1. Introduction
The similarity between tropical-like cyclones (TLCs) and tropical cyclones leads to their intensification being interpreted as driven by the latent heat release associated with convection and interaction at the air–sea interface. However, an important component of the early stages of TLC formation is related to baroclinic instability, generated by strong tropospheric depressions [
1,
2]. TLCs (as well as tropical cyclones) in their maturity phase experience wind-induced heat exchange with sea surface flows (a phenomenon generally called the
WISHE effect) [
3,
4] as the main development mechanism. Indeed, TLCs are characterized by the presence of an eye zone that identifies the predominantly warm core, fed by maximum heat exchange in the proximity of the surface. The presence of a weak vertical wind shear toward outermost regions generates strong rotation of an eyewall around the pressure minimum in the core, from which different intensity rainfalls and consequent sea storms extend [
5]. A classification of ‘Medicanes’, depending on the intensification mechanism, is consequently proposed in [
6]. TLCs of
Category A are formed from a large amount of energy transferred from air–sea interactions; the result is a self-sustainable vortex that reaches the tropical-like phase in a barotropic environment, while baroclinicity influences it only in the initial stage. TLCs of
Category B never evolve in a fully tropical-like structure but are related to the effect of strong upper-level potential vorticity streaming into a baroclinic environment. Lastly, TCLs of
Category C are considered the most common; they develop from small-scale vortices to large synoptic-scale vortices within a cyclonic circulation, and the tropical-like transition is associated with the potential vorticity in the large cyclonic structure.
Different phases of intensification and transitions of a Medicane are explored in [
7], starting from an extratropical cyclone, as a Mediterranean perturbation. A tropical-like transition with a warm core formation occurs, resulting from a warm-air mass transported from the main atmospheric flow to the center of the cyclone. This process can lead to strong convective activity near the central area of the Medicane, associated with diabatic heating [
8,
9].
From different models based on the diagnostics of TLCs, the warm core and deep convection are identified as the main features of their intensification. In some cases, the warm core is confined to the lowest parts of the troposphere [
10,
11], while in others, it is associated with low-level diabatic processes that facilitate vertical motion during deep convection close to the center, intensifying the low-pressure center through latent heat release in moist ascent [
12].
In the study of the formation and development of TLCs, one of the most strongly considered patterns is the connection between surface energy fluxes and the low temperatures generated at upper levels during cyclogenesis. This relationship contributes to cooling and moistening across all geopotential altitudes, increasing the air–sea gradient of saturation moist static energy [
13]. The same model identifies what happens during cyclogenesis, in agreement with the mechanisms of spatial self-aggregation in the presence of convective motions associated with tropical cyclones [
14,
15]. As opposed to tropical cyclones, the duration of TLCs’ action is limited to a few days, due to the confined extent of the Mediterranean Sea, which is their main source of energy. In addition, they range up to a few hundred kilometers and reach completely tropical characteristics only for a short time: the intensity rarely exceeds category 2 of the Saffir–Simpson scale [
16,
17]. Taking it into account, the German Meteorological Service proposed an unofficial classification based on the average wind speed peak of intensity
v, following the Saffir–Simpson scale for tropical cyclones [
18]: when
km/h, the event is defined as a
Mediterranean tropical depression, when 64 km/h
km/h, it is defined as a
Mediterranean tropical storm, and when
km/h, it is defined as a
Mediterranean hurricane, e.g., Medicane.
TLCs and their features are the main topics of research in MedCyclones—the European network for Mediterranean cyclones in weather and climate [
19].
In relation to the diagnostics of Medicanes, Panegrossi et al. (2023) [
20] proposed an approach based on the use of satellite tools to identify and describe their warm core and deep convection. This approach exploits passive microwave (PMW) radiometers for exploring the features and properties of the phases and structures of several TLCs.
Satellite-borne radiometric payloads are the appropriate investigation means for the synoptic characterization of the thermal components of these processes. In this work, we focus on images acquired by three satellites: Sentinel-3A/B, with their sea and land surface temperature radiometer (SLSTR) payload, and Suomi NPP, with its visible/infrared imager radiometer suite (VIIRS) sensor [
21,
22]. All of them orbit in low Earth orbit (LEO), allowing data acquisition with a low-to-moderate spatial resolution (nadir pixel
and 10 km), on total globe coverage at a moderate temporal resolution of 1–2 images/day, which fits the goal of this study.
