Premium
Spatial and temporal stream temperature prediction: Modeling nonstationary temporal covariance structures
Author(s) -
Gardner Beth,
Sullivan Patrick J.
Publication year - 2004
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2003wr002511
Subject(s) - covariance , covariance function , environmental science , abiotic component , watershed , spatial variability , hydrology (agriculture) , temporal scales , spatial ecology , streams , scale (ratio) , atmospheric sciences , mathematics , ecology , geology , statistics , geography , computer science , cartography , computer network , geotechnical engineering , machine learning , biology
Stream temperature is a measure of water quality and directly influences both the biotic and abiotic dynamics within the aquatic system. Because of its importance, there is a need to find better methods of monitoring and modeling stream temperatures. Gardner et al. [2003] derived a networked geostatistical model to explain the spatial patterns of stream temperatures based on 72 temperature loggers recording stream temperatures hourly throughout the Beaverkill watershed network, New York, for a single time period. Because temperatures and temperature relationships between stream sections change over the season, the covariance structure of the system is likely also to change with time. The covariance structure of stream temperature data collected throughout the month of July was examined and found to be nonstationary temporally. However, the observed changes in the covariance structure were found to be highly dependent on main stem stream temperatures over time. Five nested correlation models were created and compared using Mallow's C p . One model representing large‐scale variation in the sill as a linear function of the main stem temperature for each day was selected as the most parsimonious model.