Seasonal summertime upwelling east of Zhoushan Island has been known for centuries, particularly because this area is a famously rich fishing ground, one of the best in the China Seas (Figure 1
a,c). Local fishermen undoubtedly noticed that in summer this area is appreciably colder (by up to 3-to-4 °C) compared to ambient waters (Figure 1
c). The traditional importance of this area for marine fisheries is currently amplified by marine aquaculture and marine ranching. These factors and the proximity to major research centers stimulated numerous studies of this region from in situ and lately from satellite data that are briefly reviewed below.
Since upwellings usually feature low temperatures relative to ambient waters, most remote sensing studies of upwellings utilize sea surface temperature (SST) data. For decades, the bulk of SST data has been provided by the advanced very high-resolution radiometers (AVHRR) flown on polar-orbiting National Oceanic and Atmospheric Administration (NOAA) satellites. The AVHRR data from multiple NOAA satellites have been thoroughly calibrated and inter-calibrated under the Pathfinder project. Since the early 2000s, SST data are also provided by the moderate resolution imaging spectroradiometers (MODIS) flown on the MODIS-Terra and MODIS-Aqua satellites operated by National Aeronautics and Space Administration (NASA). The visible infrared imaging radiometer suite (VIIRS) radiometer aboard the Suomi National Polar-orbiting Partnership (Suomi-NPP) satellite launched in 2011 and operated by NASA provides high-quality SST data. The above-mentioned instruments provide SST data with spatial resolution down to 1 km, which is sufficient for studies of meso-scale phenomena, including most upwellings. The temporal resolution of infrared SST radiometers is compromised by cloudiness that results in data series with numerous gaps. Various techniques have been proposed to alleviate this major shortcoming of infrared SST data. One approach is to merge infrared SST data with microwave SST data because microwave radiometers can see through clouds. Unfortunately, the spatial resolution of microwave SST data is rather poor (a few tens of kilometers). Another approach involves compositing satellite images. The basic idea behind this approach is that cloud patches in individual images could be eliminated by artificial intelligence (AI) algorithms. Such pattern recognition algorithms recognize clouds as patches with different temperatures that move relatively fast over the ocean.
Geostationary satellites, owing to their stable positions over fixed points on Earth, allow their payload instruments to carry out extremely high-frequency sampling of the sea surface. The sampling frequency is limited only by the scanning speed of their imagers. For example, instruments aboard the Japanese geostationary satellites Himawari (“Sunflower”) scan the Earth’s disc every 10 min. Smart AI algorithms for compositing frequent images of sea surface could effectively filter out most clouds, thereby providing high-frequency time series of satellite imagery to study short-term variability of oceanic phenomena, including upwellings.
Lü et al. [1
], in their modeling study, inspected a few quasi-synoptical (5-day) composite AVHRR SST images and climatological (15-year) SST maps (based largely on satellite data) to note “the stable presence of a cold-water mass near Zhoushan Islands” (ibid., p. 6), which was up to 4 °C cooler than ambient water. The latter claim is not fully supported by their own data. Indeed, the three quasi-synoptic SST maps from July–August 1997–1998 reveal the absolute SST minimum of 25 °C vs. ambient SST of 28 °C (ibid., Figure 5), yielding the SST difference of 3 °C; while in the climatological SST maps (ibid., Figure 6) the same difference is <2 °C. Lü et al. [1
] noted: “The cold water is enclosed by relatively warm water, so it should not be formed by the horizontal advection. It is most likely that the water upwells from the lower layer” (ibid., p. 6). In a follow-up modeling study, Lü et al. [2
] simulated summer upwelling off the Zhoushan Islands and concluded that “in the study area about 60% of the upwelling is attributed to tidal effects, which should be the primary inducement for upwelling” (ibid., p. 472).
Hu and Zhao [3
] studied summer upwelling off northeastern Zhejiang from diverse datasets (AVHRR SST, 1987–2005; SeaWiFS Chl-a, 2002–2006; QuikSCAT wind, 1992–2006) and documented long-term variations (LTV) of the upwelling on the seasonal to interannual time scales. In a follow-up study based on the same datasets, Hu and Zhao [4
] studied short-term variations (STV) of upwelling off the Zhejiang coast in May–October 2004 and concluded that the upwelling variations are closely associated with the along-shore SW wind variations. Hu and Zhao [4
] identified two upwelling areas: one east of the Zhoushan Island, with the upwelling core near 30°N, another off the Yangtze River estuary, with the upwelling core at 31.5°N. The location of the Zhoushan upwelling core (30°N) differs from that determined by other authors.