The evolution of Ianos and Apollo TLCs is monitored by exploiting brightness temperature for each sensing time. Starting from the radiation amount taken from the sensor, brightness temperature is used to extract other meaningful atmospheric fields, providing an improved assessment of the TLC cloud systems. Satellite observations are analyzed by deriving cloud top altitude and temperature, vertical temperature gradient, and atmospheric pressure field [
23,
24]. Altitude and temperature are obtained by considering drops in air temperature and the dew point in standard atmospheric conditions, while sea surface temperature is taken into account in the evaluation of the atmospheric instability through the vertical temperature gradient. The same standard conditions are leveraged to extract pressure from the images. Air temperature and sea surface temperature, in our procedure, involves the daily recorded data obtained as outputs from the
BOLAM forecasting model [
25,
26]. From the same model, mean sea level pressure (MSLP) values are considered for comparison with values extracted for the pressure around the eye of the TLCs. The brightness temperature from two different spectral channels of VIIRS and SLSTR acquisition adds important information to describe the behavior of convection within the regions of the cyclonic structures, looking at the deep convection cloud pixels [
27,
28,
29] and the ice percentage ones in the system.
In the last part of the study, a 3D visualization is implemented by using
ParaView software version 5.11.0 (
https://www.paraview.org/ (accessed on 8 April 2024)) in order to investigate the evolution of Ianos. In this analysis, a database larger than the previous ones is exploited, acquiring the Ianos images by the spinning enhanced visible and infrared
imager (SEVIRI) sensor, on the rapid scan high-rate service onboard Meteosat second
generation-11 (MSG-11) satellite platform [
30]. MSG-11 is a geostationary satellite platform and, if compared with LEO ones, it allows images and data with a lower spatial resolution (pixel size between 3 and 13 km) on a total globe coverage, with a very high temporal resolution (each image is obtained in a 5 min timestep in the rapid scan high-rate SEVIRI service). Moreover, 384 products are acquired, representing visualizations every 5 min one a one-day total period of Ianos’s evolution (on 17 September). In this database, the temperature field is contained in the pixels of the products, capturing all the timesteps of Ianos in the z-axis of the rearranged 3D view.
Ianos represents the most powerful cyclone observed in recent years in the Mediterranean basin [
31]. It emerged from a cyclogenetic area connected to strong thunderstorms near the Gulf of Sidra, Libya, early on 15 September 2020. A rapid intensification occurred in the subsequent hours, reaching a very significant pressure minimum and a strong potential to acquire tropical features, powered by sea temperatures. Ianos gradually intensified over the Ionian Sea from late 16 September to early 17 September, acquiring the eye structure and Medicane features. The peak of intensity was reached near the Greek coasts in the first three hours of 18 September, when Ianos became equivalent to a second-category hurricane on the Saffir–Simpson scale. In the next two days, the structure started a weakening trend after making landfall in Greece, until it completely dissipated near the coast of eastern Libya.
Apollo, a little more than one year after Ianos, developed into a significantly strong Medicane with a wider track, encompassing two-thirds of the Mediterranean over a life period of about 10 days, between 25 October and 2 November 2021 [
32]. Originating from a tropical disturbance in a thunderstorm area near the Balearic Islands, Apollo developed a low-pressure center in the first few days, intensifying and moving toward the Tyrrhenian Sea. The peak of intensity was reached on 29 October, after enduring strengthening off the coast of Sicily. Its convection slowed down for the first time on 30 October, and the dissipation trend ended a few days later, with residual strong winds pushing the system toward the Turkish coast.
2. Methods
The performances and capabilities of sensors onboard satellite platforms represent developing tools for observational approaches. With technological advancement, the potential of satellites is being fully exploited to monitor the increasing instability related to the effects of climate change on different global scenarios [
33]. This study highlights a new perspective on diagnosing extreme events related to atmospheric mesoscale over the Mediterranean basin using a satellite remote sensing approach. As previously acknowledged for studying the diagnostics of a tropical cyclone at its maximum intensity category [
23,
24], a similar methodology is employed for comparing two Medicanes, aiming to seek atmospheric characterization from the medium-high resolution satellite data provided by two important LEO sensors. Their technical specifications for imagery acquisition are summarized in
Table 1.
The choice of the satellites leads to a new analysis of the evolution of event dynamics, performed through the calculation of the atmospheric parameters of interest. For SLSTR, onboard the Sentinel-3A/B satellite missions, the descending sun-synchronous orbit crosses over the equator at 10:00 UTC, capturing the pixel with an oblique conical scanning in two different swaths of revisiting interval. On the other hand, VIIRS is characterized by an ascending sun-synchronous orbit, crossing over the equator around 13:25 UTC, with parallel lines pixel scanning.
Imagery acquisition is specifically chosen to assess the two TLCs’ temporal evolution and their spatial displacement over Mediterranean latitudes with the best precision. Generally, according to Wien’s law, TIR spectral channels are the most appropriate to estimate the temperature of the Earth’s surface (LST—land surface temperature). At 27 °C (300 K), this parameter will have a wavelength peak emission at about 10 m, while wavelengths between 11 and 12 m and 15 m are more suitable for evaluating the cloud’s temperature and formation properties, up to °C (220 K). Therefore, for this analysis, the TIR spectral channels of the two sensors are taken into account, which are identified with the resolution bands S8–S9 for SLSTR and M15–M16 for VIIRS. Central wavelengths for TIR1 bands (S8 and M15) are m, while the ones for TIR2 bands (S9 and M16) are m.
