1. Introduction

The investigation of the hydrographic regime of the Gulf of Riga as well as the water exchange with the Baltic Proper is one of the basic aims of the subproject "Water exchange, nutrients, hydrography and database" of "Gulf of Riga Project". Last five years (1991/95) which overlay the project period may be characterized by essential deviations from the mean hydrometeorological conditions as well as by significant changes in the hydrographic regime of the Gulf:

• unusually long (8 years from 1988 to 1995) period of "atlantic" climatic conditions with mild winters, relatively warm summers and increased temperature of the Gulf was ended,
• the distribution of the wind directions over the Gulf of Riga have changed; the winds with NW component (causing an increased water level of the Gulf) were registered 1.3 to 1.7 times more but the winds with SE component 1.5 to 1.8 times less frequently as their long-term probability,
• the period of increased river run-off ended Year 1991, and a trend of the decrease of the salinity of the Gulf was replaced by a contrary one in 1992.
• the increased run-off and the changes of the wind regime have caused (i) the westward shift of the hydrological front in the Irbe strait, (ii) the decrease of the saline inflow in the bottom layer and the stratification of later.

The data from 48 monitoring cruises, performed by Latvian Fisheries Research Institute (LatFRI) and/or Marine Monitoring Center (MMC) are used to characterize the hydrologic regime of the period. The analysis of the hydrologic regime and its trends is performed mainly on the basis of the monthly and annual means (of the whole Gulf, its parts or particular layers), similarly to [2] in chapter 3. Bathymetric database, data processing and visualization program RJL [3] is used elsewhere in the paper. The locations of the hydrological stations visited during monitoring cruises and cross sections referred in the paper see in Fig. 1.

Figure 1. Locations of the hydrological monitoring stations and referred cross sections.

The water exchange between the Gulf and the Baltic Proper is calculated for the half-year periods using a differential enhancement of the previously used method [3], and extending time series until the end of 1995. The calculation results are given in chapter 4.

The time-development of the vertical temperature-salinity structure of the Gulf is of vital importance for the biochemical processes. The seasonal pycnocline model [4] is enhanced with the inclusion of simple accounting for the ice formation [5]. Chapter 5 is devoted to the calculation for the period under consideration by this model.
 

2. Hydrometeorological conditions

2.1. Air temperature

The last five years (1991 to 1995) were warmest during at least last 100 years with the average air temperature 7.1°C in Riga that is 1.2°C above the long term mean. Similar warm five consecutive years were also 1934/38, 1971/75, 1986/90 with the average air temperature 7.0°C. There are only 25% negative deviations from the long-term means observed for the monthly mean air temperatures. The warmest year was 1992, the coldest year was 1993, the largest annual difference in the monthly means was in 1994 as well as the coldest (Feb-94, -9.2°C) and the warmest (Jul-94, 19.3°C) months of the period. The largest positive and negative deviations from the long-term monthly means were, respectively, in Feb-95 (D t=+6.3°C) and Nov-93 (D t=-5.1°C). The seasonal deviations from the long-term means are summarized in table 1, the positive deviations are found mostly in the cold, but negative in the warm half-years.

	1991	1992	1993	1994	1995	1881-1981	1991-1995	D t
Jan	-0.1	0.9	-0.5	-0.6	-2.7	-4.8	-0.6	4.2
Feb	-3.4	0.3	-0.3	-9.2	1.6	-4.7	-2.2	2.5
Mar	1.9	2.9	0.0	-0.1	1.9	-1.6	1.3	2.9
Apr	5.9	4.7	6.5	7.7	6.0	4.7	6.2	1.5
May	9.1	12.0	15.2	9.9	11.2	10.7	11.5	0.8
Jun	14.2	16.3	13.6	13.3	17.5	14.9	15.0	0.1
Jul	17.8	18.4	16.7	19.3	17.4	17.4	17.9	0.5
Aug	18.1	17.9	15.0	17.1	17.4	16.1	17.1	1.0
Sep	12.7	13.3	9.1	13.5	12.3	11.7	12.2	0.5
Oct	8.0	3.8	6.2	5.7	9.5	6.4	6.6	0.2
Nov	4.2	1.5	-3.8	2.0	-0.2	1.3	0.7	-0.6
Dec	0.5	0.9	-0.6	-0.6	-5.6	-2.6	-1.1	1.5
Year	7.4	7.7	6.4	6.5	7.3	5.9	7.1	1.2

Table 1. The monthly and annual means of the air temperature (°C) in Riga and their deviations from the long-term means.
 

