1 Introduction

Coastal low-level jets (henceforth CLLJ or coastal jets) have been identified as important features along all eastern boundary current systems (EBCS; Winant et al. 1988; Ranjha et al. 2013), in the regions between the semi-permanent ocean anticyclones and the mid-latitude western margins of continents: along the California (Winant et al. 1988; Burk and Thompson 1996; Parish 2000) and the Canary (Winant et al. 1988; Soares et al. 2014) currents in the northern hemisphere, and along the Peru-Humboldt (Garreaud and Muñoz 2005; Garreaud et al. 2011), Benguela (Nicholson 2010) and West Australia (Stensrud 1996) currents in the southern hemisphere. The existence of a CLLJ in the Arabian Sea, along the coasts of Yemen and Oman (Ranjha et al. 2013, 2015) is an exception to this pattern, since it does not occur along an EBCS in a western continental coast. The Oman CLLJ is the result of the interaction between the summer South Asian Monsoon and the Findlater jet (also known as Somali jet; Findlater 1969). Lately, several studies revealed the existence of an Iberian CLLJ, connected with the North African CLLJ (Ranjha et al. 2013; Soares et al. 2014).

Coastal jets features are related to low-level winds of high intensity within the shallow well-mixed marine atmospheric boundary layer (MABL). The synoptic pattern, determined by the off-shore anticyclone and by the inland thermal low pressure, forces an along-shore flow that is vertically compressed by the subsidence associated both to the anticyclone and the continental thermal lows. The latter are regionally established by the land–ocean temperature contrast, which forces an onshore sea-breeze that is deflected by the Coriolis force, enhancing an along-shore wind component. Additionally, the coastal topography, often higher than the well mixed inversion capped MABL, blocks the cross-shore component of the flow and channels it along the coast, locally modulating it by coastal and orographic features, like capes, bays, valleys and mountains (Tjernström 1999; Pomeroy and Parish 2001). Typically the CLLJs core (wind speed maxima) is located near the MABL inversion (usually below the first atmospheric kilometre), and can present wind speeds higher than 15 ms−1, with horizontal offshore extensions exceeding hundreds of kilometres, restricted by the Rossby radius of deformation (Overland 1984).

The relevance of CLLJs for the regional climate has been documented in several studies. The Benguela coastal jet in Africa feedbacks positively with local upwelling and is believed to enhance the Namibe desert aridity (Lettau 1978; Nicholson 2010). Moreover, the interaction of the CLLJs and SSTs in the Benguela coast and adjacent Atlantic sectors seems to have a significant influence on the rainfall variability of other African regions, like the Sahel (Lamb and Peppler 1992; Reason and Rouault 2006; Nicholson and Webster 2007), in East Africa (Goddard and Graham 1999; Nicholson and Entekhabi 1987) and in western equatorial Africa (Balas et al. 2007). In the cases of the Californian CLLJ (e.g. Burk and Thompson 1996; Parish 2000) and of the Peruvian–Chilean CLLJ (Garreaud and Muñoz 2005) a similar role of CLLJs takes place outlining some of the most arid regional climates in the world (Lettau 1978). Helfand and Schubert (1995) highlights the CLLJ role in the regional moisture transport over California, showing its blocking effect, by inhibiting the low-level advection of moisture into the continent from the west. But, in the southern region of the Gulf of California, Douglas (1995) showed that the California CLLJ is responsible for significant moisture episodes connected to the North American monsoon. Webster and Hoyos (2004) acknowledged the low-level jet as an important component of the Indian monsoon system and its role in intra-seasonal and inter-annual variabilities.

Coastal jets are strongly linked to regional ocean dynamics (Beardsley et al. 1987) forcing coastal upwelling and nutrient-enriched waters at the coast. The dominant forcing mechanisms of the SST cooling during CLLJs is a mixture of seaward advection of temperature resulting from Ekman transport, air-sea heat exchange, and Ekman-driven coastal divergence (Renault et al. 2009). As a result, the upper thermocline rises towards the surface as the coast is approached. When the upwelling is intense or endures for a sufficient length of time, the thermocline may reach the surface and appears as a front along the coast which travels back and forth depending upon the strength of the upwelling. Observations show that in upwelling regions the Ekman layer thickness is often smaller than in the open ocean, presumably because upwelling brings the thermocline closer to the surface inhibiting wind mixing beyond a depth of 20–30 m (Tomczak and Godfrey 1994). Finally, the upwelling colder SSTs enhance the thermal contrast between land and ocean and feed-back positively on the CLLJs. They are, thus, important for coastal weather and impact directly in many societal activities and economic sectors, like aviation, shipping, fisheries, pollutant and aerosol dispersion, off-shore wind farms, tourism, etc. The societal impacts of coastal jets and its role in the definition of regional climates enforces the need for the understanding of how CLLJ will change in the context of global warming, consequence of the anthropogenic emissions of greenhouse gases. A warmer climate may lead to an intensification of the positive feedback between CLLJs and coastal upwelling (Bakun 1990), having important implications for local ocean currents and to the inter-connected climatic variability and change, making the EBCS regions particularly vulnerable to the impacts of global climate change (Snyder et al. 2003; Miranda et al. 2013; Bakun et al. 2015). Wang et al. (2015) showed that robust and consistent changes in the timing, intensity and spatial homogeneity of coastal upwelling is expected to occur, in response to future warming, in most EBCS. In fact, three out of four of the EBCS (Canary, Benguela and Humboldt, but not California), reveal consistent changes, by the end of the twenty-first century, in upwelling intensity and duration at high latitudes, resulting in a substantial reduction of the existing latitudinal variation in coastal upwelling. Furthermore, these authors’ projections point out to an earlier start and later end of the upwelling season, becoming more intense at high but not low latitudes. Recently, Semedo et al. (2016) analysed a set of global simulations performed using the EC-EARTH global climate model, for present and future climates. They have shown that significant changes in the frequency of occurrence of CLLJs, most notably for Iberia and Oman, are projected to occur by the end of the twenty-first century (but not for the California CLLJ, which is in agreement with Wang et al. 2015). These studies lead to the need of addressing thoroughly the changes in the Iberian Peninsula CLLJ properties and annual cycle in response to global warming.

