Assessing algal blooms in Taihu Lake using hourly data: Seasonal
characteristics and the relationship with environmental factors

1、Introduction：

Algal bloom is one of the indicators of water eutrophication (Smetacek and Zingone, 2013; Cao and Han, 2021). It has become an important problem for many coastal areas and inland water in China (Zhang et al., 2020a, 2020b; Yang et al., 2023). In the fresh water, the algal blooms are usually caused by cyanobacteria outbreaks, which are often toxic (Aranda et al., 2023; MacKeigan et al., 2023). Cyanobacterial toxins pose a threat to human and animal health, which may even lead

to fatalities. Moreover, the outbreak and decay of cyanobacteria deplete oxygen in the water, ultimately causing harm to aquatic plants and animals (Paerl et al., 2019; Choi et al., 2021). Therefore, it is crucial to monitor and predict the cyanobacteria outbreaks for ensuring public health and lake ecology.
Large-scale algal blooms attack Taihu Lake in China in recent years, and field measurements have revealed that cyanobacteria were the dominant phytoplankton species in Taihu Lake (Wu et al., 2013; Wang et al., 2021; Huang et al., 2015). Traditional methods based on

Hydrographic stations and ships are insufficient for monitoring the spatial distribution and change of large-scale cyanobacterial blooms (Xue et al., 2019; Oyama et al., 2015). In contrast, satellite remote sensing has emerged as an effective and efficient method for monitoring such algal bloom outbreaks. (Gholizadeh et al., 2016; Zhang et al., 2020a, 2020b). For instance, the Floating Algal Index (FAI) method (Hu, 2009) was utilized to retrieve cyanobacteria in Taihu Lake based on Moderate-resolution Imaging Spectroradiometer (MODIS) data (Wu et al., 2015). MODIS data performed well in assessing the influence of meteorological factors on the phenology of cyanobacterial blooms (Zhang et al., 2012). Long-term MODIS data could be employed to examine the impact of nutrient enrichment and meteorological factors on the response of cyanobacteria (Shi et al., 2017). MODIS data with a temporal resolution of half day performs well in daily analysis of algal blooms, while the Geostationary Ocean Color Imager (GOCI) satellite with hourly temporal resolution can monitor the dynamic changes in cyanobacterial blooms within a few hours. (Huang et al.,2015; Jin et al., 2018). Given the disparities in spectral bands between the GOCI and MODIS satellites, an Alternative Floating Algal Index (AFAI) method was proposed specifically to detect cyanobacterial blooms using GOCI data (Qi et al., 2018; Jin et al., 2018). Field measurements and GOCI data were combined to analyze the hourly dynamic characteristics of cyanobacteria outbreaks in Taihu Lake (Li et al., 2022).
For a large shallow lake, such as Taihu Lake in China, wind is one of the key factors in producing visible surface cyanobacterial blooms (Wu et al., 2013; Qi et al., 2018). Previous studies demonstrated that a low wind speed of 3–4 m/s favors algae accumulation at the water surface, whereas higher wind speeds lead to their dispersion into the water column, impeding observations (Li et al., 2022; Qi et al., 2018). Ac- cording to the field data, when the wind speed exceeds 7 m/s, the algae becomes fully mixed with the water (Wu et al., 2015). Temperature is also an important factor contributing to cyanobacterial blooms (Ma et al., 2016). According to laboratory experiments, the majority of cyanobacteria grew well at a 25–35 OC (Singh and Singh, 2015). Certain species were capable of surviving at temperatures below 5 O C, or even above 45 O C in laboratory conditions (Reinl et al., 2023; Rossi et al., 2023).
In addition to wind and temperature, nutrients, including total ni- trogen (TN) and total phosphorus (TP), also play a vital role in the occurrence of algal blooms in Lake Taihu on an annual or monthly scale. In contrast, the effects of nutrients (including TN and TP) on algal bloom at the hourly scale are not significant (Li et al., 2023; Huang et al.,2015). The concentration of fluorescent dissolved organic matter (fDOM) is highly sensitive to hydrodynamic disturbances (such as wind-waves and current scouring) (Wang et al., 2022). Its spatiotemporal distribution characteristics can more directly reflect the environmental driving mechanisms at short timescales (Foroughan et al., 2022). Therefore, this study includes fDOM in analyzing the algae bloom in Taihu Lake.
The temporal resolution of those studies also focused on the daily, weekly, and monthly changes of the algae. However, the area of algal blooms in Lake Taihu, different from large macroalgae such as the green tides in the Chinese Yellow Sea (Jin et al.,2018) may vary significantly within a short period (on an hourly scale). The difference between the daily maximum and minimum areas can exceed 500 km2 (Cao and Han, 2021), and the hourly area change can over 150 km2 (Li et al., 2022). Using hourly data allows for real-time monitoring of the coverage area of cyanobacterial bloom, provides a more accurate reflection of their dynamic processes, and enables a more precise quantification of how environmental factors influence changes in bloom area.
Understanding the influence of environmental factors on the dy- namic characteristics of algae at a higher temporal resolution and over a longer time horizon is crucial for accurately predicting and managing algal bloom events. In this study, the meteorological and water quality factors ofTaihu Lake were observed continuously over a one-year period with a temporal resolution of one hour. GOCI data were used to moni- toring changes of the coverage area of algal blooms in Taihu Lake from