2.2. River run-off

The river run-off to the Gulf is calculated from the monthly means of the discharge at LHMA monitoring stations on the four largest Latvian rivers. They control more than 77% of the whole discharge basin of the Gulf [6]. The basin can be logically subdivided into (see also table 2)

• Daugava basin with relatively low run-off module (M₀) is the largest one with several stations along the river and the observation series since 1880s. The data from Kegums station are used for the calculation of the run-off, the observations at Riga station are recalculated to Kegums for last 10 to 15 years multiplying by 0.976.
• Salaca, Gauja and Lielupe basins have greater integral M₀, the observation series are available since 1920s.
• The rest of the basin [7] referred as uncontrolled has spatially different M₀. The hydrologic stations control about 50% of its area, the data series are available since 1940s, 1960s.

2.3. Wind regime

The wind measurements above the water surface that would be the most representative for the Gulf are episodic, therefore the measurements at the coastal meteorological stations have to be used for the analysis. There are several hydrometeorological stations located around the Gulf or on the islands with long (from end of 1940s) observation series. This would allow to find out the most appropriate data to characterize the wind field over the Gulf. The wind velocity and direction measurements are influenced by the local effects (the type and location of relief macroforms, the distance from coastline, the orientation of coastline, trees, buildings) more than any other meteorological parameter. The levels of the exposure of the meteorological stations for different wind directions are summarized in table 4 according to exposure criterion (see table 5, [8]).

	N	NE	E	SE	S	SW	W	NW	Kf	W(m/s)
Virtsu	9	9	7	6	6	6	9	9	7.6	5.1
Pärnu	6	6	5	9	9	9	9	6	7.4	5.2
Kuresaar	7	7	7	9	9	9	9	7	8.0	5.9
Kihnu	7	7	7	9	9	9	9	7	8.0	6.2
Sõrve	7	9	9	9	6	10	10	7	8.4	6.2
Ruhnu	5	5	7	9	9	9	7	5	7.0	5.8
Kolka	9	9	9	8	5	5	6	6	7.1	4.9
Ventspils	10	9	7	7	7	10	11	11	9.0	6.2
Mersrags	7	10	10	10	7	5	5	5	7.4	4.1
Ainazi	9	9	9	9	8	9	10	10	9.1	5.2
Skulte	6	7	5	7	7	6	6	6	6.3	3.9
Daugavgriva	10	10	8	6	6	6	7	7	7.5	5.7
Riga HMO	7	5	5	5	7	6	7	7	6.1	3.6
Riga-airport	7	7	7	7	7	7	7	7	7.0

Table 4. The exposure K_f of the meteorological stations (for different wind direction, and integral). The long-term (until 1963) mean wind velocity W, if available. Data from [8], except Riga-airport.
The exposure of the station would be a subjective parameter, however, it can help to find out the most appropriate station for the wind data and indicate the deformations of the measured wind distributions. The influence of the time-dependent elements of the surface (trees, buildings) on the measurements has to be accounted for comparing measurements at the different stations, especially, in the context of long time series. Thus, the average measured wind velocity at Ventspils station has reduced by 1.1 m/s during the time period from 1966 until 1982, but the average velocity for S, SW and W directions even by 1.8 m/s. No trend is detected at the neighbouring Sõrve station for the same period. One can conclude that the construction works have reduced the exposure of the Ventspils station for the above directions.

Character of respective direction	Exposure Kf
Open sea	11
Closed sea	10
Bight, big lake	9
Big river	8
Land without obstacles	7
Single obstacles	6
Above obstacles	5
Between obstacles	4

There is almost no meteorological station with even K_f distribution for the all directions (see table 4). The stations on the islands (Ruhnu, Kihnu) would have the most favorable location for the wind measurement over the open Gulf. However, their exposure for the NW, N, NE and E winds is not sufficient. It can reduce the actual wind speeds, especially during the spring and summer months with prevailing winds of the above directions. The observed wind velocity at Riga-airport station is significantly lower, however the location ensures uniform K_f for all directions. The data from this station can be used as the characteristic wind velocity over the Gulf multiplying wind velocity by some transition coefficient K_w. The monthly means of the wind velocities [1] for Kihnu, Ruhnu and Riga are summarized in table 6 together with the monthly transition coefficient K_w. However, seasonal dependence of K_w requires further investigation due to higher probability of the winds from the sectors with low K_f during March-August. The K_w range from 1.4 to 1.5 is suitable for the rest of the year.

