Determining Mean Grain-Size in High Gradient Streams With Autocorrelative Digital Image Processing

Accurate surface roughness characterization of a steam bed is essential to hydrodynamic and sediment transport modeling. High gradient steams are particularly challenging in part because small changes in the mean grain size can have a demonstrative impact on its fluid dynamics. Traditionally employed methods of determining mean grain size are labor intensive, provide low temporal resolution, and are historically inaccurate. We present an automated digital image processing method for high gradient streams that uses spatial autocorrelation to determine mean grain size. Numerical results are compared to statistics obtained from sieve-based protocols. Results support this method as a potential alternative to traditional methods with a dramatic increase in temporal resolution. When optimized parameters are applied the system accurately determines mean grain size within a 50% difference from sieved samples 85% of the time.