Premium
Sensitivity of mean transit time estimates to model conditioning and data availability
Author(s) -
Hrachowitz M.,
Soulsby C.,
Tetzlaff D.,
Malcolm I. A.
Publication year - 2011
Publication title -
hydrological processes
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.222
H-Index - 161
eISSN - 1099-1085
pISSN - 0885-6087
DOI - 10.1002/hyp.7922
Subject(s) - mean transit time , sampling (signal processing) , environmental science , percentile , statistics , sensitivity (control systems) , transit time , tracer , mathematics , hydrology (agriculture) , geology , computer science , physics , detector , medicine , telecommunications , perfusion scanning , electronic engineering , perfusion , engineering , cardiology , geotechnical engineering , transport engineering , nuclear physics
Abstract Mean transit times (MTTs) can give useful insights into the internal processes of hydrological systems. However, varying model conditioning assumptions and data availability can limit the use of MTT, particularly in terms of comparing the results of studies using different assumptions and data records of varying lengths. We present a systematic analysis of sensitivity of MTT estimates to different methods of artificially extending the data record, varying model warm‐up period lengths and varying sampling intervals for a small upland catchment in the Scottish Highlands. The analysis was based on Cl − data in conjunction with the convolution integral model using the gamma distribution as transit time distribution. It could be shown that three out of four different methods to artificially extend the data record and to generate a warm‐up period give mostly equivalent results. The required minimum warm‐up period length to reliably estimate MTT for a 3‐year period of data was observed to be about 2 years or 3 times the MTT, implying that ∼95% of the tracer signal entering the stream at day 1 of the warm‐up period has to be recovered by the end of the warm‐up period in order to avoid significant deviations from the best available MTT estimates. It was furthermore found that sampling intervals of up to 4 weeks can produce MTT estimates within about 0·25 times the best available MTT estimate, albeit with potentially increased process misrepresentation in terms of the gamma distribution parameter α. Copyright © 2011 John Wiley & Sons, Ltd.