Analysis of the streamflow-gaging station network in Ohio for effectiveness in providing regional streamflow information
The streamflow-gaging station network in Ohio was evaluated for its effectiveness in providing regional streamflow information. The analysis involved application of the principles of generalized least squares regression between streamflow and climatic and basin characteristics. Regression equations were developed for three flow characteristics: (1) the instantaneous peak flow with a 100-year recurrence interval (P100), (2) the mean annual flow (Qa), and (3) the 7-day, 10-year low flow (7Q10). All active and discontinued gaging stations with 5 or more years of unregulated-streamflow data with respect to each flow characteristic were used to develop the regression equations. The gaging-station network was evaluated for the current (1996) condition of the network and estimated conditions of various network strategies if an additional 5 and 20 years of streamflow data were collected. Any active or discontinued gaging station with (1) less than 5 years of unregulated-streamflow record, (2) previously defined basin and climatic characteristics, and (3) the potential for collection of more unregulated-streamflow record were included in the network strategies involving the additional 5 and 20 years of data. The network analysis involved use of the regression equations, in combination with location, period of record, and cost of operation, to determine the contribution of the data for each gaging station to regional streamflow information. The contribution of each gaging station was based on a cost-weighted reduction of the mean square error (average sampling-error variance) associated with each regional estimating equation. All gaging stations included in the network analysis were then ranked according to their contribution to the regional information for each flow characteristic. The predictive ability of the regression equations developed from the gaging station network could be improved for all three flow characteristics with the collection of additional streamflow data. The addition of new gaging stations to the network would result in an even greater improvement of the accuracy of the regional regression equations. Typically, continued data collection at stations with unregulated streamflow for all flow conditions that had less than 11 years of record with drainage areas smaller than 200 square miles contributed the largest cost-weighted reduction to the average sampling-error variance of the regional estimating equations. The results of the network analyses can be used to prioritize the continued operation of active gaging stations or the reactivation of discontinued gaging stations if the objective is to maximize the regional information content in the streamflow-gaging station network.