Influence of Scale on Thermal Characteristics in a Large Montane River Basin

Abstract

This study monitored stream temperatures over two hydrological years at various nested scales within the large, unregulated river Dee catchment (North East Scotland). These scales were (i) the whole catchment (11 sites along main stem Dee); (ii) the tributary (single sites in main tributaries); (iii) the Girnock (five sites in one subcatchment); and (iv) the reach (26 points across single reach). The aim was to characterize the thermal regime of all locations and compare the magnitude of variation between each scale. The controls on this variation were assessed via a multiple linear regression model using Geographic Information System-derived catchment data. Temperatures were collected at 15-min resolution and for further analysis and discussion combined to daily means. At the catchment and subcatchment scales, a west to east gradient in mean and minimum temperatures was observed, largely paralleling changes in altitude. Temperature differences between subcatchments were generally greater than between the sites along the main stem of the Dee. Differences between tributaries reflected differences in their morphology and land use. However, some tributaries had similar thermal regimes, despite different catchment and riparian characteristics. Subcatchment differences in thermal regimes of one of the tributaries corresponded to riparian vegetation reduced diurnal variability in sections dominated by broadleaf woodland. Compared with the larger scales, reach differences in thermal regime were small (e.g. mean temperatures of riffle, pool and margin habitats were within 0.3 C). The most noticeable difference was in relation to the point samples within the backwater area, which has a more constant thermal regime, most probably reflecting its groundwater source. The regression analysis indicated that monthly mean temperatures can be predicted well using elevation and catchment area. Forest cover was a significant explanatory variable during the summer months. However, some of the empirical temperature data from the Dee indicate that similar thermal regimes can result from different physical controls and processes that have important implications for the extrapolation of such predictive models.

Stream temperature prediction in ungauged basins: review of recent approaches and description of a new physics-derived statistical model

“…Similarly, both Pratt and Chang (2012) and Hill et al (2013) aimed at estimating mean stream temperature in summer and winter. Very few studies have actually attempted Bogan et al (2003) Eastern USA AE 596 30 Week R 2 = 0.80, σ e = 3.1 • C Chang and Psaris (2013) Western USA MLR, GWR 74 n/a Week, year R 2 = 0.52-0.62, σ e = 2.0-2.3 • C Daigle et al (2010) Western Canada Various 16 0.5 Month σ e = 0.9-2.8 • C DeWeber and Wagner (2014) Eastern USA ANN 1080 31 Day σ e = 1.8-1.9 • C Ducharne (2008) France MLR 88 7 Month R 2 = 0.88-0.96, σ e = 1.4-1.9 Gardner and Sullivan (2004) Eastern USA NKM 72 1 Day σ e = 1.4 • C Garner et al (2014) UK CA 88 18 Month n/a Hawkins et al (1997) Western USA MLR 45 ≥ 1 Year R 2 = 0.45-0.64 Hill et al (2013) Conterminous USA RF ∼ 1000 1/site Season, year σ e = 1.1-2.0 • C Hrachowitz et al (2010) UK MLR 25 1 Month, year R 2 = 0.50-0.84 Imholt et al (2013) UK MLR 23 2 Month R 2 = 0.63-0.87 Isaak et al (2010) Western USA MLR, NKM 518 14 Month, year R 2 = 0.50-0.61, σ e = 2.5-2.8 • C Isaak and Hubert (2001) Western USA PA 26 1/site Season R 2 = 0.82 Johnson (1971) New Zealand ULR 6 1 Month n/a Johnson et al (2014) UK NLR 36 1.5 Day R 2 = 0.67-0.90, σ e = 1.0-2.4 • C Jones et al (2006) Eastern USA MLR 28 3 Year R 2 = 0.57-0.73 Kelleher et al (2012) Eastern USA MLR 47 2 Day, week n/a Macedo et al (2013) Brazil LMM 12 1.5 Day R 2 = 0.86 Mayer (2012) Western USA MLR 104 ≥ 2 Week, month R 2 = 0.72, σ e = 1.8 • C Miyake and Takeuchi (1951) Japan ULR 20 n/a Month n/a Moore et al (2013) Western Canada MLR 418 1/site Year σ e = 2.1 • C Nelitz et al (2007) Western Canada CRT 104 1/site Year n/a Nelson and Palmer (2007) Western USA MLR 16 3 Season R 2 = 0.36-0.88 Ozaki et al (2003) Japan ULR 5 8 Day n/a Pratt and Chang (2012) Western USA MLR, GWR 51 1/site Season R 2 = 0.48-078 Risley et al (2003) Western USA ANN 148 0.25 Hour, season σ e = 1.6-1.8 • C Rivers- Moore et al (2012) South Africa MLR 90 1/site Month, year R 2 = 0.14-0.50 …”

“…This approach was usually applied to average the land cover characteristics only, particularly forest coverage (e.g. Sponseller et al, 2001;Scott et al, 2002;Macedo et al, 2013;Segura et al, 2014), but also in some cases to average most of the predictor variables, including elevation or slope (Hrachowitz et al, 2010;Imholt et al, 2013). Other authors considered larger portions of the catchments as averaging areas, sometimes extending far beyond the riparian zone.…”

“…The range of methods encompasses traditional approaches such as multi-linear regression (e.g. Arscott et al, 2001;Mayer, 2012;Imholt et al, 2013) or linear mixed modelling (Macedo et al, 2013), but also more advanced techniques such as geographically weighted regression (Pratt and Chang, 2012;Chang and Psaris, 2013), networked kriging models (Gardner and Sullivan, 2004;Isaak et al, 2010;Ruesch et al, 2012) or machine learning techniques (e.g. Westenbroek et al, 2010;Hill et al, 2013;DeWeber and Wagner, 2014).…”

“…Hrachowitz et al (2010), Imholt et al (2013) and Rivers-Moore et al (2012) expressed water temperature as a linear combination of climatic and physiographic variables for each month of the year separately. Their models were derived for a particular year, but can be transferred to other years by estimating stream temperature at a few measurement points using Mohseni's logistic equation and fitting the multi-linear regression model to the resulting values (Hrachowitz et al, 2010).…”

Shield or not to Shield: Effects of Solar Radiation on Water Temperature Sensor Accuracy

“…Lack of knowledge about the potential for sunlight to affect Tw has led to inconsistency in sensor deployment [8][9][10][11][12][13][14][15]. For example, some researchers shade sensors with plastic tubes [8] whereas others do not [9], or fail to mention this detail [10].…”