The Iberian coastal low-level jet has been referred in a small number of studies (e.g. Relvas and Barton 2002; Hoinka and de Castro 2003; Torres et al. 2003; Alvarez et al. 2008), however none clearly show observational evidence or characterize its main properties. In fact, these studies focus on the on the surface wind and on the wind-driven upwelling (Relvas and Barton 2002; Torres et al. 2003; Alvarez et al. 2008); and on the surface wind and its dependence on the Iberian thermal low (Hoinka and de Castro 2003). Those authors suggest that summer surface wind maxima and its seasonality as a surface indication of the low-level jet presence. Very recently, Monteiro et al. (2016) revealed the first observational evidence of the Iberian coastal low-level jet, through aircraft measurements performed off the Lisbon coast. During a low-level jet event, aircraft measurements showed the local wind maximum at higher levels in a good agreement with numerical simulations. Soares et al. (2014) presented the first high resolution study on the Iberian CLLJ, pointing out that in present climate the Western Iberian CLLJ has a mean maximum wind speed around 15 ms−1, located between 300 and 400 m height, and has a maximum frequency of occurrence, in summer, above 30 % downwind of the main coastal features (Capes Finisterre, Roca and São Vicente). The CLLJ occurrence has a clear annual cycle, with frequencies of occurrence of around 10 % in the intermediate seasons, and being almost inexistent in winter. Furthermore, the coastal jet spatial patterns and frequencies of occurrences have large inter-annual variabilities. As aforementioned, Semedo et al. (2016) projected a jet occurrence growth for Iberia in a warming climate, nevertheless the coarse resolution limits the knowledge of the regional and local details of these changes, indicating the need for higher resolution studies. Semedo et al. (2016) associated this increase to the intensification of the Azores Anticyclone. In fact, Davis et al. (1997) examining a twentieth century surface pressure climatology revealed that the meridional flow has increased over the Atlantic, and Li et al. (2012) projected an intensification of Northern Hemisphere subtropical highs in a warming climate, predominantly caused by an increase in thermal contrast between the land and ocean. This enhanced thermal contrast is expected also to lead to an inland thermal low higher frequency.

In this study, the analysis presented by Soares et al. (2014) is applied to a full set of climate simulations, to examine emerging signals of future changes in the properties of the Western Iberian CLLJ associated with global warming. The simulations were performed using the Weather Research and Forecast (WRF) model, at 9 km resolution, for an historical period (1971–2000) and a future period (2071–2100). This high resolution investigation allows an exhaustive characterization of the expected changes of the Iberian CLLJ by the end of the twenty-first century, regarding the mean properties, the annual cycle and the spatial and temporal variabilities.

