Bay of Biscay s temperature and salinity climatology

Transcription

1 Bay of Biscay s temperature and salinity Charraudeau R., Vandermeirsch F. Laboratoire Physique Hydrodynamique et Sédimentaire, Ifremer, France Abstract Bay of Biscay is the synthesis of all the in situ data collected in the zone of the bay of Biscay as far as 15 W, in the 20th century, principally by the IFREMER, SHOM, and WDCA (about stations on T and/or S, during the period ). It shows monthly information on temperature and salinity through the whole water column with a 1/5 degree resolution. The resulting atlas was produced by an optimal analysis. This optimal analysis had the particularity of reflecting different scales: shelf and abyssal plain. A fine scale on the coast reveals all the local effects, including plume and upwelling and a large scale on the shelf where a slower signal is obtained due to the large scale circulation. Keywords:, multi-scale interaction, hydrology, optimal analysis. 1. Introduction In order to produce a reliable reference for temperature and salinity fields in the bay of Biscay, the IFREMER has decided to collect all the information from all of the profiles available for the zone and use it to produce an atlas composed of a map and numeric files. The resulting is an optimal analysis of all the historical data from the profiles of temperature and salinity from the SHOM (46%), WDCA (35%), IFREMER(17%), UKHO (1%), and MEDS (1%) in the bay of Biscay. The zone of the bay of Biscay extends from 15 W to 1 W in longitude and 50 N to 43 N in latitude. The database covers a long period from 1862 until today with an irregular distribution. Optimal analyses shown in this atlas are limited by the nature of the database (data are scattered in space, especially for high depth), characteristics of the optimal analysis techniques, and the grid used. Annual, seasonal, monthly, interannual analyses have been computed for temperature and salinity. The resulting is unique in two ways; firstly, this database is the first of its kind, indeed no such database has ever been established before for the bay of Biscay. Secondly, this is the first time that different scales have been used in the analysis, thereby reflecting the physical properties of different circulations. 2. Data distribution Profilers, buoys, bottles, CTD and XBT profiles are used in this project. Quality control has been done by selecting the data in the format and inconsistent measure are eliminated. To have enough data for each level, we have made a vertical interpolation of observed level data to standard levels which are used for our analysis. We have described the data by spatial distribution (Figure 1), distribution by depth after the vertical interpolation (Figure 2), and temporal distribution (Figure 3). 1

2 Figure 1 : Spatial distribution Bouteilles CTD Bathythermographes nb profils année Figure 3 : Temporal distribution Figure 2 : Distribution by depth for January We have deduced that these analyses are consistent in a spatial order and that they are more representative from the end of the 20th century. Below 2000m, we have got less data, so the maps produced for these depths are less precise. 3. Optimal estimation method The analysis is derived from optimal estimation methods as exposed by Bretherton & a, The solution is resolved as an anomaly relative to a monthly profile s. This profile s is a spatial mean of the data by depth as shown in Figure 7. For each monthly analysis, all the data of the month are collected and converted to anomalies relative to the to build the data vector [d]. The covariance matrices involved in the equation: x = x + C ( C + R) a f 1 ao o d are constructed using the gaussian structure functions in space including the data noise. These covariance matrices integrate different scales according to different area, like shelf or abyssal plain. The variances are deduced from the dataset. After calculating the profile s, we have produced analyzed fields in temperature and salinity on 261 standard levels between 0 and 4000m, for each month. We have deduced the seasonal and annual from the monthly by a simple mean. The grid spacing is 1/5. 4. Multi-scale influence 2

