The purpose of this study is to investigate the wintertime distribution of the fish eggs along the coast of southwestern Taiwan, and to elucidate the fish spawning ground in this area. Moreover, the distribution of striped mullet eggs was also investigated. Two kinds of sampling methods were used in this study, including the surface and vertical trawling. The former applied vertical and oblique towing along the coastline. One sampling was proceeded in November, December 2000, and January 2002 by R/V Ocean Researcher 3 at three transects vertically along the coastline at 10 to 1140 meters deep in Budai, Jiading, and Fangliau. The other was proceeded in November 2000, January, December 2001, and January 2002 parallelly along the coastline at 8 sites from Budai to Fangliau. Besides, the vertical sampling employed in December 2001 and January 2002 by R/V Ocean Researcher 1 in the Taiwan Strait was also analyzed. A total of 5909 fish eggs were obtained from three sampling nets, NORPAC, cyclindricalconic net, and bongo net. All together 4 families, 2 genera, and 2 species of fish eggs, including Myctphiformes Synodontidae, Engraulidae, Tetraodontiformes Tetraodontidae, Scorpaeniformes Triglidae, Muglidae, Clupeiformes chirocentru hyposelosoma, Salmoniformes Polyipnus polli, and Allopus japonicus were found. The fish eggs that couldn¡¦t be identified were sorted into 41 groups by the egg shape, size, and membrane pattern. The four most abundant fish eggs were Muglidae, Engraulidae, dead eggs, and group CV. The fish eggs abundance decreased from inshore to offshore at two transects in Jiading and Fangliau. But it was contrary at the transect of Budai. The same pattern also appeared in the distribution of Muglidae eggs. In the area of inshore water where waterdepth from 15 to 25 meters, higher abundance of total eggs, Muglidae, and Engraulidae were found at the southern site. Muglidae eggs were the most abundant in December, but Engraulidae eggs didn¡¦t appear in this month. 33 fish eggs that might be striped mullet were sampled by OR1 at the edge of 21 ¢J isotherm where China Coastal Current met Kuroshio Current. The hydrographic variation in the year from 2000 to 2001 was affected by tides and topography, whereas in the year from 2001 to 2002 was influenced by Kuroshio. Higher fish eggs abundance found at Linyuan may due to the turbulence of Kao-Ping submarine canyon.

Respirable dust, carbon dioxide, ammonia, hydrogen sulfide, and carbon monoxide concentrations were measured using fixed-area monitoring and contaminant mapping in a 19–crate farrowing room during the winter. Direct–reading instruments were used with fixed–area stations and contaminant mapping to evaluate concentrations during five days over a period of a three–week farrowing cycle. Concentrations were evaluated to determine the effect of the pit ventilation on contaminant concentrations, a change in concentration occurred over a sample day, and to determine if three data collection methods produce different daily respirable dust concentrations. Pit ventilation did have a significant effect on contaminant concentration in a farrowing barn during winter. Compared to when the pit fan was on, mean area contaminant concentration, with the exception of CO, was significantly higher when the pit fan was turned off (p<0.001). Mean respirable dust concentration was 79% higher, CO2 concentration was 35% higher, NH3 increased from 0.03 ppm to 10.8 ppm, and H2S concentrations increased from 0.03 ppm to 0.67 ppm. A significant change in area respirable dust (p<0.001) and CO2 (p<0.001) mean concentrations occurred over time throughout the course of a sample day. Mean area respirable dust concentrations were highest in the beginning of the sample day and decreased by 77 % (pit fan off) to 87% (pit fan on) over a five–hour sample period. Higher concentrations were likely attributed to the feeding period that occurred early in the day. When the pit fan was turned off, mean area CO2 concentrations increased by 24% by the end of the sample day due to the inefficient ventilation and the constant production of CO2 generated by the swine. Finally, comparing the three data collection methods produced similar results concerning the ranking of the daily mean concentrations of respirable dust; however, differences were seen in the magnitude of the daily average respirable dust concentrations across the three data collection methods, which might lead to different interpretations of risk. To ensure risk is not underestimated, multiple fixed–area monitors are recommended to characterize room concentrations. Throughout the study, contaminant concentration did not exceed regulatory or international consensus standards; however, recommended agricultural health limits suggested in the literature were exceeded for respirable dust, CO2, and NH3. These findings indicate the need to consider personal exposures to those working in farrowing barns and control options to reduce these contaminant concentrations in production facilities.