2 Data and methods

2.1 WRF climate simulations

As previously referred, Soares et al. (2014) carried out a high resolution numerical study devoted to the characterization of the western Iberia CLLJ for the present climate, using ERA-Interim (Dee et al. 2011) to force the Weather and Research Forecasting (WRF) model (Skamarock et al. 2008) as a regional climate model. In the current study, a full set of high resolution climate simulations is used for historical and future climates, also using the WRF and with a similar set up to the hindcast. The historical run is forced by the EC-EARTH model (Hazeleger et al. 2010) for the period 1971–2000, and the future climate simulation is also forced by EC-EARTH for the 2071–2100 time-slice. The set of simulations encompass two domains, one larger at 27 km resolution, and a nested higher resolution domain at 9 km, both centred in the Iberian Peninsula (Fig. 1a). Since a good representation of the planetary boundary layer is crucial, 49 vertical levels were used, with 20 vertical levels in the boundary layer. The lowest model sigma level is at approximately 10 m of height, the second level at 30 m, and the following distances, between levels, increase by 10 % in the first 1000 m and model top is at 50 hPa. Additionally, and as in the hindcast, the chosen physical parameterizations include the microphysics WSM 6 class single-moment (mp6) scheme by Hong and Lim (2006), the planetary boundary layer scheme of Mellor–Yamada–Janjic (Janjic 2001) and the Betts–Miller–Janjic cumulus scheme (Betts 1986; Betts and Miller 1986; Janjic 1990, 1994, 2000). The simulations use the Noah LSM 4-layer soil temperature, soil and canopy moistures model (Chen and Dudhia 2001). The radiation scheme used relies on the shortwave and longwave schemes (Collins et al. 2004) of the NCAR Community Atmospheric model. These options have been evaluated for inland surface variables, namely, maximum and minimum temperatures and precipitation in mainland Portugal (Soares et al. 2012), Iberian precipitation (Cardoso et al. 2013), solar irradiance (Magarreiro et al. 2016), and offshore wind in Soares et al. (2014). Moreover, its results were used to study the moisture recycling processes in Iberian (Rios-Entenza et al. 2014), and the diurnal cycle of the coastal clouds in western Iberia (Martins et al. 2015) both affected directly by the CLLJ dynamics. Other examples, are the studies focused on the development of wildfires propagation models and the characterization of the climatic cooling potential and energy savings for Iberia, using direct ventilation and evaporative cooling systems (Campaniço et al. 2016). All these studies showed the high quality of the results produced using WRF to describe the Iberian climate variables. Recently, the full set of simulations was also used for climate change assessment studies in Iberia, analysing the precipitation changes projected for a future climate (Soares et al. 2016). A more comprehensive description of the model set-up and parameterizations used can be found in Soares et al. (2012) and Cardoso et al. (2013). As in the hindcast, the WRF results are hourly.

Fig. 1
figure 1

a WRF model domains (27 and 9 km) and correspondent domain topographies. b Western Iberian region of interest and orography at 9 km. Boxes (All, BoxN and BoxS) used for regional analysis over land and ocean (sea boxes in full contour and land boxes in dashed contour). In blue a cross-section for detailed examination. P1, P2 and P3 are locations analysed in Soares et al. (2014), P4 location where a maximum future CLLJ occurrence anomaly is found, and P5 location where a maximum jet frequency of occurrence is found in the historical simulation

Full size image

The historical simulation is not synchronized and therefore only its statistics can be compared with observations. The future simulation follows the greenhouse gases emissions of the RCP8.5 scenario (Riahi et al. 2011). The RCP8.5 is one of three possible future scenarios and is named according to the equivalent effect, relative to pre-industrial era, of greenhouse gases concentrations in the radiative forcing values in the year 2100 (i.e. +8.5 W m−2). The other proposed emission scenarios are RCP2.6 and RCP4.5. In the RCP 8.5 the global annual greenhouse emissions rise throughout the twenty-first century, while in RCP2.6 emissions peak between 2010 and 2020 and in the RCP4.5 emissions peak around 2040 and diminish afterwards (Moss et al. 2010; van Vuuren et al. 2011; Riahi et al. 2011). Although RCP8.5 is the more extreme option, the current emission record support it a plausible scenario.

2.2 Methods

The hourly vertical profiles of the zonal and meridional wind speed components and of the temperature allows the detection of CLLJ occurrences in the region of interest (Fig. 1b), following the algorithm proposed by Ranjha et al. (2013), and used in Soares et al. (2014), Semedo et al. (2016) and Cardoso et al. (2016). This algorithm seeks for wind speed maxima associated with a sharp temperature inversion within the MABL. Following, Ranjha et al. (2013), a CLLJ occurs when the following criteria are met: (1) the jet maximum height is below the first kilometre; (2) the wind speed at jet maximum increases by at least 20 % from the wind speed at the surface; (3) the wind speed above the jet maximum decreases to below 80 % of the wind speed at the surface (i.e. a 20 % falloff) in the first five kilometres; (4) the jet maximum occurs within a temperature inversion. These criteria were applied sequentially to hourly wind-speed and temperature vertical profiles to detect the location and frequency of CLLJ. Once identified as a CLLJ occurrence in a grid point, a number of statistics are computed to portray: the local hourly CLLJ frequency of occurrence at different time scales (monthly to seasonal), the mean wind speed, the regional probability distribution functions (PDFs) of jet height of occurrence and jet speed and the jet and surface wind roses.

The referred detection algorithm is applied to the historical and future climate simulations, and their results are compared to assess the expected future changes on the properties of the Iberian CLLJ. But, first, a comparison focused in the CLLJ properties simulated in the historical run is performed against the ones of the hindcast simulation reported in Soares et al. (2014). These authors evaluated extensively the WRF wind and temperature results against observations, showing a very good model performance and its quality to simulate the CLLJs. Finally, in order to give extra weight to the quality of historical WRF results, a comparison of the historical vertical profiles (temperature and wind speed) with available radiosondes (only Lisbon and LaCoruña) was performed (Table S1 in supplemental material).