An initial check of the raw product marks the starting point of the analysis. In this pre-processing phase, the primary step involves optimizing the information in the pixels of each image. Different boundary conditions in the
grid visualization lead to the application of resizing and pixel resampling corrections. In particular, each image undergoes georeferencing [
35], involving geographical correction to achieve uniform resized images through a
nearest neighbor pixel resampling algorithm and a reprojection on the ellipsoid system UTM WGS-84.
After this first phase, the images are processed to compute the atmospheric parameters of the events. The grid of pixels in these newly resampled products contains data in brightness temperature,
. Although this metric lacks direct physical significance,
depends on the central wavelength (hence, the specific TIR spectral channel) and is proportional to the radiation amount recorded on the satellite sensor,
, at the moment of acquisition. For this reason, since the relationship is described by Planck’s law [
36,
37] (Equation (
1)),
denotes the thermal response of the total radiation recorded at the sensor, neglecting its reflection contribution along the object-satellite line of sight:
where
J s is Planck’s constant,
m/s denotes the speed of light, and
J/K is Boltzmann’s constant.
data offer initial insights into the regions observed within the satellite imagery, depending on the wavelengths of both spectral channels. To enhance the observed scene, a temperature map is applied to each image, in which the total range of values is divided into 15 sub-ranges, assigning them different colors. This tool can help identify objects in the scene according to the range values, representing the first step of focusing on the cloud cluster characterizing the cyclone.
Following this initial step, the TIR2 resolution band of the two sensors is selected to derive the first two atmospheric parameters. We propose evaluating the scene by computing the altitude
H from the top of the observed cloud system, and the associated temperature field
. Their derivation is based on the coupling of two linear equations (Equation (
2)), from which, values of
H and
are significant for the physical representation of the satellite view, particularly for the clouds associated with the cyclonic system:
The calculation involves two constant numerical parameters, representing the conditions of air temperature drop (9.84 K per 1000 m of altitude, in K/m units) and dew point drop (1.82 K per 1000 m of altitude, in K/m units) [
38]. These constants allow deriving a real temperature value, providing
, given by the daily averaged output value of air temperature at 2 m above mean sea level (AMSL), acquired from the
BOLAM forecasting model database. The overall calculation of
H and
must be completed by a generally known criterion used for assuming that the satellite-recorded
values, under suitable instability conditions, can coincide with the dew,
, at the altitudes of the condensation level [
24]. With this procedure, by imposing
, system (
2) is solved, and the quantities
H and
are obtained (Equation (
3)).
An evaluation of atmospheric instability conditions is carried out from the TLC computed values of H and
. The investigation consists of computing the values of the vertical temperature gradient
by studying the relationship between the temperature’s positive variation in the layers of the cyclone and the altitude, expressed by Equation (
4).
In this equation, the quantity,
, represents the daily averaged sea surface temperature (higher than
in all the scenes, except for the values associated with the sea surface area), obtained as outputs of the
BOLAM forecasting model database, in order to obtain a local estimation of the temperature variation with the cloud stratification. Following this interpretation, the derived values can describe how the instability evolves with the daily Medicanes evolution if compared to the gradient limit values that—in an atmospheric general context—identify the degree of instability, as follows:
- (1)
K m−1 (absolute atmospheric instability), in which the numerical value is the dry adiabatic vertical gradient;
- (2)
K m−1 K m−1 (conditional atmospheric instability), in which the vertical temperature gradient is included in the range’s moisture-to-dryness vertical gradients.
The next step entails calculating the atmospheric pressure field. In particular, considering the previously obtained altitude values alongside the standard atmospheric sea level pressure as
hPa, the pressure field can be derived using the exponential law (Equation (
5)).
where
In Equation (
6),
represents the scale height parameter,
T the temperature,
the specific gas constant in dry air, and
g the gravitational acceleration. As the initial assessment to compute the pressure field, a constant value of the scale height equal to
Km is selected, representing the typical air temperature under general standard conditions of the Earth’s atmosphere (
K) [
16,
17]. This step is supplemented by examining the pressure field in the area surrounding the eye of the cyclone, on the peak intensity days of Ianos and Apollo, followed by a direct comparison with the MSLP output data recorded in the
BOLAM databases.
As mentioned above, several studies point out the important role of convection mechanisms in the process of cyclogenesis. One approach to identifying convection, as proposed by [
27,
28,
29], involves the detection of pixels associated with deep convection clouds, in a direct way from the scene that highlights the different phases of the cyclone. Accordingly, we propose observing convection for the two Medicanes by exploiting the brightness temperature range, 230 K
K, of the georeferenced images displayed in the TIR1 spectral channel.
