Characteristics of nocturnal breezes in the Windward Islands in the Southeastern Caribbean : structure and nighttime regimes

Transcription

1 Characteristics of nocturnal breezes in the Windward Islands in the Southeastern Caribbean : structure and nighttime regimes C. d Alexis* A. Abouna** H. Berthelot** D. Bernard* *Faculty of Natural Sciences, The University of Antilles and Guyane (UAG), Geosciences and Energy Research Laboratory (LARGE) Guadeloupe, West Indies; christophe.dalexis@univ-ag.fr **Météo-France Guadeloupe Abstract : The windward Islands in the South-Eastern Caribbean have been often chosen as a field for studies on daytime convective boundary layer, sea breezes, weak mountain wakes and gravity waves. Few studies have examined the development of nocturnal stable layers, turbulent processes during these conditions, or the land breeze advent in these regions. Nevertheless, recent research suggests that processes, operating throughout the stable nocturnal boundary layer, may have a significant impact on the boundary layer winds, on the dispersion of pollutants and air quality. In this paper, time series of micrometeorological data, obtained from towers located at the east coast of the island of Guadeloupe, under a southeast trade wind regime, have been used to characterize nocturnal land breeze features. Keywords : Trade wind; Land breeze; Weibull distribution; Synoptic wind. 1. Introduction The present study is a first attempt to estimate some characteristics of a nocturnal breeze flow in a Caribbean island. In sub tropical islands, sea breeze circulation is fairly well covered, in order to understand breeze mechanisms such as formation, persistence and dissipation, in order to simulate the sea breeze structure at fixed locations, to study the effects of topography on climate and the effects of breezes on the environment [1, 2]. In this situation, a thermal internal boundary layer is formed on the land with different diffusive characteristics inside and above the layer. As sunlight warms the land in daytime hours, sensible heat fluxes allow mixing and convective processes. Temperature differences between land/sea produces a pressure gradient at a low level in the atmosphere and causes breeze circulation roughly transverse to the coast. The resulting breeze between atmospheric columns over land and sea triggers circulation with a rising motion over the land. In volcanic islands, such as Guadeloupe, complexity is added by the interaction between breeze and relief. Knowledge about some of these characteristics is vital for various environmental impact assessment studies. Some of the important concerns in this regard are the fate of pollutant release inside the sea/land breeze circulation. At night, radiative cooling inland is greater than radiative cooling of sea water. An overlying cool boundary layer may appear and participate in the dynamics, yielding a thermal land breeze. These conditions create a thin, stable, stratified boundary layer, with a land breeze flow which is much shallower and weaker than the thermal sea breeze in our latitudes [3, 4]. Furthermore, when mountains are present near the coast, katabatic flow may also occur [5]. These situations are different from the ones of sea breezes. The aim of this article is to characterize the nocturnal flow observed around a windward island. Firstly, we will recall the main climatic condition features of the area. Secondly, we will consider series coming from land using meteorological data collected for a year on masts. Then, the classification of daily weather types (RTQ) developed by Météo-France will be used to determine the weather corresponding to the breeze events. The occurrence and onset 1

2 time of the nocturnal breeze will also be presented. Other local meteorological parameters will be considered on a particular day. 2. Some characteristics of climate in the Lesser Antilles In this study, the coastal wind regime of a windward island like Guadeloupe, located between N and N and between W and W, is analyzed (Figure 1). Figure 1. The locations of the three meteorological towers on the island of Guadeloupe: Raizet (RZT) and Desirade (DSD) are managed by Meteo-France Office. Arnouville (ARN) is managed by University of Antilles-Guyane. Two distinct seasons are noticed in this region: a dry season (December-May) and a rainy season (June-November). Wind regimes are mainly associated with the northerly migration of the inter-tropical convergence zone (ITCZ) and the passage of major weather systems such as easterly waves, tropical depressions, storms and hurricanes [6]. In general, several types of winds are encountered in the coastal environments of the windward islands. The main type of wind is the one which is created by large scale synoptic features (such as high and low pressure systems). Wind speed measurements show that the annual cycle of undisturbed synoptic winds reveals a double oscillation (Figure 2). Figure 2. Annual variation in the location of the inter-tropical convergence zone (ITCZ) resulting in the wet and dry seasons. Displacement of the Azores high pressure to the south in January (a) and to the north in July (b). 2