3 The particularity of our project is that we have respected different scales. Indeed near the coast, predominant effects are local effects like plume or upwelling, and on the contrary on the shelf the variability is modified to large scale circulation. To respect these physical properties we modified the optimal analysis by changing the value of the radius of influence in the gaussian structure functions. We can see in Figure 4 that a low value for the radius of influence, for example 50km, shows the plume of the Gironde. A high value for the radius of influence, such as 300km (Figure 5), better respects the average tendency that corresponds to large scales. What we wanted to do is to blend both these maps, taking the coastal information from the first one and the large scale information from the second one. To mix this both scales we have summed two covariance matrices one with a low scale and one with a large scale (Autret and Gaillard, 2004). Figure 4 : Optimal analysis with a radius of influence equal to 50km Figure 5 : Optimal analysis with a radius of influence equal to 300km 5. Climatologies This project generated different climatologies (Figure 6). First we have the monthly profiles, used in the optimal estimation to compute the monthly and the interannual. With the monthly, we obtain different kinds of information like seasonal and annual, and monthly vertical sections. From the monthly we can compute the monthly density and the monthly geostophical stream. of profiles Multi scale optimal estimation version 1.1 Annual Seasonal vertical sections density Interannual Annual Geostophical stream Diagram T/S Montlhy Geostophical stream Figure 6 : Pattern of filiations of the different climatologies 3

4 5.1. profile s Optimal estimation computes temperature and salinity anomalies. These anomalies are calculated from a reference defined within the set of data. We have a reference for each month, which is a profile of temperature or salinity (Figure 7) corresponding to the mean by depth of all the data of the month. Figure 7 : Mean profiles for July for temperature and salinity The is defined on 261 levels from 0 m to 4000 m in temperature and salinity in the bay of Biscay (Figures 8 and 9). With each map, we supply the corresponding error field map, which give information on the reliability of the map. This map field error is linked to the density and the spatial repartition of the data Seasonal and annual From the monthly, we have deduced two other climatologies, the seasonal and the annual. Due to the monthly distribution of data, we have preferred to average the monthly information rather than to compute a seasonal or annual optimal estimation. Indeed in this case, the estimation will be biased by the month which has the most data, so the result of an optimal analysis will correspond more to the month with the most the data Interannual A new type of information which is presented by our study is the inter annual. We wanted to show the annual changes in temperature and salinity over 50 years. For a fixed month, for every year, we calculated the maps of temperature of salinity by depth, by using a temporal window of 6 years around the year, which we made slide over 50 years. Construction of this is currently underway. 4

5 0 m 100 m 1000 m Figure 8 : Maps of temperature (top panels) and maps of the confidence in the corresponding estimation (bottom panels), for January for depths 0, 100 and 1000m. January March September Figure 9: Maps of salinity (top panels) and maps of the confidence in the corresponding estimation (bottom panels), for January, March and September Vertical section We looked at the horizontal information by supplying maps with latitudes and longitudes by depth level (Figure 10). It is interesting to look at the temperature and at the salinity on the vertical line by creating profiles following latitude or longitude. With these vertical sections we can check if we have some density inversion. 5

6 Figure 10: Zoom of a vertical section of temperature for January for the latitude 46 North. 6. Conclusion Although this work is still in progress, some conclusions can already be drawn. The database is impressive and a high quality of analysis has been obtained. The multi-scale influence gives some good results. Indeed we have succeeded in representing different scales on the same map. The diversity of information (monthly, seasonal, annual, interannual and vertical information) gives a new and complete spectrum of the temperature and salinity of the bay of Biscay. Acknowledgements This work is supported by the project Défi Golfe de Gascogne and Océanographie Côtière Opérationnelle of the IFREMER. We express sincere gratitude to all of the data contributors. We thank A. Bonnat for his special contribution with regard to the data, in particular for the digitalization of historic data and M. Fichaut for her collection. We thank the SHOM for having authorized us to use their data within the framework of this study. We thank the CORIOLIS project for supplying the matlab and fortran optimal analysis model. References Bretherton, F.P., Davis R.E., 1976 : A technique for objective analysis and design of oceanographic experiments applied to MODE-73. Depp-Sea Research 23: Autret E., Gaillard F., 2004 : Système opérationnel d analyse des champs de température et de salinité mis en œuvre au centre de données CORIOLIS. Version pp. 6