Study on the Spectral Characteristics of Black-Odorous Water: A Case Study of Shenzhen City, China

摘要

The occurrence of black-odorous water (BOW) in urban waters is a common problem in China that has occasional, seasonal, and perennial rules. The treatment of urban has become an important task for water managers in China. Based on field BOW spectra and 13 types of water quality parameters measured in Shenzhen City, in this study, we processed three types of spectra, including normalized, difference, and ratio spectra, to reveal the correlation between the spectra and the water quality parameters, especially for the spectra and four indicators that are used to distinguish BOW. The four indicators include Secchi depth (SD), dissolved oxygen (DO), oxidation-reduction potential (ORP), and ammonia nitrogen (NH3-N). Furthermore, we analysed the spectral sensitivity of the different water quality parameters to build a quantitative regression model between the water quality parameters and the portrait spectrum. The results showed the following: (1) Of the four indicators, both SD and NH3-N were significantly related to the spectrum of mild and severe BOW, and the ORP was only significantly related to the spectrum of severe BOW. The correlation between DO and BOW was not significant. (2) For mild BOW, the sensitive bands were 404 to 572 nm and 704 to 864 nm for the SD and 705 to 717 nm for NH3-N. For severe BOW, they were 499 to 606 nm and 730 to 900 nm for the SD, 556 to 664 nm for NH3-N, and 642 to 709 nm for the ORP. (3) The SD had the strongest correlation with the BOW spectrum. The correlation between the SD and the BOW spectrum was symmetrically distributed in the range of 404 to 572 nm near 489 nm, whereas 545 nm appeared as the centre in the range of 499 to 606 nm. (4) The best regression relationship between the SD and BOW was R(489)/R(714) (coefficient of determination R (2) = 0.8806) and R(545) (R (2) = 0.4868) for the mild and severe grades, respectively. The best regression relationship was R(489)-R(607) (R (2) = 0.7598) between NH3-N and mild BOW. The results of this study provide important guidance significance and application value regarding the selection of bands, model construction, and result verification for urban BOW using remote sensing.