3 Results

3.1 Evaluation of the historical simulation

The maps of CLLJ frequency of occurrence results of the historical simulation, for winter (December to February—DJF), spring (March to May—MAM), summer (June to August—JJA) and autumn (September to November—SON) along the Iberian Peninsula coastal waters, can be seen in Fig. 2. The frequencies of occurrence are in close agreement with the ones of the hindcast simulation by Soares et al. (2014; their Fig. 6, see supplemental material, Figure S2). In both present climate simulations (hindcast and historical) summer presents the highest frequencies of CLLJ occurrence, with values over the full western Portuguese coast above 28 %, and local maxima around 32 % downwind of the three main capes along the western cost of Iberia (Finisterre, Roca, and São Vicente). The historical run produces a smaller frequency local maximum southwest of Cape São Vicente, in Portugal, which seems to be connected to the stronger and more persistent winds from Gibraltar. In spring the historical run shows considerable lower frequencies of occurrence than the hindcast, less than 5 % in large areas offshore Iberia; while in autumn the opposite happens, i.e. about more 2–3 % downwind of the main coastal capes. In winter both runs show the same kind of maximum values (bellow 8 % of frequency of occurrence), localized in small areas downwind the more conspicuous capes, like Cape Carvoeiro. Positive identifications of coastal jets off the Strait of Gibraltar are visible in summer and in the intermediate seasons (as in Soares et al. 2014), which correspond to a gap wind named Levante (Dorman et al. 1995). These southern coast winds are not studied here and will not be considered in the statistical analysis below; the area where occur it is masked and positive CLLJ occurrences are filtered out.

Fig. 2
figure 2

Maps of hourly CLLJ frequency of occurrence (%) of the historical simulation for each season: a winter—DJF, b spring—MAM, c summer—JJA and d autumn—SON. The same colour scale as Soares et al. (2014) is used but added with grey bars for higher values to enhance differences

Full size image

In Table 1 a summary of the main seasonal spatial CLLJ statistics for the historical run is presented. These include the mean, median and maximum values of: CLLJ frequency of occurrence, jet wind speed and jet height. Additionally, Table 1 shows a comparison of the historical and hindcast CLLJ statistics comprising: the spatial correlation of the seasonal CLLJ frequencies of occurrence and the differences between the mentioned jet properties. Values for the full domain (All) and for the north (BoxN) and south (BoxS) boxes, depicted in Fig. 1b, are shown in Table 1.

Table 1 Statistical comparison between the hindcast and the historical coastal jet properties: spatial correlation of seasonal frequencies of occurrence, relative errors of mean and median of jet wind speed and jet height. Relative differences (Dif.) between the historical and the hindcast, in percentage (historical–hindcast)/hindcast for all variables except for frequencies of occurrence (historical–hindcast) and for spatial correlation coefficients
Full size table

For summer, the spatial correlation of the CLLJ occurrences are high, 0.96 for the full area in analysis, 0.94 and 0.96, for the north and south domains (Fig. 1b), respectively. The correlations for winter and spring are also high, between 0.89 and 0.95. But, in autumn smaller spatial correlations are found, from 0.80 to 0.48. Regarding the mean and median values of the CLLJ frequency of occurrence, the differences between the control and the hindcast are rather small, ~1 % out of ~20 % for the full domain, for JJA. The larger differences are, again, found in spring, and are much smaller in autumn and winter.

In Fig. 3 the summer main statistical properties of the Iberian CLLJ result of the historical simulation are plotted. These include the jet height PDF, the jet wind speed PDF, the jet wind rose, the surface jet wind rose, a mean wind and potential temperature cross-section and a vertical wind and potential temperature cross section both when the jet occurs.

Fig. 3
figure 3

Summer CLLJ statistics for Historical (1971–2000): a jet height histogram, b jet wind speed histogram, c jet wind rose and d jet surface wind rose and East–west cross-section at 38.41°, marked in blue in Fig. 1, where the blue lines are potential temperature and colours are e wind speed and f vertical velocity, both averages when the jet occurs

