Evaluation of Diffuse Reflectance Spectroscopy for Prediction of In Situ Phototrophic Biofilm Composition

Biofilm-based algae cultivation is a compelling alternative to suspended growth due to high productivity and a simplified harvesting process. However, process monitoring of attached growth systems presents a challenge due to destructive, labor-intensive, and slow conventional sampling and analysis. This study evaluated a potential approach to in situ monitoring: diffuse reflectance spectroscopy (DRS). Visible-near infrared spectra were used to develop partial least squares regression (PLSR) models for the prediction of fatty acid methyl esters (FAME), volatile solids (VS), ash, and carbohydrate content in phototrophic biofilms. Previous DRS algae studies typically used dried biomass and/or pure algal cultures, whereas this study collected spectral data directly from wet, in situ biofilms grown on a demonstration-scale revolving algal biofilm (RAB) system treating real wastewater. Despite the challenging conditions of mixed-species, wet, in situ biofilms, the PLSR model for FAME achieved moderate predictive performance (NRMSE: 10%; RPD: 2.3), demonstrating potential for resolving relative differences in lipid content under similar operating conditions. Models predicting VS, ash, and carbohydrates did not achieve predictive performance sufficient for practical application (NRMSE: 13–14%; RPD: 1.1). When models were calibrated and validated using seasonally disparate datasets (trained on summer and tested on fall, and vice versa) rather than the combined, mixed-season dataset, prediction accuracy declined sharply, highlighting the influence of seasonal variability and the need for larger, more diverse calibration datasets and periodic recalibration. These results suggest that DRS is a promising tool for non-destructive compositional monitoring of FAME in wet phototrophic biofilms, but additional model development and validation across broader operating conditions are needed before reliable long-term field deployment.