3 The measurements come from the meteorological service of Méteo-France (Figure 3) and they reveal: - a wind flow which increases in strength but remains steady in direction from January to February, - a wind flow which decreases in May due to the deflection of both the Azores anticyclone and the South American depression, - a further increase in intensity from June to September due to both the Azores anticyclone reinforcement and ITCZ displacement towards the north, - and a minimum speed in October generated by the concurrent effects of the Azores high pressure waning and the displacement of ITCZ towards Guadeloupe. J F M A M J J A S O N D a) Temperature RZT ( ) Mean ( C) Mean amplitude ( C) First quintile ( C) Median ( C) Fourth quintile ( C) Interquintile range ( C) b) Wind speed ( ) DSD (m/s) RZT (m/s) Table 1. a) Monthly average surface air temperatures in RZT over 30 years ( ), b) Monthly average wind speed over 4 years ( ) for RZT and ARN. Source : Météo- France. As far as air temperature is concerned, its annual change reaches a minimum in January- February (24.7 C) and reaches up to 27.8 C in July-August (Table 1). This annual cycle of temperature undergoes thermal inertia of the ocean. Indeed, using NOAA data buoys ( W and N at about 800km east of the coast), from 2006 to 2009, there is a similarity between the cycles of annual air temperature and sea surface. The processing of marine data has allowed us to see that sea surface water temperatures close to the island can vary by several degrees from 26 C in February to 29 C in September (Figure 5). 3

4 Figure 3. Wind rose plots (left) and distribution tables (right) for DSD at the top (Number of data studied: 11687, Missing values: 1) and for RZT at the bottom (Number of data studied: 10367, Missing values: 1321). Source: based on meteorological observations provided every three hours by Météo-France. Figure 4. Annual mean sea surface temperatures in the Atlantic and the Caribbean sea. Source: Nasa. 4

5 Above is a graph of sea surface temperature for the Caribbean Sea and the Atlantic Ocean. With no measurement closer to our shores, we have used data derived from an analysis of one year of oceanographic satellite measurements on the NASA website. Annual average around Guadeloupe is about C. We also found it useful to present the spatial distribution of annual mean temperatures on the island and those of February and September. Figure 5. Spatial distribution of temperature on February (a), for an average year (b) and on September (c). Source: Météo-France. 3. Observational datasets The data sets used to study land breezes on the east coast are provided by 10m towers, soundings and Anasyg-Presyg maps. 3.1 Meteorological towers and quality control method Two towers are located on the east coast: the first one, Arnouville (ARN), O and N, is on the east coast of the main island, and the second is located on a small island called Désirade (DSD), in front of the flux O and N. The third meteorological tower is the tower of the national weather station (Raizet, RZT, station identifier: TFFR, station number: 78897). It is located inside the island at, "O and "N (Figure 1). The data collected meet the WMO specifications and are available on the international network (University of Wyoming, College of engineering, Department of Atmospheric Science: All towers measure temperature, pressure, wind, humidity, at various levels ranging from 2 to 10m. The primary measurement level is between 1.5 and 2m for temperature, humidity and pressure. Winds are measured at 10m (synoptic winds). The first two towers, RZT and DSD, provide data with an hourly frequency while measurements at the station ARN give data per second (2 and 10m) and per every 0.05 second (6m). 3.2 Synoptic situation: graphics and soundings To understand the plausible causes of thermal breeze formation on the east coast, we present the analysis of the vertical structure of the atmosphere at 12UTC and 2000UTC. Radiosonde data were obtained from balloons released at RZT station. Another relevant information is a new graphic product of Méteo-France calling Anasyg-Presyg. These graphics include the crucial upper-level dynamic elements superimposed on the surface phenomena: plotting of upper-level jets and tropopause anomalies. Various kinds of fronts and other surface 5