	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec	Year
Kihnu (K)	7.4	5.7	5.3	5.1	4.9	5.2	5.2	5.3	6.8	7.7	8.1	7.8	6.2
Ruhnu (R)	7.1	5.1	4.1	3.6	3.5	3.8	3.7	3.8	5.4	6.6	7.4	7.2	5.1
Average K, R	7.3	5.4	4.7	4.3	4.2	4.5	4.4	4.6	6.1	7.1	7.8	7.5	5.6
Riga	5.0	4.2	4.2	4.2	4.0	3.9	3.7	3.6	4.3	4.7	5.1	4.9	4.3
KW	1.44	1.30	1.11	1.03	1.05	1.15	1.19	1.27	1.42	1.52	1.51	1.52	1.31

Table 6. Monthly means of the wind velocities at Kihnu, Ruhnu and Riga stations (Yrs. 1974/78, 1981/85, 1987), and transition coefficient K_w for Riga station.
The uncorrected wind data from the Riga-airport meteorological station is used to compare the characteristics of 1991/95 with the long-term means. The seasonal means of the wind velocities are summarized in table 7. The average wind velocities during the last five years were close to the long-term means, the higher wind velocities are found for the winters with increased cyclone activity, the lower are characteristic for the summers and, especially, autumns with higher anticyclone activity as usual. The largest seasonal mean wind velocities are observed during winter-92, the smallest during summer-94.

The distribution of the wind directions for the last five years is essentially different from the long-term mean. Both during the particular seasons and the whole period there is an enlarged repetitiveness (1.3 to 1.7 times above the long-term mean during winters, springs and autumns) of either of W, NW and/or SW winds. These winds are caused by a higher amount of the cyclone events, they increase the water level of the Gulf. At the same time the winds that tend to decrease the level of the Gulf (NE, E, SE and S directions) have become more rare, especially, during winters and summers (repetitiveness is 1.5 to 1.8 times below the long-term mean). The shift of the mean wind direction to the western sector indicates an essential change of the general character of the air mass movements (see table 8).

	1991	1992	1993	1994	1995	1977/90	1991/95	D W
Winter	4.8	5.4	5.3	3.6	4.3	4.5	4.7	0.2
Spring	4.1	4.1	3.5	3.9	3.6	3.7	3.8	0.1
Summer	3.8	4.0	3.6	3.0	3.3	3.6	3.5	-0.1
Autumn	5.0	4.7	3.8	3.9	3.9	4.8	4.2	-0.6
Year	4.4	4.6	4.0	3.6	3.7	4.2	4.1	-0.1

Table 7. The mean seasonal and annual wind velocities W(m/s) at the Riga-airport observation station and their deviation D W (m/s) from the long-term means.

	N	N E	E	S E	S	S W	W	N W	Direction (° )
Winter (%)	5.8	4.7	7.1	15.5	20.2	22.4	13.3	11.1	SW
Spring (%)	14.5	6.4	8.4	12.5	9.7	12.7	10.8	25.1	NW
Summer (%)	12.0	5.7	10.0	8.3	10.9	19.6	15.2	18.5	W
Autumn (%)	3.3	4.6	7.8	18.5	26.0	18.0	13.4	8.5	S
YEAR (%)	8.9	5.3	8.3	13.7	16.7	18.2	13.2	15.8	SW
Winter (D %)	-0.5	-1.8	-5.4	-8.2	+0.3	+5.7	+5.7	+4.3	+530
Spring (D %)	-3.8	-1.3	-1.9	+1.6	-1.3	-0.5	+1.3	+6.1	-10
Summer (D %)	+1.7	+0.1	+3.5	-2.7	-5.8	-2.8	+3.2	+3.0	+210
Autumn (D %)	-1.5	-1.6	-2.9	-1.5	+2.3	-2.5	+5.4	+2.5	+50
YEAR (D %)	-1.1	-1.2	-1.7	-2.7	-1.1	0.0	+3.9	+3.9	+310

Table 8. Seasonal and annual distribution of the winds by direction (%) and the deviations of this distribution (D %) from the long-term means for 1974/1990. The same for mean wind direction.
 

3. Hydrographic regime

3.1. Water temperature

The small depth of the Gulf of Riga ensures a full vertical mixing of the water masses in the cold part of the year and the low thermal inertia. The last causes a good correlation between the mean water and air temperatures, the correlation coefficient for 1963/95 is 0.88. Therefore also the water temperature was above the long-term mean for the 1991/95 period (see table 9). The autumn temperatures were close to the long-term mean, the winter and spring temperatures were definitely increased for all the water column. During summers the water temperatures in the upper layer (0 to 30 m) were more increased (0.8 to 1.1° C) above the long-term mean than temperatures of the deeper layer(0.3 to 0.4° C abovethe mean for 30 to 50 m layer). The largest positive deviation from the long-term mean is observed in the end of Aug-92 (+3.0° C) whilst the largest negative deviation in autumn-93 and winter-94 (-0.5 to -1.0° C).

