气候变化领域集成服务门户

Wind-driven coastal upwelling along the Pacific Northwest Coast of the USA results in oceanic water that may be periodically entrained into adjacent estuaries and which possesses high nutrients and low dissolved oxygen (DO). Measurement of water quality indicators during these upwelling water entrainment events would represent extreme values for water quality thresholds derived from typical estuarine conditions. Tools are therefore needed to distinguish upwelled waters from other causes of exceedances of water quality thresholds within estuaries of the region. We present an example application of logistic regression models to predict the probability of exceedance of a water quality threshold, using DO data from the Yaquina estuary, Oregon, USA. Models including water temperature and salinity correctly classified exceedances of DO about 80 % of the time. Inclusion of in situ fluorescence in the logistic regression model for DO improved the model performance and reduced the rate of false positives.