High-resolution grain size analysis using photos
Summary
Grain size is an important factor that influences sediment transport at beaches. Traditional methods of obtaining grain size properties, such as sieving or settling tube, are time-consuming and labour-intensive, which limits the ability to map large sections of a beach in the spatial and temporal domains. This leads to significant simplification of the spatio-temporal variability of the beach sediment, whereby an entire
section of the beach is defined by a single property.
In this study, grain size distributions are analysed using pyDGS. This method involves capturing images of sediment samples, significantly reducing fieldwork while enabling higher-resolution sampling. This study addresses the main challenge when using various grain size analysis methods, where the areal (pyDGS) and volumetric (sieve) measures are not directly comparable. To effectively utilise pyDGS as an alternative to mechanical sieving, we introduce a novel correction method. In contrast to the traditional correction method, which relies on correction exponents obtained through additional sampling methods, our method utilises information extracted from the images. It involves a three-step process for all grain size classes within an image: 1) calculate the total number of grains in the photo frame, 2) estimate the volume of each grain class, and 3) compute the mass of each grain size category by multiplying the number of grains by their volume and density. In this study, the grain shape was estimated and calibration was done with sediment samples that were photographed and sieved. In total, we analyzed 380 sediment photographs, with 43 sediment samples also sieved for direct comparison. The results of our novel correction method were compared to those of the traditional correction method. Our study reveals that the errors associated with D90, D84, D50, and D25 are similar. However, for D16 and D10, our novel method achieved a 4% reduction in RMSE and a 7% reduction in MAE for
D16, and a 24% reduction in RMSE and a 19% reduction in MAE for D10. During the most recent field campaign, we photographed a total of 24 cross-shore rows. This extensive dataset demonstrates the potential of our approach, which would be unfeasible with traditional grain size analysis methods.
The main contribution of this research is the introduction of a novel correction approach, enabling the conversion of area-based measurements into volume-by-weight measurements. This approach eliminates the need for generalisation and offers a more precise correction method for individual images. Making it particularly valuable for larger beach sections. Despite challenges related to grain shape uncertainties and sediment variability, the new method provides a more effective way of correcting the images for improved accuracy.
While pyDGS does exhibit significant errors that prevent it from providing absolute values, it cannot fully replace mechanical sieving. However, its high spatial resolution capabilities offer a valuable tool for studying the temporal evolution of grain size at beaches. This method provides insights into relative patterns in grain size evolution, enhancing our understanding of beach sediment dynamics and contributing to more effective coastal management strategies.