Journal cover Journal topic
Hydrology and Earth System Sciences An interactive open-access journal of the European Geosciences Union
Hydrol. Earth Syst. Sci., 21, 4727-4745, 2017
© Author(s) 2017. This work is distributed under
the Creative Commons Attribution 3.0 License.
Research article
21 Sep 2017
Can spatial statistical river temperature models be transferred between catchments?
Faye L. Jackson1,2, Robert J. Fryer3, David M. Hannah1, and Iain A. Malcolm2 1School of Geography, Earth and Environmental Science, University of Birmingham, Birmingham, B15 2TT, UK
2Marine Scotland Science, Scottish Government, Freshwater Laboratory, Faskally, Pitlochry, PH16 5LB, UK
3Marine Scotland, Marine Laboratory, 375 Victoria Road, Aberdeen, AB11 9DB, UK
Abstract. There has been increasing use of spatial statistical models to understand and predict river temperature (Tw) from landscape covariates. However, it is not financially or logistically feasible to monitor all rivers and the transferability of such models has not been explored. This paper uses Tw data from four river catchments collected in August 2015 to assess how well spatial regression models predict the maximum 7-day rolling mean of daily maximum Tw (Twmax) within and between catchments. Models were fitted for each catchment separately using (1) landscape covariates only (LS models) and (2) landscape covariates and an air temperature (Ta) metric (LS_Ta models). All the LS models included upstream catchment area and three included a river network smoother (RNS) that accounted for unexplained spatial structure. The LS models transferred reasonably to other catchments, at least when predicting relative levels of Twmax. However, the predictions were biased when mean Twmax differed between catchments. The RNS was needed to characterise and predict finer-scale spatially correlated variation. Because the RNS was unique to each catchment and thus non-transferable, predictions were better within catchments than between catchments. A single model fitted to all catchments found no interactions between the landscape covariates and catchment, suggesting that the landscape relationships were transferable. The LS_Ta models transferred less well, with particularly poor performance when the relationship with the Ta metric was physically implausible or required extrapolation outside the range of the data. A single model fitted to all catchments found catchment-specific relationships between Twmax and the Ta metric, indicating that the Ta metric was not transferable. These findings improve our understanding of the transferability of spatial statistical river temperature models and provide a foundation for developing new approaches for predicting Tw at unmonitored locations across multiple catchments and larger spatial scales.

Citation: Jackson, F. L., Fryer, R. J., Hannah, D. M., and Malcolm, I. A.: Can spatial statistical river temperature models be transferred between catchments?, Hydrol. Earth Syst. Sci., 21, 4727-4745,, 2017.
Publications Copernicus
Short summary
River temperature (Tw) is important to fish populations, but one cannot monitor everywhere. Thus, models are used to predict Tw, sometimes in rivers with no data. To date, the accuracy of these predictions has not been determined. We found that models including landscape predictors (e.g. altitude, tree cover) could describe spatial patterns in Tw in other rivers better than those including air temperature. Such findings are critical for developing Tw models that have management application.
River temperature (Tw) is important to fish populations, but one cannot monitor everywhere....