Les vallées alpines urbanisées sont régulièrement soumises à des épisodes de pollution aux particules fines, en particulier sous des conditions hivernales anticycloniques. Ces épisodes se développent du fait de la conjonction de l'augmentation des émissions et de la stratification de l'atmosphère qui inhibe le mélange vertical et isole l'atmosphère de vallée de la dynamique de grande échelle. Le transport des polluants devient alors principalement piloté par les écoulements locaux d'origine thermique. Ces écoulements se caractérisent par une forte dépendance aux spécificités locales de la zone et sont difficiles à représenter dans les modèles numériques de prévision du temps, tout comme les conditions stables qui les accompagnent. L'amélioration de la prévision des situations de pollution hivernale en zone de montagne nécessite donc une meilleure compréhension de la dynamique locale en condition stable. Cette thèse s'inscrit dans ce contexte et vise à améliorer la compréhension de la structure des circulations locales à l'échelle de la vallée. Pour cela, l'étude s'appuie sur les données acquises lors de la campagne Passy-2015 et sur des simulations numériques haute résolution réalisées avec le modèle Méso-NH. La campagne s'est déroulée durant l'hiver 2014-2015 dans le bassin de Passy, situé à proximité du Mont-Blanc et à la confluence de trois vallées. Les concentrations en PM10 observées dans ce bassin excédent régulièrement les seuils réglementaires et montrent des hétérogénéités marquées au sein du bassin et avec les vallées adjacentes. L'étude de la dynamique met en évidence le rôle des circulations locales vis-à-vis des disparités dans la distribution spatiale des polluants. En particulier, les écoulements dans le bassin sont organisés selon différentes strates et génèrent des niveaux de ventilation hétérogènes. En journée, les échanges de masse s'opèrent de manière préférentielle entre les segments de vallée les plus ensoleillés. [...] / Air quality issues are frequent in urbanized valleys, particularly in wintertime under anticyclonic conditions. Pollution episodes occur due to the combination of increased emissions and atmospheric stratification that inhibits vertical mixing and isolates the valley atmosphere from large-scale dynamics. The transport of pollutants then becomes mainly driven by local thermally driven flows that largely depend on local characteristics and are difficult to represent in numerical weather prediction models. Improving the forecasting of winter pollution situations in mountain areas therefore requires a better understanding of local dynamics under stable conditions. This thesis fall within this objective and aims at improving the understanding of local wind dynamics at valley scale. It is based on high-resolution numerical simulations performed with Méso-NH and data from the Passy-2015 field experiment that took place during the winter of 2014-2015 within the Passy basin, located near Mont-Blanc and at the confluence of three valleys. The PM10 concentrations observed in this basin regularly exceed the regulatory thresholds and show marked heterogeneities within the basin and with adjacent valleys. The wind dynamics study highlights local flow characteristics that are consistent with the PM10 heterogeneities observed within the valley. In particular, flows within the basin show a stratified structure and give rise to heterogeneous ventilation levels. During the day, mass exchanges preferentially occur between the sunniest valley sections. At night, the convergence of flows from tributary valleys, along with the local orography, induces a very heterogeneous flow structure on the vertical and horizontal in the Passy basin. These characteristics tend to reduce ventilation in the basin especially in the eastern sector, which is also the most polluted sector during wintertime episodes. As spring approaches, the increase of solar radiation balances inter-valley mass exchanges, thus reducing pollutant accumulation within the basin. The analysis of mechanisms controlling local circulations underlines the importance of fine scale characteristics of topography and surface (snow cover) that determine the distribution of energy received at the surface.