	1991	1992	1993	1994	1995	1963-1990	1991-1995	D Tw
10th Feb	0.6	1.8	1.5	-0.3	0.7	0.0	0.9	+0.9
10th May	4.2	4.4	4.6	3.0	3.9	2.9	4.0	+1.1
10th Aug	13.6	12.7	12.7	13.4	12.8	12.1	13.0	+0.9
25th Oct	10.0	9.3	7.7	8.4	9.4	9.0	9.0	0.0
YEAR	7.1	7.0	6.6	6.1	6.7	6.0	6.7	+0.7

Table 9. Annual and long-term average seasonal temperatures of the Gulf of Riga (° C). Deviations D T_w from the long-term mean.
The seasonal cycle (see Fig. 2) for the upper 0 to 10 m layer was usual with the largest positive deviations +2 to +3° C during May and August for the time period under consideration whilst the temperatures of the deeper 30 to 50 m layer indicated the large positive deviation from the long-term mean, especially from August until October (up to 4° C above mean).

Figure 2. Long-term mean seasonal cycle of the temperature of upper, lower layers and whole water column (lines). Temperature measurements during 1991/95 (points).

The horizontal distribution of the surface temperature is close to the long-term mean. It can be characterized by (i) the increased temperatures of the deeper part and decreased temperatures of the shallow zones during autumns and winters; (ii) maximum temperatures near river mouths but minimum temperatures in the northern deeper zone (see Fig. 3a) during springs; (iii) almost uniform surface temperatures during summers with exception of possible upwelling situations near the coastline (see Fig. 3b).

Figure 3a. Surface temperature distribution during spring (May-94).	Fig.3b. Surface temperature distribution during summer (Jul-92). Upwelling event near western coast included.

Horizon (m)	0	10	20	30	35	40
Tw (° C)-West	18.9	16.3	11.0	5.2	3.2	3.4
D Tw (° C)-West	+3.1	+1.6	+2.6	+2.3	-0.3	-0.7
Sw (PSU)-West	5.16	5.19	5.38	5.57	5.79	5.97
D Sw (PSU)-West	-0.59	-0.65	-0.69	-0.97	-1.01	-1.00
Tw (° C)-East	18.6	16.4	12.2	4.7	3.3	2.7
D Tw (° C)-East	+2.7	+2.0	+3.9	+2.7	+1.7	+1.1
Sw (PSU)-East	5.80	5.89	6.06	6.34	6.50	6.70
D Sw (PSU)-East	-0.67	-0.61	-0.59	-0.68	-0.79	-0.88

Table 10. Average temperature (T_w) and salinity (S_w) at the different depths for the western (station 142) and the eastern (station 121A) parts of the Gulf during Jul/Aug 1991/95 and their respective deviations (D T_w,_D S_w) from the periods Jul/Aug 1976/80.
A presence of three different water masses (summer, winter and diluted Baltic water) in vertical temperature profiles can be detected during the end of the springs and throughout the summers in the western part of Gulf and often also in the central and southern regions. Then the temperature minimum of the depth profile can be found at the intermediate 20 to 30 m depths, whilst the temperature of the lower 30 to 40 m layer is up to 6° C higher [2]. This is caused by the movement of the Baltic water along the western coast to consecutively southern, central, eastern, and northern parts of the Gulf [9]. The presence of the Baltic water in the central deepest part of the Gulf is observed only occasionally for the 1991/95 period, therefore the vertical temperature (and also salinity and oxygen) structure differs from the one from 1960s until 1980s. The peculiarities of the vertical temperature distribution are the most distinct during the periods of maximum stratification. Therefore the averaged depth profiles of temperature and salinity from the observations during July and August are used for comparison of two hydrologically different periods, namely 1976/80 (cold period with high salinity) and 1991/95 (warm period with low salinity). The averaged temperatures for both periods in different horizons are summarized in table 10 for western and eastern parts of the Gulf. The comparison indicates the large increase of the temperatures of the upper 30 m, this increase is lower for the deeper layer of eastern part. Contrary, the deeper layers of the western part have become even colder (and also less saline) that indicates the decrease of the salinity of the inflow from the Baltic.
 

3.2. Salinity

The time-dependence of the salinity can be characterized by higher inertia in comparison with the thermal inertia. It is related to the intensive water exchange with the Baltic Proper, and the influence of the river run-off is observable with essential shift in time. The decrease of the salinity of the Gulf of Riga started Year 1977 (with maximum salinity since 1924) and continued with minor fluctuations until 1992 when the lowest average annual Gulf salinity since at least a middle of 50s was observed (5.46 PSU). Similar low salinity was observed during 1928/31 [10]. The trend of the decrease of the salinity was caused mainly by the increased river run-off from the end of 70s until the beginning of 90s (see Fig. 4).

Figure 4. Perennial mean water salinity and river run-off.

The salinity during last five years was 0.4 to 0.5 PSU below the long-term mean (see table 11) the largest deviation is found during May-92.

