BULLETIN OF MARINE SCIENCE, 54(3): 602-ti09, 1994 LONG-TERM GULF-TO-ATLANTIC TRANSPORT THROUGH TIDAL CHANNELS IN THE FLORIDA KEYS Ned P. Smith ABSTRACT Current meter time series collected between 1987 and 1993 from tidal channels in the Middle and Lower Keys quantify long-term net displacement and volume transport between the Gulf of Mexico and Hawk Channel on the Atlantic Ocean side of the Florida Keys. NOAA-supported field studies conducted near Looe Key Marine Sanctuary during autumn, winter and spring months of 1987 and 1988 reveal a quasi-steady nontidal flow into Hawk Channel. Resultant speeds in Newfound Harbor Channel, Bahia Honda Channel and Moser Channel are 0.05,0.11 and 0.04 m·s-I, respectively. More recent studies of Long Key Channel and Channel No. Five show temporal variability over time scales on the order of 1-2 weeks; seasonal variations are not well defined. The long-term net flow is consistently out of the Gulf of Mexico. Investigations conducted as part of the SEAKEYS program included trans- lating cumulative net displacement into cumulative volume transport. Current profiles at an- chor stations under both flood and ebb conditions are used to calibrate two tidal channels, For a 34-day time period in October and November 1990, the resultant volume transport through Bahia Honda Channel is 620 m3·s-l• The tidal contribution to the total is isolated and found to provide a resultant volume transport of 78 m3·s-1 in the opposite direction- into the Gulf of Mexico. A similar analysis of a one-year record from Long Key Channel indicates a resultant volume transport of 262 m3·s-1 with strongest Gulf-to-Atlantic transport in winter and spring months, An understanding of advective and diffusive transport processes can be applied to a wide variety of issues of importance in marine science. Transport mechanisms act as a least common denominator relating studies that deal more specifically with larval transport, nutrient transport, suspended sediment transport and pollut- ant transport among other things. Transport processes in the Florida Keys include the wind-driven circulation in shallow-water areas on both the Atlantic and Gulf sides of the keys; the ebb and flood of the tide, especially through the major tidal passes separating the keys; and the poorly understood density currents, arising from the enhanced response to sensible and latent heat fluxes in the shallower waters on the Gulf side of the keys. Lee (1986) has described wind-driven cir- culation between the keys and the reef tract off the Upper Keys near Key Largo. Lapointe et al. (1993) have reported on similar studies in the waters off Big Pine Key. Pitts (1994) has compared current patterns recorded simultaneously off Key Largo and Bahia Honda Key and found high coherence in along-channel transport. The emerging pattern includes a wind-forced seasonally-varying circulation in the shallow waters on the Atlantic side of the keys. In the Upper Keys, the long- term net flow is northerly in summer months and southerly in winter months. In the Lower Keys along-channel flow toward the northeast is rare. Southwesterly flow is fastest in winter months, when winds are out of the north-easterly quadrant. In summer months, winds are more directly shoreward. Neither the circulation of the shallow waters on the Gulf side of the keys nor the flow through the major tidal channels has been systematically studied and reported. The purpose of my paper is to extend previous work on the regional circulation of the keys in a geographical sense by summarizing results that have come from a series of studies of major tidal passes (Fig. 1). Results of field studies conducted in 1987 and 1988 in the northern part of the Lower Keys are combined with 602 SMITH: CHANNEL TRANSPORT IN FLORIDA KEYS 603 I I ,I ,I , ,, I I I ,I , "," " ,.... '" -. .",; ..,....:. .- ~.'","" ••••• ~., ><'\-" ~ •• . •• }<o • ,. ..F ,"": ",'"'.. L '. ;. :;:,>: ---:N1''1-..•.•.B..MT" -_ ..... ,- . ..,-------..,-----, o 50 100Km Figure 1. Map showing locations of tidal channels investigated between 1987 and 1993: Newfound Harbor Channel (N), Bahia Honda Channel (B), Moser Channel (M), Long Key Channel (L) and Channel No. Five (F). The approximate position of the reef tract is indicated by the hatched line that parallels the keys on the Atlantic side. Hawk channel lies between the keys and the reef tract. results from more recent studies conducted as part of the SEAKEYS program (Ogden et aI., 1994) to characterize long-term net flow patterns. Results show a quasi-steady nontidal flow from the Gulf side of the keys to the Atlantic side, however the data base assembled at this time is not sufficient to establish either the long-term variability or its underlying causes. An estimation of the associated volume transport suggests that the flow through tidal channels may represent a flushing mechanism for waters landward of the reef tract on the Atlantic side of the keys. DATA The data