Full size image

The comparison of Fig. 3 with Figure S2 (Fig. 7 from Soares et al. (2014)) reveals that the statistical features of the historical simulation are rather similar to the hindcast ones. The histograms of jet heights and of jet wind speed are almost identical (Fig. 3a, b), where the height median is around 290 m for both simulations, and the median jet speed in around 15 ms−1, with only 3 % of relative difference between the two simulations. Likewise, the most persistent jet wind direction is north–north-easterly for more than 35 % of occurrences and north with values less than 25 % (Fig. 3c). The historical surface, wind when the jet occurs (Fig. 3d), shows a striking persistence above 40 % from north and around 25 % from NNE, and less than 15 % NNW. The potential temperature (contour lines) and wind speed (colours) cross section, when the jet occurs, of the historical simulation (Fig. 3e) is analogous to the one presented by Soares et al. (2014). The thermal structure of the lower atmosphere shows a well-mixed MABL capped by warmer air. This well mixed layer is under the jet structure which slopes towards the coast. The mean jet core is between 200 and 400 m heights and has wind speeds above 12 ms−1. The potential temperature (contour lines) and vertical wind (colours) cross section, when the jet occurs, (Fig. 3f) shows the typical subsidence air over the ocean associated with the Azorean anticyclone. This prevalent subsidence air over the ocean is stronger above the jet eastern limit, which compresses the isentropes and enhances the horizontal wind speeds and by definition of the jet core is located where higher wind speeds are found. It is relevant to mention that the CLLJ vertical structure and wind intensity are highly dependent on the vertical resolution and on the planetary boundary layer scheme used (Bravo et al. 2008; Hu et al. 2010). As Bravo et al. (2008) pointed out a not very high vertical resolution may generate a weaker and too vertically wide low-level jet. Moreover, the setting of a relatively high value for the minimum TKE, as considered in the Mellor–Yamada–Janjic planetary boundary layer scheme, may exacerbate mixing and contribute as well to a too high and too weak LLJ.

Although in the historical simulation WRF is forced by a free running GCM (EC-EARTH) and not by a reanalysis (ERA-Interim), where offshore and onshore observations are assimilated, and two different time windows were covered (1971–2000 for the historical run and 1989–2007 for the hindcast) the historical results follow closely the ones for the hindcast (Soares et al. 2014). This resemblance is essential to increase the confidence in WRF results, in its ability to describe the present and future climate properties of the Western Iberian CLLJ, and on its suitability to disclose the climate change signal on the coastal jet features.

3.2 Future projections for the Iberian CLLJ

According to the WRF simulation following the RCP8.5 future scenario, the overall values and the spatial pattern of future frequencies of occurrence of the Iberian CLLJ show considerable changes for the end of the twenty-first century. The maps of the projected seasonal CLLJ frequency of occurrence for the end of this century, as well as the seasonal coastal jets frequency of occurrence anomalies (future minus present) are shown in Fig. 4. A Student’s t statistical significance test was applied to the spatial distributions, and the results show that the differences between future and historical simulations are statistically significant. Additionally, in Table 2 a summary of the main coastal jet future properties and correspondent changes, comparing the historical and the future simulation, are listed.

Fig. 4
figure 4

Maps of seasonal hourly CLLJ frequency of occurrence (%) for future climate, for a DJF, b MAM, c JJA and d SON; the same colour scale as Soares et al. (2014) but added with grey bars for higher values. Maps of the differences (Future minus Historical) of seasonal hourly CLLJ frequency of occurrence (%) (2071–2100 minus 1971–2000), for e DJF, f MAM, g JJA and h SON. Shaded areas specify not statistically significant changes using a Student’s t test at the 90 % confidence level

Full size image
Table 2 Summary of the jet properties changes projected for the end of the twenty-first century. Future values and relative differences (Dif.) to the historical results, in percentage (future–historical)/historical for all variables except for frequencies of occurrence (future–historical)
Full size table

Future CLLJ climate (Fig. 4a–d) reveals significant occurrences along almost the full extent of the western Iberian coast, with frequencies above: 36 % in the summer (values in grey), 10 % in the spring and 12 % in autumn. These 3 seasons show maximum values of 46, 14 and 16 %, respectively, which are mainly localized in the northern half of the western coast.

For summer, the amplification on the persistence of the CLLJ is most notable. In the northern sector, the frequency of occurrence anomaly (future minus present) has a maximum around 18 %, which corresponds to almost doubling its presence in this region. An increase in the frequency of occurrence of above 10 % extends over the full western Iberian coast (Fig. 4; Table 2), pointing to not only a more persistent CLLJ but also to an enlargement of its region of influence, especially in higher latitudes, in agreement with the results of Wang et al. (2015) and Semedo et al. (2016). The northern area where larger increases are expected, ~15 % on average in the BoxN (Table 2) are in agreement with the regions that in present climate revealed higher temporal variability (Soares et al. 2014). The consistency of the augment of persistence of summer coastal jets can also be observed in Fig. 5a, where boxplots of the full seasonal distributions of frequency of jet occurrence, for the selected areas in Fig. 1b, for present and future climates, are presented. In fact, there is a clear distribution shift for much higher values in the full area and in both sub-domains, which is stronger in the BoxN. For summer, in all areas, the present climate median frequencies are surpassed by future 25th percentile frequencies, and for the BoxN the present climate 75th percentile has similar values to the future 25th percentile, stressing the major changes projected for the future Iberian jet.

Fig. 5
figure 5

Boxplots of the CLLJ results for the historical (black) and the future projections (red), for the different regions of Fig. 1b, for all seasons, for a jet frequency of occurrence, b jet wind speed, and c jet height. Individual boxes span from the 25th to the 75th percentile, with the median represented by a straight line, with the absolute maximum and minimum indicated by crosses

Full size image

In Fig. 5b, c, the seasonal boxplots of jet wind speed and jet height, for present and future climates, are also plotted. For summer, the jet wind speed boxplots show a shift for higher jet speeds in the full area, in spite of some reduction in the southern sub-area (BoxS), which is compensated by a stronger increase in the northern sub-area (BoxN). In fact, the jet speed mean value increases around 2, 6 % and diminishes 2 %, respectively, for the full area, the north and south boxes (Table 2). Overall, there is a reduction of values ranging between 5 and 17 ms−1, and increasing occurrences of enhanced speeds from 17 to 27 ms−1.