Lou et al. [5
] and Lou et al. [6
] reported summer upwelling off Zhejiang from MODIS SST data, 2007–2009. They noted that the upwelling appears in June, which is consistent with the results by Hu and Zhao [4
] from 2004, yet inconsistent with the latest results, including this study. Lou et al. [5
] and Lou et al. [6
] pointed out that the Zhejiang upwelling vanishes in September, in accordance with Hu and Zhao [4
] and later studies, including this paper.
Liu and Gan [7
] investigated dynamics of intensified upwelling caused by a topographic promontory (“coastal headland”) east of the Zhoushan Islands at ~30°N, ~122.4°E. In the observational part of this paper, Liu and Gan [7
] presented maps of SST from June–August 2000–2009 and from August 2009. Both maps feature the coldest SST spot at ~30.7°N, ~122.5°E (ibid., Figure 2b,c), centered around the Shengsi Islands (part of the Zhoushan Archipelago). Both maps reveal that the upwelling SST intensity (defined as the core SST minus ambient SST) is ~2 °C or less.
Huang et al. [8
] analyzed all-weather (cloud-free) monthly mean data derived from the Multiscale Ultrahigh Resolution (MUR) SST dataset and concluded that the Zhoushan upwelling emerges in April to June and vanishes in September. They also ran the Regional Ocean Modeling System (ROMS) to investigate the wind influence on the upwelling. The simulation revealed that the wind is essential in inducing the Zhoushan upwelling.
Xu et al. [9
] investigated interannual variations of the Zhoushan upwelling using MUR 1 km resolution SST data from 2003–2014 and Pathfinder 0.05° × 0.05° resolution SST data from 1985–2002. The upwelling occurred from April through September with about the same strength over 30 years (1985–2014). The coldest SST spot was usually located around the Shengsi Islands, while the SST upwelling intensity (from SST climatology) was ~2 °C (ibid., Figure 3).
Yang et al. [10
] proposed a dimensionless upwelling intensity index (NUI) by combining the area of upwelling and the intensity of cold center. Unfortunately, as all indices, the NUI cannot be interpreted in terms of observables (directly measured quantities).
The above concise review of (mostly) observational studies reveals that our present knowledge of the Zhoushan upwelling is unsatisfactory and has focused on seasonal variations. Variations on other timescales, especially the short-term variability of the Zhoushan upwelling remain virtually unexplored.
In this study, the Level 3 daily and hourly SST data from Himawari-8 (JAXA-SST, Figure 1
c) were used to explore statistical features, seasonal variations, and short-term variations of the Zhoushan upwelling. The Himawari-8 is a new-generation Japanese geostationary meteorological satellite, whose instrument, the Advanced Himawari Imager (AHI) has a sampling interval of 10 min and a spatial resolution of 2 km. Its high temporal resolution allows AHI to capture short-term (hourly and daily) variations in surface manifestations of upwelling. To quantify surface features of upwelling, an SST-gradient-based upwelling detection algorithm which was successfully applied in a study of upwelling off northeastern Taiwan [11
] was used to demarcate the Zhoushan upwelling. An upwelling intensity index was developed to quantify the Zhoushan upwelling’s surface manifestation and its short-term variations. Such statistical results have never been reported before. The structure of this paper is as follows. Section 2
describes the data and methods used in this study. Section 3
and Section 4
present results and discussion. The main results are summed up in Section 5
The only criterion for judging the existence of upwelling in the detection algorithm is whether there is an upwelling center in the center-searching box. The upwelling center is defined as the coldest temperature minimum point in the center-searching box. The existence of the temperature minimum point depends on the SST distribution within and around the center-searching box, especially the CDW located north of the box and the shallower nearshore water mass in the west. If the water temperature outside the box is less than that inside the box, the temperature decreases from the inside to the outside, then there is no temperature minimum point in the box (such as the monthly SST image in October in Figure 4
), the algorithm in this paper will automatically determine that there is no upwelling phenomenon in the satellite image at this moment. This usually occurs in autumn and winter, because when the net heat flux at the sea surface is negative, the seawater on the west side of the box cools down faster due to the shallow topography, while the water mass on the north side comes from colder river water. This is the reason why the algorithm used in this study cannot find the upwelling from the SST satellite images from October to March of the following year, although the upwelling exists all year round. It is the main inability of the algorithm, and one of the directions of future work.
Previous studies pointed out that the upwelling appears in June and vanishes in September [4
], while the result in this study shows the upwelling occurs as early as April (Figure 5
), which is consistent with the results by Xu et al. [9
]. Although He et al. [25
] noted that the wind-induced upwelling in July is the strongest, which is consistent with the result by Hu and Zhao [3
], and the temperature difference (near 1.5 °C) in August is larger than in July (near 1 °C) (ibid., Figure 1); the result of UPI
monthly variations in this study also confirmed this (Figure 5
). In addition to this, the results of this study also show the location of the upwelling center has a significant monthly migration, which has not been reported in previous studies.