2020 to 2021 based on AFAI. The coverage area of algal blooms in Taihu Lake was assessed hourly using GOCI images and the associated impacting factors were evaluated with measured data, including fluo- rescent dissolved organic matter (fDOM), water temperature (Tw), hourly air temperature (Ta), hourly maximum air temperature (Tm), wind speed (V), and remote sensing products such as photosynthetically active radiation (PAR). A method was proposed to investigate the coverage area of algal blooms in Taihu Lake based on environmental factors. The maximum likelihood estimation method was adopted in this proposed method, and the results were determined using BIC methods.

2、Data and Methodology：

2.1. Study area
Taihu Lake between 30O 55,40"N–31O 32,58"N and 119O 52, 32"E–120O 36,10" E is the third largest freshwater in China (Fig. 1(a)). It is a large shallow lake with an area of about 2400 km2 and an average depth of 1.9 m (Zhang et al., 2012). The floating leaf vegetation in the eastern coastal area of Taihu Lake greatly interferes with the extraction of cyanobacterial blooms, potentially leading to inaccurate results and misleading conclusions (Zhu et al., 2018; Wu et al., 2015). Thus, in order to ensure the accuracy and reliability of our study on algal blooms, the eastern bays ofthe lake are not included in this research. The study area is as shown in Fig. 1.

2.2. Data resources and processing
To monitor the cyanobacterial blooms in Taihu Lake, the GOCI level 1B data is adopted, which can be obtained from the Korea Ocean Sat- ellite Center (https://kosc.kiost.ac.kr). The GOCI satellite has a spatial resolution of 500 m and a temporal resolution of 1 h (8 times/day, 0,00 UTC-7,00 UTC). All the calculation in this study is based on UTC time. For Taihu Lake in China, the local time is UTC + 8.
To ensure the precise registration between the locations of ground objects derived from satellite data and their real-world locations, the GOCI data is geometrically corrected. The geographic lookup table is made with the longitude and latitude information of the GOCI data which is also available on the website of Korea Ocean Satellite Center (https://kosc.kiost.ac.kr). The ground control points (GCPs) based on the 1:250,000 basic geographic information dataset (2021) provided by the National Catalogue Service for geometric accurate correction (https://www.webmap.cn/main.do?method=index) are used for Image-to-map Registration. The GLT geometric correction method and Image-to-map applications method are available in ENVI version 5.3 (Exelis Visual Information Solutions, Boulder, Colorado).
To reduce the atmospheric interference, the images are corrected using Rayleigh correction, which can remove the molecular (Rayleigh) scattering effects (Hu, 2009). The correction method is expressed as follows:

where L is the calibrated sensor radiance after adjustment for ozone and
other gaseous absorption, F0 is the extraterrestrial solar irradiance at data acquisition time, θ0 is the solar zenith angle, and R r is the Rayleigh reflectance estimated by Second Simulation of the Satellite Signal in the Solar Spectrum (6S) atmospheric correction. The 6S atmospheric correction model is available in: http://6s.ltdri.org/pages/downloads. html.
The hourly PAR data are obtained from the Himawari-8 L3 hourly PAR data. The data are downloaded from JAXA's P-Tree system, Japan Aerospace Exploration Agency (JAXA)(https://www.eorc.jaxa. jp/ptree). The near-real-time PAR data has a spatial resolution of 0.05O and a temporal resolution of 1 h.

The water quality data, including Tw and fDOM were measured by a water quality buoy with an EXO water quality monitoring platform. The water quality buoy located at (31◦ 08,24"N, 120◦ 10,48"E) in Taihu Lake (Fig. 2) measured Tw and fDOM every half an hour, from May 2019 to November 2021.
Meteorological data, including hourly air temperature (Ta), hourly maximum air temperature (Tm),wind speed (V) and maximum wind speed were recorded by three automatic meteorological stations (Fig. 2). The data was collected on an hourly basis at these stations, and the re- ported Ta, Tm, and V values for Taihu Lake represent the average of the measurements at these three stations. The “maximum wind speed” in this paper adheres to meteorological standards, specifically defined as the maximum value of 10-min running average wind speeds over a designated observation period.