Similarly, the jet height boxplots (Fig. 5c) for summer exhibits a shift towards higher levels in a future climate in the full area, and strongly in the northern sub-area. This shift is accompanied for an enlargement of the distributions, which doesn’t alter substantially neither the mean nor the median values of the jet height. The southern sub-area only shows a small widening of the distribution. The changes of jet wind-speed and jet height were tested for their statistical significance, through a Kolmogorov–Smirnov test, and all the changes are statistically significant at a 95 % confidence level.

Looking at some other future properties of the summer coastal jets, Fig. 6 displays the future CLLJ wind rose, the surface wind rose when the jet occurs and the mean cross sections of potential temperature and jet speed and vertical velocity. The projected jet direction change points out to an increased northerly frequency (Fig. 6a) resulting from a reduced NNE jet wind direction when compared to present climate (Fig. 3c). Accordingly, at the surface, the wind, when the jet occurs, shifts from NNE to N, and the highest frequencies remain from the northern sector (Fig. 6b). The jet height-speed histograms (not shown) for present and future are similar, but accordingly, with the two previous PDFs, revealing a common frequency change for the occurrence of the jet at higher vertical levels and with superior speeds. Finally, Fig. 6c–d presents cross sections for the potential temperature and wind speed, and potential temperature and vertical velocity, averages when the jet occurs (in blue in Fig. 1b). It can be seen a general warming that corresponds roughly to 2 K in the MABL and 4 K aloft. For both present and future climate, the jet within the MABL is easy to identify, but with greater speeds and larger spatial coverage in the future scenario. The CLLJ slopes upward offshore, accompanying the MABL inversion slope.

Fig. 6
figure 6

Summer CLLJ statistics for future simulation (2071–2100): a jet wind rose, b Jet surface wind rose and East–west cross-section at 38.41°, marked in blue in Fig. 1b, where the blue lines are potential temperature and colours are c wind speed and d vertical velocity, both averages when the jet occurs

Full size image

The MBL sloping downward toward the coast is associated with the cross-shore pressure gradient, which maintains a near geostrophic equilibrium with the Coriolis deflection of the jet (Garreaud and Muñoz 2005). This upward slope offshore is due to the increasing sea-surface temperatures offshore, the weakening to the west of the synoptic-scale subsidence (Fig. 6d) and the enhanced subsidence closer to the coast due to the daytime regional subsidence associated to the return flow of warm air from land to sea (Beardsley et al. 1987). The CLLJ maintains a semi-geostrophic balance, in which the background pressure gradient plus the induced mesoscale pressure perturbation are balanced by turbulent and acceleration terms (Cui et al. 1998).

The future anomalies of the large-scale subsidence, the synoptic pattern flow and the thermal contrast between land–ocean, may contribute to the understanding of some of the mechanisms responsible for the CLLJ changes. It is important to refer that CLLJs are not in geostrophic balance but are deeply rooted in the synoptic large and regional scale flows. Moreover, the momentum turbulence role should not be forgotten in a more detailed view, which is out of the reach of the current study, based on a long climate simulation.

The inspection of present and future vertical velocity cross-sections (Figs. 3d, 6d) does not reveal any significant change on the large-scale subsidence. Figure 7 presents the future mean sea level pressure (mslp) and the correspondent geostrophic wind results for EC-EARTH, and their respective anomalies (future compared to present). The mslp shows a positive anomaly in the north sector, centred in the British Islands, revealing a north-easterly broadening and larger persistency of the Azorean anticyclone, and consequently a larger influence over north Iberia. Simultaneously, a negative anomaly coexists in inland Iberian, which corresponds to an intensification of the thermal low over land. This synoptic pattern has an intense correspondent signal on the geostrophic wind, which amplifies over 4 ms−1 offshore the northwest of Iberia, where the higher increases of the coastal jet occur.

Fig. 7
figure 7

Summer mean sea level pressure and correspondent geostrophic wind results of the EC-EARTH model future run (2071–2100) and anomalies against the historical simulation (1971–2000). Shaded areas specify changes not statistically significant using a Student’s t test at the 90 % confidence level

Full size image

The annual cycle of surface temperature land–ocean differences, for the historical (1971–2000) and the future (2071–2100) simulations, for the marked boxes of Fig. 1b, is plotted in Fig. 8. In the boxes, concerning ocean or land temperatures, only ocean or land grid-points are taken in account, respectively. The surface temperature differences between land and ocean expose increasing values in future climate throughout the full annual cycle, but higher in summer, and larger in the north regions. Subsequently, it is clear that the two main CLLJ forcing ingredients, the Anticyclone-Thermal Low dipole and the land–ocean thermal contrast, appear enhanced in future climate, and especially in the north half of western Iberia. All this, adds to an increase of the CLLJ persistency. Obviously, the important feedbacks between the CLLJ and upwelling are not represented in these simulations, since sea surface temperature is updated according to the forcing data. An ocean–atmosphere coupled simulation would be very important to analyse these feedbacks in a future climate, but is out of the scope of the current study.