En hiver, par condition anticyclonique hivernale, les régions montagneuses urbanisées font l'objet de conditions atmosphériques stables et découplées, qui conduisent à des épisodes de forte pollution particulaire. Ce travail de thèse traite de la caractérisation des facteurs de pollution de l'air aux particules par condition stable en vallée alpine encaissée, du point de vue de la dynamique atmosphérique. Ce travail repose sur des simulations numériques idéalisées et en conditions réelles à l'aide du modèle Weather Research and Forecasting (WRF), les particules étant modélisées par un champ de traceur passif.Dans une première partie, des simulations numériques à haute résolution ont été réalisées, en utilisant deux configurations différentes de vallées tridimensionnelles idéalisées ouvrant sur une plaine. La première configuration correspond à une vallée de largeur constante dans la direction de l'axe de la vallée ouvrant sur une plaine. Dans la seconde configuration, la vallée est composée de deux sections, la section aval, qui ouvre sur la plaine, étant plus étroite que la section amont. Cette configuration est appelée "pooling". Quelle que soit la configuration, la variation du profil vertical de température le long de l'axe de la vallée conduit à d'un gradient de pression horizontal générant un vent de vallée. Les configurations de type "pooling" sont associées à des concentrations de polluants plus élevées que dans le cas où la largeur de la vallée ne varie pas, car le vent de vallée est plus faible dans la configuration "pooling". L'impact de la largeur de la section aval sur la concentration de polluants dans la section amont est remarquable: la ventilation de la section amont peut être complètement bloquée pendant la majeure partie de la simulation (six heures) lorsque le rapport de largeur de la section amont à la section aval est égal à 10.La deuxième partie concerne une situation réelle en vallée alpine, durant un épisode fortement pollué de février 2015. La dynamique atmosphérique et la concentration de particules fines (PM) ont été modélisées dans une section de la vallée de l'Arve autour de la ville de Passy par le modèle WRF-Chem durant cet épisode. Pour cela, le cadastre d'émission préparé par l'agence de qualité de l'air de la région Auvergne Rhône-Alpes a été implémenté dans le modèle. La dynamique de la couche limite a été comparée à des profils verticaux de vent et température mesurés lors d'une campagne de terrain durant l'épisode pollué, et la concentration de PM comparée aux données de stations de qualité de l'air. La contribution des vallées tributaires à la pollution atmosphérique dans cette section de la vallée de l'Arve a d'abord été étudiée. Les résultats montrent que cette contribution est très faible par rapport à celle des sources locales; ceci implique que les émissions $in-situ$ sont principalement responsables de la forte concentration de particules polluantes enregistrée dans la vallée car le fond de vallée est découplé de l'atmosphère au-dessus de la vallée. Les processus de ventilation et leur influence sur la concentration de particules (PM) ont ensuite été étudiés. L'analyse montre qu'en raison de la ventilation limitée autour de la ville de Passy, la variabilité horaire de la concentration de PM dans la ville est contrôlée par les émissions. / During wintertime anticyclonic regimes, urbanized mountain areas often experience stable and decoupled atmospheric conditions, resulting in severe episodes of particulate air pollution. This study deals with the characterization of drivers of particulate air pollution in deep alpine valleys under such stable conditions from the point of view of atmospheric dynamics. The work has been carried out through the implementation of both idealized and real case numerical simulations using the Weather Research and Forecasting (WRF) model. Particulate air pollution has been modelled by implementing passive tracers in the simulations.In a first part, high-resolution numerical simulations have been performed using two different configurations of three-dimensional idealized valleys opening onto a plain. The first configuration corresponds to a valley of constant width in the along-valley direction, directly opening onto a plain. The second one consists in a valley of varying width in that direction, with an upstream section of larger width than the downstream section, which opens onto the plain. The latter configuration is referred to as a pooling case. The change in the vertical temperature profile along the valley axis in all configurations results in a horizontal pressure gradient that leads to the development of an along-valley flow. Configurations in which the upstream section of the valley opens onto a narrower one have shown to be prone to higher pollutant concentrations than in the case of a valley opening directly onto a plain, due to the weaker down-valley flow developing in the variable width configurations. The impact of the downstream valley section on the concentration of pollutants in the upstream valley section is remarkable: the ventilation of the upstream valley section can be completely blocked for most of the simulation (six hours) when the ratio of the upstream to downstream valley section width is equal to 10.The second part deals with a real case situation, focusing on a section of the alpine Arve River valley around the city of Passy. The numerical modelling of the atmospheric dynamics and particulate matter (PM) concentration in that section has been performed during a strongly polluted wintertime episode of February 2015. For this purpose the emission inventory prepared by the air quality agency of the région Auvergne Rhône-Alpes has been implemented in the WRF-Chem model. The boundary layer dynamics has been compared to wind and temperature profiles collected during a field campaign conducted during the episode, and PM concentration has been compared to data recorded by air quality stations. The contribution of the valley tributaries to air pollution within the section considered has been first studied. Results show that this contribution is very small compared to that from local sources; this implies that $in-situ$ emissions are primarily responsible for the high PM concentration recorded in the valley since the valley bottom is decoupled from the atmosphere above the valley. The ventilation characteristics of the valley and their influence on the particulate matter (PM) concentration have next been investigated. The analysis indicates that due to the limited ventilation around the city of Passy, the hourly variability of PM concentration therein is driven by that of the emissions.