	1991	1992	1993	1994	1995	1963-1990	1991-1995	D Sw
10th Feb	5.59	5.56	5.62	5.62	5.71	6.06	5.62	-0.44
10th May	5.50	5.25	5.49	5.50	5.55	5.90	5.46	-0.44
10th Aug	5.52	5.46	5.61	5.43	5.48	5.94	5.50	-0.44
25th Oct	5.50	5.56	5.55	5.62	5.63	5.99	5.57	-0.42
YEAR	5.53	5.46	5.57	5.54	5.59	5.97	5.53	-0.44

Table 11. Annual and long-term average seasonal salinity of the Gulf of Riga (PSU) and its deviations D S_w from the long-term mean.
The seasonal variation of the salinity was usual (higher salinity during winters but lower during summers) except for weakly developed summer maximum of the salinity in the deep layer. Thus, the mean spring and summer salinity of upper 20 m was 0.36 to 0.40 PSU below the long-term (1963/90) mean whilst for the deeper layer the same comparison gives decrease by 0.49 to 0.53 PSU. This is caused by the more intensive vertical mixing and the lowered salinity of the inflow from the Baltic Proper.

The distribution of the surface salinity is dependent on the character of the water mass dynamics [9, 11]. During autumn, winter and the beginning of spring the prevailing winds with southern component cause movement of the brackish water of the southern part of the Gulf along its eastern coast. Contrary, during the end of spring and throughout the summer the higher repetitiveness of the northern winds causes inflow through the northern straits and spreading of the brackish water along the western coast. Typical spring and summer distributions of the surface salinity are shown on Figs. 5ab. The largest horizontal salinity gradients during springs are characteristic for the southern part of the Gulf (Fig. 5a).

Figure 5a. Surface salinity distribution during spring (May-94).	Figure 5b. Surface salinity distribution during summer (Aug-93).

The brackish water spreads along the eastern coast, wind events with northern component cause the saline inflows through the northern straits. These inflows usually are more developed during summers, when the fluctuations of the governing winds lead to frequent interchanging of the circulation type of the water masses. As a result the lowered salinity can be found both near eastern and western coasts of the Gulf but the largest horizontal gradient is typical for the Irbe strait (Fig. 5b).
 

3.3. Upwelling

The upwelling situations in the Gulf of Riga can be found from May until September, i.e. during a presence of the vertical stratification. An upwelling situation is caused by the long time period of unidirectional wind that establishes the divergence region in a coastal zone. There are five situations found by the monitoring cruises with the presence of the water from deeper layers in the coastal zone during 1991/95. The most distinct was the upwelling event in the second half of September 1995.

The increased repetitiveness of the winds of E and SE sectors was observed since the beginning of Sep-95, followed by pure eastern winds from 12th until 17th of September (see Fig. 6).

Figure 6. The time dependence of wind projections (Riga station) and selected water levels preceding upwelling event on Sep-19, 1995.

This led to intensive decrease of the water level in the eastern and southern part of the Gulf followed by an establishment of the prolonged upwelling zone from the mouth of river Gauja to Ragaciems. This event is especially interesting due to the presence of thick upper mixed layer (up to 30 m in northern part of the Gulf, see also Fig. 7b) at the end of September.

Figure 7a. Temperature distribution in SW-NE cross section (Engure-Vitrupe) during upwelling event Jul-92, see fig.1, cross section 1.	Fig.7b. Temperature distribution in NW-SE cross section (central part of the western region-Daugavgriva) during upwelling event Sep-95, see fig.1, cross section 2.

The summary of the upwelling situations during last five years is given in table 12 but the typical temperature distribution in the vertical cross sections in Fig. 7. Each event is characterized by the water temperature in the upwelling region (T_wa), its difference from the temperature of upper 5 m in the central part of the Gulf (D T_w), thickness of the upper mixed layer (h), area of the upwelling region (F), means of the wind velocity (W) and direction (a_w) during five preceding days, the mean water levels at Daugavgriva (h_D), Salacgriva (h_S) and Kolka (h_K) stations.

Date	7.May-91	24.May-92	28.Jul-92	17.Sep-94	19.Sep-95
Region	E	SW	W	S	S
Twa (°C)	2.8	7.5	7.1	11.4	7.5
D Tw (°C)	-2.8	-2.6	-8.9	-3.6	-6.5
h (m)	10-15	5-10	20	10-15	15-20
F (km2)	500	350	250	100	1200
W (m/s)	3.8	3.1	4.0	2.8	4.2
aw	NE	none	W	S	E
hD (cm)	-18	0	9	4	-39
hS (cm)	-35	-4	3	-2	-51
hK (cm)	-22	7	9	6	-23

Table 12. Summary of the upwelling events 1991/95.
The analysis of the table 12 allows to draw the following conclusions:

• The upwelling follows about five days long unidirectional wind event, it may occur after 2 to 3 days during springs when the upper mixed layer is relatively thin.
• The typical area of an upwelling zone is 200 to 500 km²; it can increase up to 1000 km² in extreme cases.
• During all the upwelling events the water level in the western part of the Gulf is essentially higher than in the eastern part.

