are associated with low values of wind intensity, atmospheric pressure, and RH. Finally, on June 15 (see column 3, Table 1), the associations presented on June 14 were maintained. Results obtained in the associations of the variables studied are consistent with the characteristics exposed by the local inhabitants when the "El Torito" event is related to increases in wind intensity and humidity transport.

Table 1 Mathematical association analysis. The percentages of variability explained (PVE) for the days before (June 13), during (June 14), and after (June 15) the dust event is presented. For all analyses, the significance criterion yielded that the first component is significant with an eigenvalue ≥ 1
Full size table

Desert Dust Storm, June 14, 2016

The AOT is a satellite product commonly used as proxies for the identification of atmospheric intrusions of desert dust (DD) (Prospero 1996; Kalashnikova and Kahn 2008; Kassianov et al. 2017; Bullard and Baddock 2019) and validation of models for the study of this same variable (Liu et al. 2003; Wang et al. 2011; Chen et al. 2014; Heinold and Tegen 2019). The response of the meteorological conditions exposed above allows us to identify from the spatial and temporal distribution of the AOT (Fig. 7) that on June 14, 2016 (Fig. 7b), the "El Torito" event, characterized by transporting humid air masses and with higher wind intensities. When these air masses interact with the land surface, it gives rise to a DD emission that is then transported towards the GC until reaching the central region where it appears to be redirected to the north; the values for the AOT are up to 0.14 and are distributed between the coastal region, center, and north of the GC. These values are comparable with those reported by Llamas et al. (2013) for northwestern Mexico during June 2002–2003 (AOT < 0.24) and the mean values reported by Raman et al. (2016) in North America during May–June 2005 to 2014 (mean AOT = 0.13). After the atmospheric intrusion of DD (Fig. 7c), it appears that the size of the DD particles remaining in the peninsular zone did not allow the emission process due to the suspension mechanism; for such reason, no relevant atmospheric intrusions of DD are identified.

Fig. 7
figure 7

Spatial and temporal distribution of 869 nm AOT for June 13 (a), 14 (b), and 15 (c), from products provided by MODIS-Aqua

Full size image

To thoroughly examine dust transport from the BCP during the June 14 event, the HYSPLIT model with the forward trajectories option is applied (Fig. 8). The simulation indicates that dust-laden air parcels are transported from the foothills of the San Felipe region into the north-central region of the GC and then redirected into northern Mexico and the southern United States. Because the trajectories do not present recirculation as in the work of Morales-Acuña (2015), we can infer that high dust concentrations would not occur in the region of the trajectories. It can also be evidenced that the vertical distribution of the dust plots is dominated by trajectories with heights between 0 and 500 m during June 14, demonstrating that the emission process was at the canopy layer level. Furthermore, from 22:00 h on June 14, it is noticeable to see how the height of the dust plots increases as the trajectory progresses as a function of time; such variations in the vertical trajectory concerning time have been reported in works such as those of Ashrafi et al. (2014), Ganbat and Jugder (2019).

Fig. 8
figure 8

Horizontal and vertical departures of a dust plot simulated with HYSPLIT for June 14, 2016, under the trajectory approach. The simulation starts at 9 o'clock and the simulated time is 24 h

Full size image

Conclusions

The analysis of the hydrometeorological variables identified that a decrease of ~ 7 °C in the surface temperature of the Pacific Ocean off the PBC on June 14 causes an increase in the thermal contrast between the PO and the GC, which leads to a longitudinal (west–east) decrease in atmospheric pressure. In addition, the air masses coming from the North Pacific are redirected towards the PBC and channeled through the mountain lows, where they increase their average intensity by up to 1 m s−1. Due to the interaction of humid air masses coming from PO with the orography of the PBC, a decrease (< 50%) in relative humidity is produced, such that, on the western coast of the GC, dry air trails are perceived. The mathematical association analysis for day 14 suggests that SST could be a control variable that favors wind erosion events and atmospheric transport of dust from the deserts located in the PBC to the GC. The analysis of the spatial distribution of AOT 869 for June 14 established the existence of a dust plume whose origin is in the foothills of the San Felipe desert and extends between the central and northern regions of the GC. Finally, the model simulations for the trajectory confirmed that the distribution and geometry of the dust plume are due to displacement of the air masses in that area of GC.