base assembled for describing long-term net flow through tidal channels includes current meter time series, time series of bottom pressure used to represent water level and flow meter profiles needed to quantify bottom roughness. General Oceanics Mark I and Mark II recording current meters were moored approximately 2 m above the bottom in depths ranging from 4 to 5 m. Sampling stations were generally in mid channel at a point where the channel cross-section was at a minimum between the Gulf and Atlantic sides of the keys. Currents were recorded hourly, following a burst sampling of four readings. A vector-average of the burst reduced high-frequency wave effects. The accuracy of the inclination and direction is ± 0.5° and ± 2°, respectively, according to manufacturer's specifications. Within the range of incli- 604 BULLETIN OF MARINE SCIENCE, VOL. 54, NO.3, 1994 nation angles characteristic of the tidal channels studied, the ± 0.50 accuracy translates into an accuracy of approximately ±0.9 cm·s-I• Current vectors were decomposed into along-channel and across- channel components. The along-channel axis was defined as the heading for which the average of the across-channel components was zero. Only the along-channel components were used in the study. Bottom pressures were recorded using a Sea Data TDR-3 pressure recorder. The resolution is 0.004 decibar (db), according to the manufacturer's specifications. A 1 db change in bottom pressure cor- responds approximately to aIm change in water level. Current profiles were obtained using a General Oceanics Model 2035 Mark III direct readout flow meter. Flow meter readings were calibrated at the surface using a tethered drogue. Profiles were constructed by taking readings at approximately I-m intervals. At each level, current speeds were integrated over a one-minute sampling interval to reduce effects of wave noise and boat motion. METHODOLOGY Cumulative net displacement plots have been constructed to show the long-term net flow through the tidal channels. Displacement, DO'is defined to be the product of the hourly along-channel com- ponent current speed and the time interval it represents. Thus, for the m'h hour of the time series In Dc = L u, d.t, (1) 1=1 where u, is the along-channel component of the ith observation recorded by the current meter. Net displacement is obtained by defining the flow leaving the Gulf to be negative, and values are accu- mulated over the entire study period. Plots show both the flood and ebb of the tide (high-frequency positive and negative displacements, respectively), lower-frequency displacement associated with me- teorological forcing in a variety of forms, and any quasi-steady flow due to regional scale slopes in sea level. The local transport at each channel study site is obtained by extrapolating the measured current to the surface and bottom. Current profiles indicated that the current speed, u, as a function of height above the bottom, z, can be represented by the logarithmic expression u* Z Uz = -k In-, (2) Zo where u* is the friction velocity, k is the Karman constant (0.41) and Zo is the roughness length. When current speeds were regressed against the natural log of the height above the bottom to determine zo' the correlation coefficient averaged 0.92. The friction velocity was then obtained using (2), and the vertically-integrated transport at the jth anchor station was determined from Z Tj = Luzdz = u* [z In(~) - (Z - Zo)j, (3) " k(Z - zo) Zo where Z is the sum of the time-averaged water depth and the time-varying deviation about the mean. Z and u* are functions of time and space; Zo varies in space only. The transport for a given segment was obtained by assuming that the measured water depth and current profile at each anchor station represented the entire segment. The segment transport was then obtained by multiplying the vertically integrated transport by the segment width, Ay, and the total transport was obtained by summing the contributions from each segment: 5 TT = L Tj Ay. (4) j=! By sub-dividing the channel and occupying anchor stations during both flood and ebb conditions, ebb- dominant and flood-dominant parts of the channel were identified. Temporal variations in tidal current speed occurring during the approximately 2-h time period needed for channel calibration were removed by interpolation. The current speed at each anchor station was adjusted to a common time using measurements made during two visits. The interpolation procedure assumed that the tidal current was in phase through the entire cross-section of the channel, and that the nontidal flow was constant during the calibration. Given the relationship between the surface current speed at the reference station and the surface current speed at each anchor station, and given the water depth and roughness length at each anchor station, current meter time series from the reference station were converted to volume transport for the channel as a whole.