6 discontinuities are specified, using a wider set of symbols and the activity of the weather systems [7]. 4. Nocturnal breezes: April 2007-April 2008 To understand the coastal breeze characteristics, events were identified through April April This period includes the two major seasons known in these latitudes: the wet season going from July to October and the dry season, from February to March. 4.1 Monthly histogram and hourly occurrence The number of monthly land breezes from April 2007 to April 2008 is summarised on Figure 6a. Except for the hurricane season or maintenance periods, data were collected throughout the year. The green squares represent the days of measurement, while the burgundy ones refer to the breezes. The measure of dark red bars was established by categorizing every day of the study period. Blue bars were established by categorizing every day of the study period. For 261 measurement days, 121 days show a breeze onset. Figure 6. a) Weather types in landward breeze period (121 days of breeze), b) Histogram with hours and lengths of nocturnal breeze. As we can see, this type of flow can appear throughout the year with a predominant period from November to March. This period coincides with a southern shift of the ITZC and a strengthening of the Azores high pressure; over this period, various phenomena have arisen: a favourable East-northerly trade wind is generated on all the islands of the Lesser Antilles, a decrease in sea and air temperatures, about 1.6 C/3 C, with a minimum in February, a strengthening of the wind, from 2.5 to 3.6m/s for RZT and from 6.3 to 8.3m/s at DSD, with a wind direction rotation toward the north east. The histogram, in Figure 6b, shows the time frequency of nocturnal land breeze occurrence. We notice that the breeze may exceptionally appear, in the late afternoon, but it generally begins around 08 pm. We also note that the maximum frequency corresponds to the time of the minimum temperature which is between 4 and 5 am. It can extend up to the early morning. 6

7 Analysis of monthly average daily range temperatures provides an additional indication (Table 1). During this period, the average daily values are the highest, about C while C, from April to October. To complete this analysis we used the classification provided by Météo France. Thanks to daily data ( : 6 years, 2192 days), 13 types of weather were selected. This technique was first applied to the data of the year studied (Figure 7a) and then to the breeze periods (Figure 7b). Only the most common are represented. For the observed year, the dominant type of weather is governed by an undisturbed trade wind flow (ANP 53%), swift trade winds (AV 9%) and disturbed trade winds AP (8%). These three cases explain 70% of the observed types of weather. The following two types correspond to specific categories, the first (PAsaRo) characterizes the cyclone period, the second characterizes the lack of trade winds (AP). This type of weather appears throughout the year but it is less common in June-July. The same method has been applied for days with breeze. We observe that 95% of the days with breeze appear for 4 weather types defined at Figure 7c: ANP, AV, Ap, Pa. Another histogram is plotted for these different weather types in Figure 7d. We see that the breeze can extend throughout the night when the weather is like PFA. It starts instead at 10pm with a weather such as Ahino. It is more common in the middle of the night with an Anppn situation. Figure 7. a) The 11 most frequent weather types met in Guadeloupe through 397 days of measurement b) A summary of the number of land breeze identified for each month from April07 to april08 c) Classification by weather types d) Examples of breeze duration according to some types of weather. 7

8 6. Nocturnal breeze of the 04th December 2007: a long event. A landward breeze event from 04 to 05 December This event illustrates a long term breeze, blowing for 10 hours. During the day, the prevailing surface wind flow was light (3.5m/s) and from the southeast (SE), and the skies were mostly clear. No rainfall was observed from 6am overnight. Only patchy high clouds were observed during the day, creating the best conditions for the development of a land breeze. Figure 8. Anasyg/Presyg (04 December 2007). Symbols for features of the sea level pressure field (A being the symbol of weakening high pressure). 6.1 Weather situation and Mean sea level pressure On the night of 04 December, a weak high pressure dominated the island arc. Figure 8 presents the 1200UTC ANSYG-PRESYG (standing for Graphic SYnoptic ANAlysis or PREdiction) allowing a summarizing of meteorological information at the synoptic scale on 04 December These images were subjectively examined to classify the large scale flow associated with land breeze events. The decisive elements of synoptic conditions are: The intertropical convergence zone is below 10 N. At this mid-december period, it moves toward the equator. Around, there is a red surface delimiting a zone of convective activity. A stationary high pressure area was centred to the northeast of Puerto Rico. This situation had been prevalent for two days. The tropopause is marked by a jet, upper-level jet, crossing the Atlantic, with a strong upper-level confluent zone near Jamaica, meaning a main baroclinic zone. This analysis also shows, upstream of the islands, a pseudo-warm front, an area of low baroclinic layers without forcing wind field on the horizontal temperature gradient. These zones mark an area of mass change of surface air (temperature, humidity and wind). 6.2 Soundings The profiles of meteorological and thermodynamic parameters - wind speed and direction, virtual potential temperature (θ v ), specific humidity (q) - are shown in Figure 9. They can be divided vertically into four layers: starting from the bottom, the nocturnal stable layer, the residual trade wind layer, a low inversion layer transition layer and an upper layer. Trade wind inversion appears at about 3500m. This structure is characteristic of a moderately cloudy 8

