Open AccessAnalysis of National Weather Service Stage Forecast Errors
Author(s) -
Grace Zalenski,
Witold F. Krajewski,
Felipe Quintero,
Pedro J. Restrepo,
Steve Buan
Publication year - 2017
Publication title -
weather and forecasting
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.393
H-Index - 106
eISSN - 1520-0434
pISSN - 0882-8156
DOI - 10.1175/waf-d-16-0219.1
Subject(s) - national weather service , forecast skill , quantitative precipitation forecast , environmental science , stage (stratigraphy) , flood myth , meteorology , forecast verification , climatology , upstream (networking) , lead time , consensus forecast , mean squared error , flood forecasting , drainage basin , forecast error , upstream and downstream (dna) , statistics , econometrics , computer science , mathematics , precipitation , geography , operations management , geology , cartography , paleontology , computer network , archaeology , economics
This paper explores the skill of river stage forecasts produced by the National Weather Service (NWS). Despite the importance of the verification process in establishing a reference that allows advancement in river forecast technology, there is relatively little literature on this topic. This study aims to contribute to this subject. The study analyzed the North Central River Forecast Center’s river stage forecasts for 51 gauges in eastern and central Iowa between 1999 and 2014. The authors explored forecast skill dependence characteristics such as upstream area, water travel time, and the number of gauges located upstream of each forecasting point. They also assessed the influence of rainfall uncertainty on stage error by examining the relationship between the forecast skill and its antecedent 24-h observed rainfall. The results show that when using persistence as a reference for comparison with NWS actual forecasts, the NWS forecasts are better for predictions below and above flood stage. The difference in root-mean-square error (RMSE) between the actual and persistence forecasts ranges between 0.04 and 1.24 ft, and it increases with lead time. Locations with fewer upstream gauges exhibit greater variation in forecast skill than locations that are well gauged, especially at high flood levels. Strong predictive relationships between the physical characteristics of a basin (travel time, upstream drainage area), rainfall quantities, and forecast skill have not been identified.