To validate the presence of strongly convective clouds in the structure of the Medicanes, the
split window spectral method is applied [
39], which is generally associated with cirrus cloud detection in the system, and is hereby proposed to search for the ice percentage pixels in the cloud layers of the cyclone:
The method is based on the difference in brightness temperature values of both spectral channels of the TIR, as expressed in Equation (
7). Each image is the result of subtracting the two TIR views of the same scene, where the values within the range of
K
K identify the pixels of ice in the clouds.
By exploiting some technical features of the
ParaView software, it is possible to obtain a 3D visualization of the temporal evolution of Ianos, precisely by considering time as the third vertical dimension, together with the two horizontal dimensions provided by the longitude and latitude in the SEVIRI images. The database products are characterized by a 5-minute timestep over a one-day total period of the cyclone’s life, with a spatial pixel size of 4 km. Each image is the result of a pre-processing step that allows locating the cyclone in its impact area, representing the temperature field (extracted by solving Equation (
3)) in the TIR spectral channel of SEVIRI (central wavelength
m).
4. Discussion and Conclusions
Satellite remote sensing tools represent fundamental resources for scientific research on the characterization of a cyclonic system in Mediterranean latitudes, facilitating a detailed investigation of such complex meteorological events. Satellite-collected data facilitate the study of Mediterranean cyclone formation and development, particularly by observing clouds, air currents, and thermal variations with advanced space-based instruments.
Satellite payloads, providing information in TIR spectral bands, enable studying these kinds of events on different life days through the characterization of their cloud systems, with a focus on stratification and rotating air mass circulation. Atmospheric parameter derivation primarily relies on radiance-at-satellite responses, providing insights into upper cloud layer atmospheric conditions.
LEO platforms, such as Suomi NPP and Sentinel-3, in addition to the GEO MSG-11, yield data about the brightness temperature field in the TIR spectral channels, by observing large and intermediate scales of cyclones from the top of the cloud [
23].
Daily and data recorded by the BOLAM forecasting model allow for the assessment of the vertical thermal gradient, highlighting the central large local stratification in different regions, and its correlation with the instability degree for both Medicanes examined in this work.
BOLAM-recorded data are also useful in analyzing the atmospheric pressure field extracted from the images. A comparison with the daily MSLP shows consistency with the satellite-extracted pressure values, identifying the area of the cyclonic warm core in both cases. The derivation of pressure values in the eye area (or in the surrounding areas of the warm core) is consolidated by studying the daily spatial trend of the pressure, even if not all associated graphs are shown. However, it can be seen that the pressure field obtained from VIIRS data appears to be slightly closer than SLSTR, with respect to those recorded. For Apollo Medicane, estimates of MSLP that were obtained from
BOLAM, VIIRS, and SLSTR can be compared with the results obtained by Menna et al. [
40]. Comparing their
Figure 2 (panels b,c), the values extrapolated from the plots, from 26 October to 30 October, yield results absolutely similar to ours reported in
Table 7, particularly in the measure trends and the absolute values, with relative differences, ∼1 %.
As in the analysis for the case study of a tropical cyclone [
24], the high-quality scanning technique operated by VIIRS, combined with daily nighttime acquisition at 00:00 UTC, enhances accuracy in the results by minimizing reflected radiance effects.
Additionally, the fusion of TIR spectral channels from multiple sensors can help to identify the image pixels associated with the clouds responsible for deep convection. By replicating the procedure previously used in [
27,
28,
29], it can be seen that convection evolves in most cases within the clouds rotating around the central structure of the inner core. This helps provide a qualitative contribution to understand the dynamics of the two cyclones. Furthermore, using the
split window technique, the absence of ice percentage pixels associated with the cloud systems on the TLCs’ life days is revealed. Future efforts will involve exploring new satellite data sources and potential combinations with operational PMW sounders to refine parameter extraction for mapping Medicanes from the inner part, as suggested in [
20]. Further improvements in the efficiency of this type of study can be achieved by comparing and combining data from observations with data obtained from numerical simulations, reanalysis models, or the acquisition of ground-based measurements.
Our aim is to delve into high-quality data obtained from cutting-edge satellite platforms, encompassing their polar low orbits. Specifically, we consider measuring and understanding the impact of relative humidity and wind speeds within the cyclone system [
41], with an important focus on the several phases of its life period. By incorporating these variables, our analysis would provide a more comprehensive perspective on the intricate dynamics. Nevertheless, this type of analysis applied to TLC evolution is also a key point for us, with significant ongoing investigations about the role of convection and moist static energy in the cyclogenesis processes [
14,
15], leading up to the formation of the warm inner core.
The integration of several satellite data sources contributes to a holistic approach, paving the way for more effective strategies in predicting, managing, and responding to extreme meteorological events.