3.4. Hydrologic front in Irbe strait

The hydrologic front in the Irbe strait or near it that separates the saline Baltic water from the brackish Gulf water plays important role in the water exchange between the Gulf of Riga and the Baltic Proper. The bathymetry of the strait (shoal western part with the narrow channel of maximum depth 22 m, and increasing depths towards eastern part up to 30 to 32 m) in a combination with the location of hydrological front determine the intensity of the dense Baltic inflow. Thus, location of the front in the shallow western part or outside the Gulf increases probability of the vertical mixing, reduces the horizontal gradient on the meridional axis of the strait as well as the vertical density gradient.

The investigations of the front location and orientation, gradients of oceanographic parameters, and currents indicate strong temporal and spatial variation of these parameters. Typical are S-shaped form of the front, its orientation from SE to NW with geostrophic flow patterns near frontal zone [12, 13, 14]. The results of above papers agree with the general concept about the water circulation in estuaries, i.e. less dense water leaves the Gulf along the northern coast, but denser water enters the Gulf along the another coast. This scheme is superposed by the morphometric peculiarities (depth distribution, coastline configuration and orientation) and the wind forcing.

The analysis of perennial and seasonal variations of the water exchange parameters in relation with the hydrological parameters in the Irbe strait and external (forcing) factors is performed on the basis of indirect data due to the lack of the detailed studies of the frontal zone. The regular measurements of the vertical TSO profiles are performed at the station 114A (see Fig. 1, h=32 m, 236 profiles during last 25 years) in the deeper part of the Irbe strait. The analysis of the salinity series in this station allows to draw the following conclusions concerning the features of the frontal zone:

• The seasonal cycle of the upper layer (0 to 10 m) salinity is weak, whilst it is well developed for the lower layer (20 to 30 m) due to the thermal stratification during the warm half-year, and the seasonal variations of the wind velocity and direction. Thus, the salinity difference D S_v for both layers is only 0.1 to 0.3 PSU from November until March, the maximum D S_v is characteristic for May and June (0.7 to 0.8 PSU), i.e. the period of the initial thermal stratification, D S_v»0.6 PSU during July and August, it continuously decreases during autumn.
• There exist a good correlation between D S_v and the cumulative mean wind parameters. The D S_v increases with the increase of the value of the cumulative wind projection on the axis of the strait (i.e. NE component of the resulting wind vector) but decreases with the increase of the mean absolute wind velocity. The maximum correlation coefficients are found for the cumulating period 15 to 20 days. These coefficients are 0.55 and -0.46 for NE component and the absolute velocity of the wind, respectively; the limit of statistical significance for 168 trials are ra _=0.01=0.21. Time period 15 to 20 days can be chosen as an adaptation period of the water masses to the external dynamic influence. The cumulating period 20 days are used elsewhere further for the calculation of the means for the wind characteristics before the hydrological measurements.
• The winds of NE sector are favourable for the two-layer exchange, contrary to SW and W winds. The last ones usually have 1.5 to 2 times higher velocity, therefore these winds ensure homogenization of the shallow western part of the strait thus separating the water of the deeper layer of the Irbe strait (and also the Gulf of Riga) from the Baltic water.
• The influence of NE winds on D S_v is higher during the summers, mainly due to the wind regime. Firstly, the higher wind velocities during winter reduce D S_v mixing the water column. Besides, during winters the governing S winds cause higher repetitiveness of outflow events through the northern straits whilst inflow predominates there during the summers. The last changes the structure of the upper layer flow in the Gulf of Riga, i.e. brackish water spreads along the western coast reducing the upper layer salinity in the Irbe strait during summers [9, 11].
• The vertical salinity difference in the Irbe strait has reduced during last five years for the periods of thermal stratification, from May to August. Respective D S_v was 0.49 PSU that is 1.3 times below D S_v for summers 1971/90 and even 1.6 times below D S_v for summers 1976/80 with the maximum salinity of the Gulf of Riga. The reasons of this decrease are (i) high winter run-off that leads to mixing of increased amount of the riverine water in the whole water column and consequently, weakens the spring-summer vertical stratification, and (ii) increase of the repetitiveness of western sector winds during 1991/95.

Figure 8. Distribution of salinity during June 1978 in the vertical cross-section from Irbe strait (station 114a) to the central part of the Gulf (station 135), see fig.1, cross section 3.

