A Triangular Fuzzy Number-Based Water Quality Assessment Model for Evaluating the Impacts of Floating Photovoltaic Projects on Reservoir Water Quality Under Uncertainty

This study developed a Photovoltaic–Reservoir Water Quality Impact Model (PVRWQIM) based on triangular fuzzy number theory (TFN) to address data sparsity and measurement uncertainty in conventional water quality assessment. The model consists of two components: a TFN-based exceedance-risk module for quantifying the likelihood of water quality parameters exceeding predefined risk thresholds, and a transitional TFN module for evaluating changes in exceedance risk before and after floating photovoltaic (FPV) construction. The model was applied to Junshan Reservoir in Jiangxi Province, China, using 15 observations each from before and after FPV construction. The results indicate the following: (i) Before construction, the exceedance probabilities for water temperature (T), dissolved oxygen (DO), permanganate index (CODMn), total phosphorus (TP), total nitrogen (TN), chlorophyll a (Chla), and cyanobacterial density (CD) were 16.7%, 16.1%, 31.3%, 33.7%, 33.3%, 32.7%, and 31.3%, respectively. After construction, the exceedance probabilities for T, DO, CODMn, TP, and TN increased to 93.1%, 89.7%, 82.5%, 83.5%, and 83.3%, respectively (p < 0.01), while the probabilities of exceedance for Chla and CD decreased to 14.8% and 14.2%, respectively (p < 0.05). (ii) T and DO are the primary threat factors triggered by FPV construction, with a probability of increased ecological risk of approximately 90%. (iii) The model reveals a statistical correlation pattern: the probability of reduced ecological risk for CD and Chla is 62%, while the probability of increased ecological risk for CODMn, TP, and TN is 77%. The two sets of indicators exhibit opposite trends, suggesting that FPV shading may directly suppress algae growth while simultaneously weakening self-purification capacity indirectly through cooling and reduced water flow, thereby contributing to nutrient accumulation. It should be noted that, based on limited observational data, this study reveals correlations rather than proven causal mechanisms; the aforementioned causal interpretations require further validation through controlled experiments or mechanistic models in the future. (iv) The proposed PVRWQIM provides a practical tool for quantifying reservoir water quality risks under sparse data and measurement uncertainty and can support environmental assessment of similar FPV projects.