Forecasting drought probabilities for streams in the northeastern United States

Maximum likelihood logistic regression (MLLR) models for the northeastern United States forecast drought probability estimates for water flowing in rivers and streams using methods previously identified and developed. Streamflow data from winter months are used to estimate chances of hydrological drought during summer months. Daily streamflow data collected from 1,143 streamgages from April 1, 1877, through October 31, 2018, are used to provide hydrological drought streamflow probabilities for July, August, and September as functions of streamflows during October, November, December, January, and February. This allows estimates of outcomes from 5 to 11 months ahead of their occurrence. Models specific to the northeastern United States were investigated and updated. The MLLR models of drought stream-flow probabilities utilize the explanatory power of temporally linked water flows. Models with strong drought streamflow probability correct-classification rates were produced for streams throughout the northeastern United States. A test of northeastern United States drought streamflow probability predictions found that overall correct-classification rates for drought streamflow probabilities in the northeastern United States exceeded 97 percent when predicting July 2019 drought probability using February 2019 monthly mean streamflow data. Using hydrological drought probability estimates in a water-management context informs understandings of possible future streamflow drought conditions in the northeastern United States, provides warnings of potential future drought conditions, and aids water-management decision making and responses to changing circumstances.