ALBERT

Description: ABSTRACT We investigated which landscape and climate-related data (including information on hydrological source of flow) were statistically significant predictors of channel wetted width (WW) across a sizeable (2200 km 2 ) region of the UK. This was conducted specifically when flow was less than mean daily flow (MDF) and where channels are in a near natural state. Orthorectified air photos at 25 cm spatial resolution were used to measure WW, with the magnitude of the errors in these measurements quantified. We used flow information from local gauging stations to ensure that channels were below MDF for the days on which the air photos were captured. The root mean squared difference between the field and air photo measurements of WW ( n  = 28sites) was small (0.14 m) in comparison to median WW (3.07 m). We created points along sections of channels visible in air photos and used a terrain model to create drainage catchments for these points and computed their catchment area (CA). We selected a subset of points ( n  = 472) and measured their WW from air photos, and computed landscape-related data for each of their catchments (mean slope, mean annual rainfall, land cover type, elevation) and also mean BFIHOST, a quantitative index relating to hydrological source of flow. We used a linear mixed model to predict WW by including the landscape data (including CA 0.5 ) as fixed effects, plus a spatial covariance function estimated by residual maximum likelihood (REML) to determine unbiased estimates of the predictors. There was no evidence for retaining the spatial covariance function. With the exception of land cover, all the predictors were statistically significant and accounted for 76% of the variance of WW. When CA 0.5 alone was used as a predictor it captured 54% of the variance. The vast majority of this difference was due to inclusion of an interaction between CA and hydrological source of flow (BFIHOST). As catchment area increases, those channels with larger mean catchment BFIHOST values (greater proportion of baseflow contribution) have narrower WW by comparison to those with smaller mean BFIHOST for the same CA. Improved predictions of channel WW (based on our findings) could be used in channel restoration. This article is protected by copyright. All rights reserved.