Cao, M.; Qing, S.; Du, Y.; Yuan, R.; Shun, B.; Hao, Y., and Zhao, W., 2020. Remote sensing classification of aquatic vegetation in Ulansuhai Lake based on discrete particle swarm optimization ...algorithm. In: Jung, H.-S.; Lee, S.; Ryu, J.-H., and Cui, T. (eds.), Advances in Geospatial Research of Coastal Environments. Journal of Coastal Research, Special Issue No. 102, pp. 176-186. Coconut Creek (Florida), ISSN 0749-0208. As the largest natural wetland in the same latitude on the earth, water ecological environment of Ulansuhai Lake, China is seriously threatened, and there is various aquatic vegetation spread over the lake. Remote sensing is considered to be an effective way to map the distribution of aquatic vegetation. Different from the method used in most of the previous studies, discrete particle swarm optimization (DPSO) algorithm was used to identify and classify emergent vegetation (EV), yellow algae (YA), submerged aquatic vegetation (SAV) and water in Ulansuhai Lake based on Landsat-8 Operational Land Imager (OLI) in this research. The classification results were validated by 284 investigation sites data and visual interpretation of Gao Fen 2 (GF-2) image. The results indicated that determination coefficient (R2) for EV, YA, SAV and water were greater than 0.91, root mean square error (RMSE) were less than 0.025 km2. Besides, the overall accuracy (95.4 %) and Kappa coefficient (0.93) of DPSO algorithm are superior to spectral index, unsupervised classification methods and supervised classification methods. In addition, DPSO algorithm to other regions (the Yellow Sea) and sensors (Sentinel-2) have been successfully applied, which further proves the applicability of DPSO algorithm. The research provides a new tool to assist people in locating and quantifying aquatic vegetation, so that purposeful actions can be taken to control the eutrophication of lake water and improve the water ecological environment.
•The spectral characteristics of aquatic vegetations and yellow algae were analyzed.•Several spectral indices were proposed to distinguish aquatic vegetations and yellow algae.•Seasonal and inter ...annual dynamics of aquatic vegetations and yellow algae were investigated.•The responses of aquatic vegetations and yellow algae to air temperature were discussed.•The MAI concept is considered to be extendable and applicable.
Global lakes have suffered from algae blooms and loss or expansion of aquatic vegetations in recent decades. Remote sensing is considered as an effective approach to monitor aquatic vegetations and algae blooms. However, individual spectral index is unable to separate them due to the similarity in spectral features. In this paper, spectral characteristics analyses were conducted to find suitable spectral indices for distinguishing emergent vegetation, submerged aquatic vegetation and floating yellow algae in a complex aquatic environment, Ulansuhai Lake, China. It was found that near infrared band was appropriate to extract open water, and short-wave infrared band was suitable for extracting emergent vegetation, whereas the green and red bands were the characteristic spectra for distinguishing submerged aquatic vegetation and yellow algae. Hence, we firstly used the normalized difference vegetation index (NDVI) to extract open water, and the emergent vegetation spectral index (EVSI) to extract emergent vegetation. Then, a new developed macroalgae index (MAI) was used for distinguishing submerged aquatic vegetation and yellow algae. The applicability of the spectral indices was tested against both in situ measurements and Landsat-8 Operational Land Imager (OLI) data. The results indicated that the combination of these spectral indices was an effective method to separate aquatic vegetations and yellow algae. The proposed method was then applied to time-series Landsat images for investigating the seasonal and inter annual dynamics of aquatic vegetations and yellow algae and their responses to air temperature in the Ulansuhai Lake. The results show that emergent vegetation area increased from May to its maximum in July. The submerged aquatic vegetation area gradually increased from May to its maximum coverage in August, and decreased from late September. The yellow algae appeared in late May, and reached its maximum area in June or July, and disappeared in October. The long-term variation analyses showed that emergent vegetation area increased from 1986 to 2014, and was decreasing from 2014. The area of submerged aquatic vegetation increased during 1986–2008, and sharply decreased from 2009 to 2013, followed by a significant increasing from 2014. The yellow algae bloom mainly outbroke during the period of 1998 to 2010. We also found that the yellow algae area was more sensitive to short term mean temperature, while the area of emergent vegetation was sensitive to longer timescale of temperature. The submerged aquatic vegetation area had no significant correlation with air temperature. Besides, our study also indicated that the MAI concept can be extendable and applicable to other high-resolution satellite sensors (e.g., GF-2 PMS) and other regions with different algae blooms (e.g., Yellow Sea).
PDF