9 and almost clear sky. The first significant inversion of θ v concerns the nocturnal stable layer, which is the lowest part of the atmospheric boundary layer, 192m deep, with a gradient of about 1.6 C/100m. The second one, the transition or low inversion layer with a small gradient, is due to descent of the warmer and dryer air above. Between the two layers, we find the residual layer, characterized by a quasi-constant θ v and q, due to the turbulent mechanism of the day. Profile of wind speed and wind direction, show a large gradient, particularly for the 04 December 1200UTC. This is the signature of a radiation inversion formed near the ground with a change of wind direction and variation of speed. The low level jet can sometimes be observed at about 200m. Figure 9. Profiles of (a) wind speed, (b) wind direction (WD), (c) virtual potential temperature (θ v ), (d) specific humidity (q), on 4 December 2007, 1200UTC (black), 2000UTC (blue) and 5 December, 1200UTC (red). 6.3 Hourly time series Hourly time series of wind intensity, wind direction and temperature plots at the three towers are shown in Figure 10. Early in the morning, 1200UTC, until about two hours after sunset, 2000UTC, all meteorological towers show the same trend. Winds blow predominantly from the southeast, at an average speed of about 3-4m/s. At 2000 UTC, two wind curve directions change: for the RNA tower, a rotation about 180 degrees is observed, a breeze blows from the land toward the sea; for the RZT tower, the zero value corresponds to the special case called "calm wind ". The latter has been defined by Météo-France. The calm land wind is different from the light wind at sea. It s a non-existent wind or nearly imperceptible one, which reaches its maximum value at around 0.3m/s. No change has been observed for DSD tower, except for a strengthening of the wind intensity. 9

10 Figure 10. Hourly time series plots for the DSD tower (blue), for the RZT tower (red) and for the ARN tower (green), from December 4th at 1200UTC to December 5th 2007 at 1200UTC. The variables are : (a) 10 m wind speed, (b) 10 m wind direction, and (c) 1.8 m temperature ( C). Temperature time series at the RZT and ARN towers show a decrease by approximately 8-9 C, due to nocturnal cooling. This amplitude is similar to those frequently published [1]. For DSD, air temperature near the surface never decreases less than 26 C. It s close to sea surface temperature measured in the Atlantic Ocean. Hourly time series observations are insufficient to show there is a breeze flow. An observation of short time scales is required. It will be detailed in the following paragraph. Just prior to 0000UTC, the abrupt wind direction shift is concurrent with a change in the intensity of wind speed and temperature. We witness a 2 C drop within 10 minutes. 10

11 7. Statistical wind speed analysis In the literature, many authors use the Weibull probability density distribution (pdf) to construct an adequate statistical model for describing the wind speed frequency [8]. We will determine this law of probability, expressed as the probability density function (pdf), to characterize and describe fully and unequivocally the wind speed modulus variation regarded as random variable. All are plotted for a 20 min series during the night. Weibull Night λ confidence confidence p κ (m/s) interval interval value Mode 12/17/07 Eastward 4.71 [4.68;4.75] 8.44 [8.07;8.82] /04/07 Westward 1.07 [1.06;1.08] 4.63 [4.44;4.83] Rayleigh Night σ confidence p (m/s) interval value 12/04/07 Westward 0.32 [0.31;0.33] 0.23 Table 2. Estimation values of parameters, λ for scale and κ for shape, of Weibull pdf estimate. and estimation value of parameter σ for Rayleigh pdf. Comparison between theoretical and observed distributions using Kolmogorov-Smirnov statistical test. Two types of distributions have been identified: a Weibull distribution and a Rayleigh one. These observed functions have been compared with the theoretical Weibull and Rayleigh functions defined below using the Kolmogorov-Smirnov test (Table 2). Figure 11 shows the parameters for the densities plotted in for the westerly and easterly nocturnal wind flow regimes. Weibull distribution is found in both sets of night winds from east (Figure 10 a,b) and west (Figure 10 c,d). The most general expression of the Weibull pdf is given by the two-parameter Weibull distribution expression κ (shape parameter) and λ (scale parameter), or by its pdf equation : x f ( x) x exp ( 1) (1) Values found in the literature, in nocturnal seaward flow conditions, are λ = 5.24m/s and κ = 2.34 from December to May, and λ = 4.49 m/s and κ = 1.83 from June to November [8]. These values are of the same order of magnitude as the ones measured in ARN (Table 2). 11