Fig. 8
figure 8

Annual cycle of land–ocean differences of surface temperature (2 m) for the marked boxes in Fig. 1b, for the Historical (1971–2000) and the Future (2071–2100) simulations, future minus historical

Full size image

In the intermediate seasons, the jet frequencies of occurrence anomalies, between future and historical simulations (Fig. 4f, h; future minus present), reveal a small intensification of occurrences, on average around 4 % (Table 2), extending offshore over 300 km, in spring, and more than 200 km, in autumn. The seasonal boxplots of the jet frequency (Fig. 5a) show, for both intermediate seasons, a distribution shift for a larger persistence of coastal jets, but with less expression than in summer. Moreover, a stronger anomaly in the extreme values frequencies can be identified, for both the southern and the northern boxes, with maximum values increasing over 8 %, and in particular, in autumn in the southern region an 18 % growth is found (Fig. 5a; Table 2). The plots regarding the projected changes of the jet properties for the intermediate seasons (Figs. 5, 9; Table 2) highlight the common trend to the summer, but with smaller magnitude. The shift for higher jet wind speeds is clear in Fig. 5b, where future spring and autumn are expected to have an increase in the frequencies of greater jet wind intensities, when compared to the historical period. This augment is larger in autumn, in the northern area (BoxN), where the jet mean speed is expected to rise over 7 %. Concerning the jet heights (Fig. 5c), in future spring and autumn a modification for higher altitudes for jet occurrence is also expected. In fact, for both seasons, the boxplots reveal wider distributions and increases in all medians and mean values. In what regards to the jet wind direction, the projected changes are less noticeable than in JJA. In spring, the historical and future jet wind roses are almost identical, where, in both cases, the most persistence jet direction is from NNE, with frequencies of occurrence of about 25 %, and the other directions are almost completely comprised in the full NE-N quadrant (Fig. 9a, b). For autumn, in agreement with the Fig. 4, where a more expressive increase of jets is projected for autumn than in spring, also here a partial shift to northerly wind appears, like in summer.

Fig. 9
figure 9

CLLJ statistics for historical (1971–2000) and future (2071–2100) simulations: a, b jet wind rose for the historical and future simulations both for MAM, and c, d Jet wind rose for the historical and future simulations both for SON

Full size image

The CLLJ season seems also to temporally enlarge its influence in future climate. For the contiguous months to JJA, May and September, the frequency of jet occurrences grow significantly, around 10 % along all the coast extension, almost doubling its influence in these months (Figs. 10, 11). The summer months that experience a greater increase of frequencies of occurrence are June and July, reaching absolute values of around 50 % (Fig. 10), but enlarging its spatial influence, covering the full western offshore region of Iberia. Still it can be seen that for the full annual cycle the CLLJ occurrences grow, roughly 5 % in March, April, October and November, with the exception of the winter months (DJF). In general, the variability of the CLLJ frequency also increases spatially and temporally, with increased values of standard deviation further away from the shore, and with the exception of August where a decrease is found (not shown). This weakening of the standard deviation shows August as the month when the CLLJ possess steadier prevalence.

Fig. 10
figure 10

Maps of monthly means of the hourly frequency of CLLJ occurrence (%) for future simulation 2071–2100. The same colour scale as Soares et al. (2014) but added with grey bars for higher values

Full size image
Fig. 11
figure 11

Maps of differences of the monthly means of the hourly frequencies of CLLJ occurrence (%). Future minus Historical simulations (2071–2100 minus 1971–2000). Shaded areas specify changes not statistically significant using a Student’s t test at the 90 % confidence level

Full size image

To further illustrate the local changes in the CLLJs, the monthly frequencies of occurrence and mean jet wind speeds are plotted in Fig. 12, for 5 selected locations (Fig. 1b). The first three positions (P1, P2 and P3) are the points where Soares et al. (2014) identified the highest CLLJ frequencies of occurrence in the hindcast simulation. P4 is located where the maximum future anomaly is found, and P5 is the site where a maximum jet frequency of occurrence is found in the historical simulation. For all the sites a temporal enlargement of the CLLJ season is detected, for earlier and later than JJA. It is visible that local frequencies of occurrence over 20 % occur from May to October. In 4 out 5 locations the maximum frequency is anticipated one month, from August to July, in future climate. All the locations present frequencies of occurrence higher than 10 % from April to October in future projections, anticipating 1 month the CLLJ season and delaying also 1 month it’s ending. The southern point (P3) is the one showing the lower increase in the CLLJ frequency of occurrence (~4 % in July) The northern points have much larger projected increases, which in the extreme case of P4 reaches approximately 18 %. Regarding, the jet wind speed annual cycle most of the locations present an increase of the jet mean wind speed, less than 1 ms−1, that corresponds to an anticipation and delay of the CLLJ season.

