Climate change raises concern that risks of hydrological drought may be increasing. We estimate hydrological drought probabilities for rivers and streams in the United States using maximum likelihood logistic regression (MLLR). Streamflow data from winter months are used to estimate the chance of hydrological drought during summer months. Daily streamflow data collected from 9,144 stream gages from January 1, 1884 through January 9, 2014 provide hydrological drought streamflow probabilities for July, August, and September as functions of streamflows during October, November, December, January, and February, estimating outcomes 5-11 months ahead of their occurrence. Few drought prediction methods exploit temporal links among streamflows.
We find MLLR modeling of drought streamflow probabilities exploits the explanatory power of temporally linked water flows. MLLR models with strong correct classification rates were produced for streams throughout the U.S. One ad hoc test of correct prediction rates of September 2013 hydrological droughts exceeded 90% correct classification.
Results:
As the global climate has warmed concern has increased that climate change will influence water flow in streams and rivers. In this study we use maximum likelihood logistic regression (MLLR) to estimate hydrological drought probabilities for gaged rivers and streams throughout the U.S. from 5 to 11 months ahead of their occurrence. Our results demonstrate the utility of using well-formed MLLR models to estimate summer month hydrological drought streamflow probabilities from streamflow data collected during previous winter months. We find that modeling hydrological drought streamflow probabilities in this way exploits the explanatory power of temporally linked water flows, even without full knowledge of the nature and timing of surface and groundwater flows specific to each stream.
Equations developed using MLLR to calculate hydrological drought probabilities exhibit strong correct classification rates in most areas of the U.S., predicting hydrological drought outcomes up to 11 months in advance of their occurrence. The predictive equations may be used to communicate local and regional drought conditions, improve drought awareness, implement drought management plans, and test water allocation protocols and decision rules. Work to improve MLLR model performance using new combinations of hydrological drought threshold percentiles and new explanatory variables will leverage the explanatory power of temporally linked water flows, helping drive progress in the assessment and prediction of hydrological drought.
While past studies have reported low-flow characteristics and have applied various methods to drought modeling and evaluation, few have dealt specifically with forecasting hydrological drought. Estimates of summer month hydrological drought probability provide advanced warning of drought conditions, extending the lead time for drought awareness, management response, and making water withdrawal decisions.
Terms, Statistics, and Performance Measures for Maximum Likelihood Logistic Regression Models Estimating Hydrological Drought Probabilities in the Northeastern United States (2019)
Terms, Statistics, and Performance Measures for Maximum Likelihood Logistic Regression Models Estimating Hydrological Drought Probabilities in the Delaware River Basin (2020)
Terms, Statistics, and Performance Measures for Maximum Likelihood Logistic Regression Models Estimating Hydrological Drought Probabilities in the United States
Modeling summer month hydrological drought probabilities in the United States using antecedent flow conditions
Modeling summer month hydrological drought probabilities in the United States using antecedent flow conditions
Methods for estimating drought streamflow probabilities for Virginia streams
Low-flow characteristics of Virginia streams
Interactive Map: Northeast Region Drought Streamflow Probabilities
This application allows the display and query of drought probability. Maximum likelihood logistic regression is used to estimate drought probabilities for selected Northeast region streams.
Interactive Map: Estimating Drought Streamflow Probabilities for Virginia Streams
Maximum likelihood logistic regression is used to estimate drought probabilities for selected Virginia rivers and streams 5 to 11 months in advance. Hydrologic drought streamflow probabilities for summer months are provided as functions of streamflows during the previous winter months. This application allows the display and query of these drought streamflow probabilities for Virginia streams.
- Overview
Climate change raises concern that risks of hydrological drought may be increasing. We estimate hydrological drought probabilities for rivers and streams in the United States using maximum likelihood logistic regression (MLLR). Streamflow data from winter months are used to estimate the chance of hydrological drought during summer months. Daily streamflow data collected from 9,144 stream gages from January 1, 1884 through January 9, 2014 provide hydrological drought streamflow probabilities for July, August, and September as functions of streamflows during October, November, December, January, and February, estimating outcomes 5-11 months ahead of their occurrence. Few drought prediction methods exploit temporal links among streamflows.
We find MLLR modeling of drought streamflow probabilities exploits the explanatory power of temporally linked water flows. MLLR models with strong correct classification rates were produced for streams throughout the U.S. One ad hoc test of correct prediction rates of September 2013 hydrological droughts exceeded 90% correct classification.
Results:
As the global climate has warmed concern has increased that climate change will influence water flow in streams and rivers. In this study we use maximum likelihood logistic regression (MLLR) to estimate hydrological drought probabilities for gaged rivers and streams throughout the U.S. from 5 to 11 months ahead of their occurrence. Our results demonstrate the utility of using well-formed MLLR models to estimate summer month hydrological drought streamflow probabilities from streamflow data collected during previous winter months. We find that modeling hydrological drought streamflow probabilities in this way exploits the explanatory power of temporally linked water flows, even without full knowledge of the nature and timing of surface and groundwater flows specific to each stream.