4. Water exchange

The water exchange with the Baltic Proper is among the most important factors as river run-off and the processes on the sea-atmosphere, influencing the hydrological regime of the Gulf of Riga. The water exchange is affected by freshwater budget, ice cover, fluctuations of density gradient in the straits, water mass dynamics of the Baltic Proper, the circulation of the Gulf due to wind, air pressure and the water density fields. The water budget calculation results depend on the calculation method, selection of the input parameters, and time scale. As a consequence the published data differs 2 to 5 times [3, 15, 16]. The summary of the previous investigations of the water exchange allows to distinguish three basic water exchange modes with different impact on the hydrological regime of the Gulf:

1. The advective two-layer exchange in the Irbe strait determined by the river run-off to the Gulf and the water density difference between the Gulf and the Baltic Proper [16]. The one-layer two-direction exchange (inflow along southern, outflow along northern coast) can be considered as a special case of this mode. This type of exchange ensures continuous saline water inflow that penetrates the deeper layers of the Gulf through the western region affecting its vertical stratification, especially, during the summer months.
1. The reversive one-layer water exchange characteristic mostly through the northern straits driven by the wind projection on S-N direction [17].
1. Pulsating exchange due to the daily or synoptic fluctuations of the currents [16, 18, 19]. This exchange mode determines the short-period water level fluctuations but its impact on the salinity regime, stratification and long-term circulation patterns would be minor.

(1)

(2)

(3)

1. Monthly river run-off and the water level at Kolka station for r_q and V, respectively.
1. Average salinity of the upper 50 m at the Baltic observation stations BY15 and 40A available four times per annum for S_B.
1. Average salinity of the Gulf calculated by RJL from all cruises in the period under consideration (if number of hydrological stations per cruise exceeds 7) for S.

The calculations by the differential Knudsen¢ s method generally do not differ much from the previously used method [3]. The average annual water exchange for the whole time period Q_in =111 km³, Q_out =144 km³ that is 1.5 to 5 times less than the results of other authors [15, 16]. This disagreement is mainly due to the different selection of S_B and S values. The average values of the water exchange for the different half years almost do not differ (52.7 km³ and 57.2 km³ for winter and summer Q_in, respectively), but the difference between the minimum and maximum values for the winter periods is 1.4 times higher than for the summer periods. The extreme periods are found in the winters of 70s (maximums during winters 1972/73, 1978/79, minimums during 1973/74, 1979/80) when the very intensive inflows of the Baltic water were followed by minimum water exchange during next 6 to 12 months.

No correlation of the calculated water exchange values with the Baltic and/or Gulf of Riga salinity, river run-off and water level fluctuations was found. Analysing the correlation of the calculated water exchange with the averaged wind vector projections, the maximum of the correlation function was found for the wind projection on NE-SW axis. The correlation is statistically significant for summers (r=0.71, significance limit is ra_=0.01=0.54). Due to similar correlation between D S_v and the wind northeast component (see chapter 3.4) it may be concluded that the wind regime affects not only the location of the hydrological front and the vertical salinity difference but also the intensity and character of the water exchange. During the winters when the water column is mixed and the unidirectional flows may prevail the water exchange dependence on NE winds is weaker. Contrary, the two-layer advective water exchange during the summers is dependent on the favorable wind events.

The average annual water exchange during last 5 years was 9 to 10% higher than the long-term mean. It was 3.8 km³ lower for the winter periods but 14 km³ higher for the summers. In comparison with the maximum water exchange years (1975/79) the period under the consideration indicates the same water exchange for summers but 13% decrease of water exchange for the winter periods. The increase of the western wind repetitiveness and salinity regime of the deeper layer allow to conclude that the two-layer advective water exchange is reduced during the period under consideration and the increase of the water exchange is related with the one-layer exchange mode. This conclusion agrees with ones about decrease of the vertical density difference and the movement of the hydrological front westwards.
 

5. Dynamics of vertical TS structure

5.1. Forcing data

The following data are used for the forcing of the model of the dynamics of the vertical structure of the temperature and salinity [5]:

1. Initial conditions, i.e. the temperature and salinity depth distributions from the sampling at 121 hydrological station on 19th May, 1992, at 18⁰⁰.
1. Meteorological conditions: cloudiness, air temperature and humidity, wind velocity from the observations at the Riga station, eight times daily.
1. The river run-off, inflow of the Baltic Proper water, and its salinity as in chapter 4.

Figure 9. Monthly means of forcing data (May-92 for last decade)

5.2. Calculation for 1992/95

The calculated time development of the temperature depth distribution is shown on Fig. 10.

Figure 10. Calculated time-depth distribution of temperature.

 The time dependence of the surface and bottom temperature including comparison with the observations at station 121 (see fig. 1) is given on Fig. 11.

Figure 11. Time-dependence of calculated and observed surface (Ts) and bottom (Tb) temperatures.

 The calculated development of fluxes on the atmosphere-sea interface is shown on Fig. 12.

Fig.12. Calculated means of heat flux (Q) components: insolation (I0), back radiation (Qbr), sensible (Qs) and latent (Ql) heat fluxes.

