Sensing of free-surface patterns in opaque flows using Kinect

The field of remote sensing is important in industrial and academic applications as it can present a low-cost and low maintenance option for observation of natural or artificial processes. Recent studies have shown that the dynamic pattern on flowing water can used to infer the underlying flow conditions, and that the more detailed the measurement, the more accurately the flow conditions can be determined. Most current methods of free-surface detection are intrusive to the flow and/or generally measure in one dimension (wave probe) or two dimensions (laser induced fluorescence). This paper assesses the viability of 3D free-surface sensing techniques using a Microsoft Kinect sensor to measure surface patterns of amplitude below 15 mm. The data is validated against a traditional wave probe device. Results show that whilst the signals do not match precisely, the general wave height and dominant length scales appear similar. Detailed statistical analysis shows that the Kinect device is able to detect the average water depth with a maximum error of 3.9 % The rootmean-square wave height is consistently under-reported by the Kinect sensor, by approximately 50 %, but the Kinect and the wave probe data both show similar frequency spectra, with similar dominant component, spectral gradient, and increasing trend with flow rate.