Premium
Thermal regime metrics and quantifying their uncertainty for North American streams
Author(s) -
Jones N. E.,
Schmidt B. J.
Publication year - 2018
Publication title -
river research and applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.679
H-Index - 94
eISSN - 1535-1467
pISSN - 1535-1459
DOI - 10.1002/rra.3257
Subject(s) - percentile , environmental science , range (aeronautics) , standard deviation , metric (unit) , mean radiant temperature , statistics , ecoregion , confidence interval , hydrology (agriculture) , mathematics , climate change , geology , ecology , biology , operations management , materials science , oceanography , geotechnical engineering , economics , composite material
Understanding and characterizing thermal regimes is gaining popularity, but there has been little assessment of the sources and magnitudes of uncertainty among different thermal metrics. Understanding how the quantity of data influence estimates of metrics and the characterization of thermal regime is critical to resource management. We examine the influence of record length on the uncertainty of estimation for commonly used thermal metrics including mean annual maximum and minimum, timing of the annual maximum and minimum, mean annual temperature range, mean weekly maximum temperature, July maximum, minimum, and range. We selected 19 sites from U.S. Geological Survey hydrometric station network to represent stations with both small and large drainage areas across the ecoregions of the contiguous United States with at least 20 years of daily stream temperature data. We also selected 54 sites from Water Survey of Canada's hydrometric network with at least 7 years of sub‐daily data for the province of Ontario. Randomizing a progressively increasing set of years used to calculate estimates of each metric provided the percentile confidence bands that were compared with various thresholds of acceptable certainty. Bootstrap confidence bands quickly decreased in width with increasing record length and approached an acceptable level at an average of 12 years for daily data metrics. Metrics calculated using the sub‐daily data required approximately 3 years of data. The timing of annual minimum and maximum temperatures required the greatest amount of data to reduce bias to an acceptable level.