STATISTICAL MODELING OF INTENSIVE TBT MONITORING DATA IN TWO TIDAL CREEKS OF THE CHESAPEAKE BAY

Abstract Results are presented from a three year program of weekly tributyltin (TBT) monitoring at nine stations in two tidal creeks of the Chesapeake Bay. Most stations are near dense moorings of recreational boats. The logarithm of TBT + concentration is shown to have appropriate statistical and mathematical properties for modeling. All stations with a relatively high concentration of TBT + (> 10 ng/liter) show a significant annual variation superimposed upon a decreasing linear trend. The trend slope corresponds to a half-life of about 2–3 years. There is strong evidence that the model residuals are uncorrelated in time. All the high concentration stations in a creek system can be adequately modeled by the same annual component amplitude and phase values and the same linear trend slope, while individual stations differ only in overall concentration.