Predicting grain size in gravel-bedded rivers using digital elevation models: Application to three Maine watersheds

Abstract

Riverbed grain size controls suitability of spawning habitat for threatened fish species. Motivated by this relationship, we developed a model that uses digital elevation models (DEMs) to predict bed grain size. We tested the accuracy of our model and two existing models with channel measurements from high-resolution airborne light detection and ranging (LiDAR) DEMs. All three models assume that bed grain size is a function of reach-average high-flow channel hydraulics (measured by shear stress or stream power). Our test data are field measurements of median grain size ( D 50 ) at 276 stations along four rivers in Maine. Pleistocene continental glaciation strongly influences the longitudinal profiles, which have alternating steep and gradual segments. We exploit the resulting variations in sediment supply to understand the controls on model success or failure in predicting bed grain size. Results show that all three models have ∼70% success in predicting D 50 within a factor of two overall, and better where the rivers are coarse gravel bedded (∼80% success where D 50 ≥ 16 mm). This similarity is unsurprising given that the models primarily rely on channel gradient (S) and drainage area as inputs. Measurements of S from LiDAR DEMs yield only a modest improvement in model success over those from topographic maps. We find that our model works best in sediment-starved steep reaches. Model failures fall into two broad categories: (1) relatively fine-grained ( D 50 D 50 . We argue that models based on airborne infrared LiDAR DEMs may reach a maximum around 80%–85% accuracy due to these sub-reach-scale factors, which cannot be easily measured from DEMs. The overall success of the models in predicting grain size indicates that the morphology of these channels has adjusted to the imposed S and sediment load during the ∼15 k.y. since deglaciation and through the period of anthropogenic channel change over the past three centuries.