Previous studies have different views on the location of the upwelling center. The SST maps of Liu and Gan [7
] feature the coldest SST spot at ~30.7°N, ~122.5°E (ibid., Figure 2b,c), centered around the Shengsi Islands (part of the Zhoushan Archipelago). Hu and Zhao [4
] pointed out that there have two upwelling centers: one is near 30°N and another is at 31.5°N. Xu et al. [9
] noted that the coldest SST spot was usually located around the Shengsi Islands. Different from the previous statement of single-center or double-center upwelling, the statistical result of UPF show that Zhoushan upwelling has multiple potential upwelling points and can be divided into four clusters: located between Gouqi Island and Lvhua Island, in the southeast sea off Shengsi Island (same to the result of Liu and Gan [7
] and Xu et al. [9
]), around the Zhongjieshan Islands, and off the east coast of the Taohua-Liuheng Islands. This new finding shows that the upwelling around Zhoushan is a multi-center upwelling system, which was combined by multiple isolated upwelling. The upwelling core area in this study is located at 122°E–123°E, 29.5°N–31.15°N with an irregular ellipse extending from SW to NE (Figure 7
b), which is consistent with the results by Hu and Zhao [4
] and He et al. [25
], and smaller than Lou et al. [5
]. In addition to this, it is found that the direction of the major axis of the core area is basically consistent with the direction of the 30 m isobath while the core area distribution is confined to the water depths between 10 m and 50 m. This finding confirms a view reported in the previous studies that topography is a key factor in the formation of Zhoushan upwelling [7
Due to limitations in the temporal resolution and coverage of the observed data, the short-term variations (daily and hourly) of the Zhoushan upwelling are occasionally mentioned. By analyzing three years’ data of temperature differences from surrounding non-upwelling waters, Lou et al. [5
] pointed out that the upwelling has short-time (several-day) variations (ibid., Figure 7). After checking the AVHRR SST images between 18 July and 9 August 2004 (ibid., Figure 3), Hu and Zhao [4
] concluded that the Zhoushan upwelling has a short-term daily variation, and is related to the variation of southwest alongshore wind. Compared with previous results (daily), the results in this study finely characterize the quasi-24 h periodic process of surface upwelling on a smaller timescale (hourly). This finding is beyond expectation and has never been reported before. A preliminary discussion suggests that in addition to tidal mixing, changes in the intensity of upper ocean stratification due to diurnal variations in solar radiation are also playing an important role in the quasi-24 h periodic variation of surface upwelling. These new findings and mechanisms may contribute to future theoretical and numerical studies.
The Level 3 daily and hourly SST data from Himawari-8 AHI were used to quantitatively document statistical features, seasonal variations, and short-term variations of the Zhoushan upwelling. To quantify the features of surface upwelling, an SST-gradient-based upwelling detection algorithm, which has been successfully applied in a quantitative study of upwelling in northeastern Taiwan, was adapted for this study. The results once again demonstrate the usability of the algorithm and, therefore, can be extended to other regions to detect surface upwelling.
Similar to previous studies, our results show that the Zhoushan upwelling appears in April, usually peaks in August, and rapidly disappears in September. It was also found that the location of the upwelling center also has a significant monthly migration: it appeared first in the coastal area with steep terrain near the northern Daishan Island in April, then moved eastward to the northern sea of Zhongjieshan Islands in May; finally it moved to the northeast of Shengsi Island, in between Gouqi Island and Lvhua Island in June, and remained quasi-stationary until the upwelling disappeared in October.
The statistical results show that potential upwelling points are clustered in places with large topographic gradients and can be divided into four aggregation areas: (1) between Gouqi Island and Lvhua Island, (2) SE of Shengsi Island, (3) around the Zhongjieshan Islands, and (4) east of the Taohua-Liuheng Islands. The core area of Zhoushan upwelling runs from SW to NE between 10 m and 50 m isobaths. These findings have never been reported before and may provide observational support for site selection of potential marine ranching and artificial upwelling.
The unexpected short-term variations of upwelling in the actual observations were captured by Himawari-8. The lifecycle was shown to be quasi-24 h and can be divided into two stages: intensification and decay. In addition, an hourly migration of the upwelling center under the advection of tidal current was also found. A preliminary analysis suggests that the quasi-24 h periodic variations may be caused by the competing effect between tidal mixing and the stratification in the water column. The quantitative analysis of the detailed features in this process will greatly improve our observational understanding of the details in the short-term variation and provide an empirical basis for future theoretical and numerical studies.