Fig. 12
figure 12

Local mean monthly of a–c CLLJ frequency of occurrence (%) and b–d jet wind speed (m/s). P1, P2 and P3 are locations analysed in Soares et al. (2014), P4 location where a maximum future anomaly is found, and P5 location where a maximum jet frequency of occurrence is found in the historical simulation

Full size image

4 Conclusions

The EBCS regions are amongst the most productive ocean ecosystems, where the atmosphere–land–ocean feedbacks play a key role in defining the regional climate (Bakun et al. 2015; Wang et al. 2015). In the EBCS regions, one can find some of the most arid regions of the world, namely, in the Mojave, Namibe, and Atacama deserts, and a crucial component is the presence of coastal jets (Nicholson 2010). Ranjha et al. (2013) using the ERA-interim reanalysis characterized the main CLLJs properties globally, and were the first to quantify the persistency of the Western Iberian coastal jet, reporting frequencies of occurrence above 20 %. Semedo et al. (2016), by means of a set of global climate simulations, identified robust and clear changes of the CLLJs properties in response to future warming, namely for the Iberia and the Oman CLLJs. This study projected future rise of CLLJ occurrences, in particular during summer, but also in the intermediate seasons, enlarging the duration of the CLLJ season, and an increase of the jet wind speeds. Recently, the western Iberian present climate CLLJ properties, were extensively described by Soares et al. (2014) using a high resolution regional climate simulation with WRF forced by ERA-Interim, where a maximum frequency of occurrence in summer above 30 % was found, with mean jet wind speeds around 15 ms−1 and a jet core located between 300 and 400 m heights.

In this study, the first climate change assessment study using a high resolution (9 km) regional climate simulation, focused in the Western Iberian CLLJ, is presented. Two simulations with the WRF model were analysed, one historical for the period 1971–2000, and another for future climate (2017–2100), in agreement with the RCP8.5 scenario, both forced by the EC-EARTH model. The CLLJ properties were computed for these two runs and compared to investigate their changes in the context of global warming. However, firstly a comparison of the CLLJ properties result of the historical simulation with the ones of the hindcast run presented by Soares et al. (2014), was performed. This comparison clearly demonstrates the similarity between these simulations regarding the western Iberian costal jet. The comparative analysis between future and historical simulations suggest significant modifications for the Western Iberian CLLJ. The hourly frequency of occurrence of the CLLJ increases in the full Western Iberian coast, from values of around 35 % to almost 50 % in summer. The relative increases of the frequency of occurrence are stronger in the north sector of Western Iberia, where the frequency of occurrence nearly doubles. The occurrence of coastal jets is expected to cover larger areas, both latitudinal and longitudinal. The CLLJ season is projected to be enlarged, extending to more months of the annual cycle, only the winter months continue to have negligible occurrences. There is a shift for higher jet wind speeds and for the jet occurrence at higher heights, in all three seasons, but in particular in summer. Also, the inter-annual variability is expected to increase significantly, with the exception of August.

The projected changes for the Western Iberian CLLJ are substantial and are in accordance with Wang et al. (2015) and Semedo et al. (2016), emphasizing the increased impacts at EBCS higher latitudes. Here, an explanation is proposed to understand the future evolution the main CLLJ ingredients, based on the changes in the main circulation ingredients, from the synoptic to the regional scales. In fact, in future climate, the Azorean Anticyclone is more persistent and its influence extends further northeast encompassing in a greater measure the areas to the north of Iberia, forcing more frequent and more intense along-shore winds, especially in the north sector of Iberia. At the same time, the ocean-land thermal contrast appears enhanced in the end of the twenty-first century, and again in a larger extent in the north part. These two CLLJ conditioning effects add to the general CLLJ augments mentioned. The future changes in the Iberian CLLJ may have a direct impact on the upper ocean properties and in the upwelling regime, and thus on the biological production. The enhancement of low-level jets would lead to an increased upwelling rate, greater turbulence in the upper-ocean, and faster offshore transport of surface waters (Bakun et al. 2015). These effects may be, however, counteracted by upper-ocean warming which could increase water column stratification. The interaction between the upwelled waters, thermocline depth, turbulent mixing and the strength of the wind jets is still an unresolved question (Bakun et al. 2015)

To our knowledge the current study is the first one using both high resolution and temporal sampling to address the signal of climate change on a relevant CLLJ system. The extension of this work to other CLLJs and the identification and attribution of the role of the changes on the regional climate it is a challenging and promising investigation for a near future. Obviously, the important feedbacks between the CLLJ and upwelling should be taken in account and a high resolution ocean–atmosphere coupled simulation would be desirable to analyse these feedbacks in present and future climates.