 Analysing calculations one can conclude that

• the forcing with high time-resolution reasonably fetches short-term meteorological peculiarities and single events, for instance, storms;
• each of individual years has different scenario of stratification development;
• the calculation results indicate a reasonable agreement with the measurements at single station, all basic characteristics as the surface and bottom temperatures, the moments of spring retreat of pycnocline and full mixing of Gulf in autumns, the duration of the ice periods and their existence is present;
• the model is sensitive to the spring development of the pycnocline that determines the bottom temperatures during summers;
• the calculation results are suitable for the filling of gaps between the measurements, for instance to calculate the primary production;
• the components of the heat flux on the sea-air interface are side product of the model; the high time-resolution of them allows their utilization for different purposes, for instance, regional climate model.

Fig.13. Calculated dynamics of real (hreal) and effective (heff) thicknesses of ice cover over Gulf of Riga from 1-Feb-94 until 31-Mar-94

Acknowledgments

This work has been in part supported by grant No. 93.353 of Latvian Science Council and the Finnish Institute of Marine Research within the framework of the "Gulf of Riga Project" funded by the Nordic Council of Ministers. Authors acknowledge Dr. Jevgeniy Zaharchenko (LHMA) for the access to the hydrometeorological data necessary for performed analysis, and Edmunds Jula (LatFRI) for the permission to use the salinity observations in the Baltic Proper.
 

References

1. Meteorological bulletins of the Latvian Hydrometeorological Agency. Tables TGM-1, TGM-1M. Technical report, LHMA, 1974-1995.
2. V. Berzins. Hydrology. In: Ecosystem of the Gulf of Riga between 1920 and 1990. Ed. Ojaveer. Estonian Academy Publishers, Tallinn, 1995, pp. 7-31.
3. V. Berzins, U. Bethers, J. Sennikovs. Gulf of Riga: bathymetric, hydrological and meteorological databases, and calculation of the water exchange. Proc. of the Latvian Academy of Sciences. (7/8):107-117, 1994.
4. J. Sennikovs, U. Bethers. Modelling of the vertical temperature and salinity structure of the Gulf of Riga. Latvian Journal of Physics and Technical Sciences, (1):19-41, 1995.
5. J.Sennikovs, U.Bethers. Model for the vertical structure of physical fields in Gulf of Riga with account for ice formation. This volume.
6. Surface water resources of USSR. Basic hydrological characteristics. Vol. 4, Iss. 1. Estonia, Iss. 2. Latvia, Gidrometeoizdat, Leningrad, 1965, 66, 73, 76, 78. In Russian.
7. Surface water resources of USSR. Hydrological inventory. Vol. 4, Iss. 2. Latvia, Gidrometeoizdat, Leningrad, 1964. In Russian.
8. Handbook of USSR climate. Iss. 4. Estonia, Iss. 5. Latvia. Part 3, Wind. Gidrometeoizdat, Leningrad, 1966. In Russian.
9. V.Berzins. Hydrological division into districts of the open part of Gulf of Riga. In: Hydrochemical and hydrobiological descriptions and division into districts of the littoral part of the Baltic Sea, the Gulf of Riga, and the Gulf of Finland. Zinatne, Riga, 1987, pp. 7-20. In Russian.
10. E. Aunins. Ion content and salinity of the water of the Gulf of Riga. RGMO Series, (10): 55-79, 1966. In Russian.
11. V.Berzins. Interannual and seasonal changes of water salinity in the Gulf of Riga. In: Fisheries' research in basin of Baltic Sea. Zvaigzne, Riga, pp. 3-12.
12. U.Lips, M.-J. Lilover, U.Raudsepp, L. Talpsepp. Water renewal processes and related hydrographic structures in the Gulf of Riga. EMI Report Series, (1): 1-34, 1995.
13. U.Lips, J.Laenearu, L.Maegi. Hydrographic measurements by R/V Kiir. EMI Report Series, (2): 12-29, 1995.
14. T.Kouts. Thermohaline structures in Gulf of Riga. SMHI RO Series, (23): 4-19, 1995.
15. A.Pastors. Water and heat budget of the Gulf of Riga. In: Gulfs as run-off targets. Riga, 1967, pp. 8-33. In Russian.
16. V.Petrov. Water budget of the Gulf of Riga and water exchange with Baltic Proper. RGMO Series (18): 20-40, Leningrad, 1979. In Russian.
17. H.Maardiste. A note on the currents in the Muhu straits. In: Dynamics of the Baltic sea water. Tallinn, 1975, pp. 88-93. In Russian.
18. A.Pastors. Water circulation in the Gulf of Riga and the Irbe strait. RGMO Series (15): 99-123, Leningrad, 1976.
19. A.Baranov. Currents in the Gulf of Riga and the Irbe strait. Riga, 1970, 86 p.