12 Figure 11. Best-fit Weibull pdf (green line), Rayleigh pdf (black line) and normal kernel density pdf (red line) estimates on the left and cumulative distribution functions, F(x) on the right. The first line (a and b) concerns nocturnal trade wind speed. The two others are about breeze cases. The third line is the best fit for Rayleigh pdf. The distinction between westerly and easterly winds overnight has not been referenced in the consulted bibliography. Whatever the distribution, parameters are much lower in westerly windy conditions. Indeed, in these circumstances, the scale parameter is changed, the distribution is shifted to the left and its height increases. Furthermore, cases of Rayleigh distribution are relatively rare but nevertheless present (Figure 10 e,f). It is important to note that wind speed pdf in an isotropic situation would have a shape parameter = 2 for a Rayleigh distribution [9]. It means that the magnitude of wind speed is issued from a vector with two independent components. 12

13 2 x x f ( x) exp (2) In many meteorological situations, such as in the trade winds, the distributions observed are rather modeled by Weibull. However, for nocturnal landward breezes, in the case of preferred wind direction, the flow is close to isotropic conditions, and is close to 2. Thus wind speed values are Rayleigh distributed. 8. Conclusions and outlooks We have seen that on the east coast of Guadeloupe, a nocturnal wind flows toward the sea. On mountain coasts exposed to trade wind, night cooling associated with undisturbed flow is responsible for the decrease in speed and rotation of the wind. 121 breeze episodes have been detected during the measurement period. They were most common from December to March. Analysis of Ana-Syg map, weather types and experimental data, showed that the breeze is observed during weak synoptic pressure gradients. It has been observed for both undisturbed or/and failure trade wind, it appears most frequently around 10pm. It is a weak western wind characterized by a sudden approximate 2 C decrease in temperature. Analysis of tower data shows that the Weibull-pdf gives a more reliable pattern fitting of the empirical nocturnal wind speed frequency data. These findings are helpful in that they are essential for establishing computerized models to improve the understanding the coastal dispersion of pollutants at night. Acknowledgement: The authors are grateful to C. Montout for his constructive suggestions and J. Lacroix for participating in proofreading the manuscript. References: 1. Case, J. L. and Wheeler, M. M. and Manobianco, J., A 7-Yr Climatological Study of Land Breezes over the Florida Spaceport. Journal of applied meteorology, 44, Information on disaster risk management, case study of five countries : Case study Jamaica. Inter-American Development Bank, Economic Commission for Latin America and the Caribbean Rotunno, R., On the linear theory of the land and sea breeze. Journal of atmospheric sciences, 40, Yan, H. and Anthes, R.A., Effect of latitude on the sea breeze. Monthly weather review, 115, Mapes, B. E. and Warner, T.T. and Xu, M., Diurnal pattern of rainfall in northwestern south America. Part III: Diurnal gravity waves and nocturnal convection offshore. Monthly weather review, 131,

14 6. Brévignon, C., L'environnement atmosphérique de la Guadeloupe, de Saint- Barthélémy et Saint-Martin. Météo-France, Santurette, P. and Joly, A., ANASYG/PRESYG, Météo-France s new graphical summary of the synoptic situation. Meteorol. Appl., 9, Weisser, D., A wind energy analysis of Grenada: an estimation using the 'Weibull' density function. Renewable Energy, 28, Pavia, G.E. and O Brien, J.J., Weibull statistics of wind speed over the ocean. Journal of climate and applied meteorology, 25,