Spectral characteristics of CDOM (Chromophoric dissolved organic matter) in water columns are a key parameter for bio-optical modeling. Knowledge of CDOM optical properties and spatial discrepancy based on the relationship between water quality and spectral parameters in the Yinma River watershed with in situ data collected from highly polluted waters are exhibited in this study. Based on the comprehensive index method, the riverine waters showed serious contamination; especially the chemical oxygen demand (COD), iron (Fe), manganese (Mn), mercury (Hg) and dissolved oxygen (DO) were out of range of the contamination warning. Dissolved organic carbon (DOC) and total suspended matter (TSM) with prominent non-homogenizing were significantly high in the riverine waters, but chlorophyll-a (Chl-a) was the opposite. The ternary phase diagram showed that non-algal particle absorption played an important role in total non-water light absorption (>50%) in most sampling locations, and mean contributions of CDOM were 13% and 22% in the summer and autumn, respectively. The analysis of the ratio of absorption at 250–365 nm (E250:365) and the spectral slope (S275–295) indicated that CDOM had higher aromaticity and molecular weight in autumn than in summer, which is consistent with the results of water quality and the CDOM relative contribution rate. Redundancy analysis (RDA) indicated that the environmental variables OSM (Organic suspended matter) had a strong correlation with CDOM absorption, followed by heavy metals, e.g., Mn, Hg and Cr6+. However, for the specific UV absorbance (SUVA254), the seasonal values showed opposite results compared with the reported literature. The potential reasons were that more UDOM (uncolored dissolved organic matter) from human sources (wastewater effluent) existed in the waters. Terrigenous inputs simultaneously are in relation to the aCDOM(440)-DOC relationship with the correlation coefficient of 0.90 in the summer (two-tailed, p < 0.01), and 0.58 in the autumn (two-tailed, p < 0.05). Spatial distribution of the CDOM parameters exhibited that the downstream regions focused on dry land have high CDOM molecular weight and aromatic hydrocarbon. Partial sampling locations around the cities or countries generally showed abnormal values due to terrigenous inputs. As a bio-optical model parameter, the spectral characteristic of CDOM is helpful in adjusting the derived algorithms in highly polluted environments. The study on organic carbon and pollutants in highly polluted waters had an important contribution to global carbon balance estimation and water environment protection.