Equations developed using MLLR to calculate hydrological drought probabilities exhibit strong correct classification rates in most areas of the U.S., predicting hydrological drought outcomes up to 11 months in advance of their occurrence. The predictive equations may be used to communicate local and regional drought conditions, improve drought awareness, implement drought management plans, and test water allocation protocols and decision rules. Work to improve MLLR model performance using new combinations of hydrological drought threshold percentiles and new explanatory variables will leverage the explanatory power of temporally linked water flows, helping drive progress in the assessment and prediction of hydrological drought.
While past studies have reported low-flow characteristics and have applied various methods to drought modeling and evaluation, few have dealt specifically with forecasting hydrological drought. Estimates of summer month hydrological drought probability provide advanced warning of drought conditions, extending the lead time for drought awareness, management response, and making water withdrawal decisions.
- Data
Terms, Statistics, and Performance Measures for Maximum Likelihood Logistic Regression Models Estimating Hydrological Drought Probabilities in the Northeastern United States (2019)
Tables are presented listing parameters used in logistic regression equations describing drought streamflow probabilities in the Northeastern United States. Streamflow daily data, streamflow monthly mean data, maximum likelihood logistic regression (MLLR) equation explanatory parameters, equation goodness of fit parameters, and Receiver Operating Characteristic (ROC) AUC values identifying the utiTerms, Statistics, and Performance Measures for Maximum Likelihood Logistic Regression Models Estimating Hydrological Drought Probabilities in the Delaware River Basin (2020)
Tables are presented listing parameters and fit statistics for 25,453 maximum likelihood logistic regression (MLLR) models describing hydrological drought probabilities at 324 gaged locations on rivers and streams in the Delaware River Basin (DRB). Data from previous months are used to estimate chance of hydrological drought during future summer months. Models containing 1 explanatory variable useTerms, Statistics, and Performance Measures for Maximum Likelihood Logistic Regression Models Estimating Hydrological Drought Probabilities in the United States
A table is presented listing: (1) USGS Gage Station Numbers, (2) Model Identification Tags, (3) Model Term Estimates, (4) Model Term Fit Statistics, and (5) Model Performance Indices for Maximum Likelihood Logistic Regression (MLLR) Models estimating hydrological drought probabilities in the United States. Models were developed using streamflow daily values (DV) readily available from the U.S. Geo - Publications
Modeling summer month hydrological drought probabilities in the United States using antecedent flow conditions
Climate change raises concern that risks of hydrological drought may be increasing. We estimate hydrological drought probabilities for rivers and streams in the United States (U.S.) using maximum likelihood logistic regression (MLLR). Streamflow data from winter months are used to estimate the chance of hydrological drought during summer months. Daily streamflow data collected from 9,144 stream gaAuthorsSamuel H. Austin, David L. NelmsModeling summer month hydrological drought probabilities in the United States using antecedent flow conditions
Climate change raises concern that risks of hydrological drought may be increasing. We estimate hydrological drought probabilities for rivers and streams in the United States (U.S.) using maximum likelihood logistic regression (MLLR). Streamflow data from winter months are used to estimate the chance of hydrological drought during summer months. Daily streamflow data collected from 9,144 stream gaAuthorsSamuel H. Austin, David L. NelmsMethods for estimating drought streamflow probabilities for Virginia streams
Maximum likelihood logistic regression model equations used to estimate drought flow probabilities for Virginia streams are presented for 259 hydrologic basins in Virginia. Winter streamflows were used to estimate the likelihood of streamflows during the subsequent drought-prone summer months. The maximum likelihood logistic regression models identify probable streamflows from 5 to 8 months in advAuthorsSamuel H. AustinLow-flow characteristics of Virginia streams
Low-flow annual non-exceedance probabilities (ANEP), called probability-percent chance (P-percent chance) flow estimates, regional regression equations, and transfer methods are provided describing the low-flow characteristics of Virginia streams. Statistical methods are used to evaluate streamflow data. Analysis of Virginia streamflow data collected from 1895 through 2007 is summarized. Methods aAuthorsSamuel H. Austin, Jennifer L. Krstolic, Ute Wiegand - Web Tools
Interactive Map: Northeast Region Drought Streamflow Probabilities
This application allows the display and query of drought probability. Maximum likelihood logistic regression is used to estimate drought probabilities for selected Northeast region streams.
Interactive Map: Estimating Drought Streamflow Probabilities for Virginia Streams
Maximum likelihood logistic regression is used to estimate drought probabilities for selected Virginia rivers and streams 5 to 11 months in advance. Hydrologic drought streamflow probabilities for summer months are provided as functions of streamflows during the previous winter months. This application allows the display and query of these drought streamflow probabilities for Virginia streams.