This paper develops a deep learning classification method with fully-connected 8-layers characteristics to classification of coastal wetland based on CHRIS hyperspectral image. The method combined spectral feature and multi-spatial texture feature information has been applied in the Huanghe (Yellow) River Estuary coastal wetland. The results show that: (1) Based on testing samples, the DCNN model combined spectral feature and texture feature after K-L transformation appear high classification accuracy, which is up to 99%. (2) The accuracy by using spectral feature with all the texture feature is lower than that using spectral only and combing spectral and texture feature after K-L transformation. The DCNN classification accuracy using spectral feature and texture feature after K-L transformation was up to 99.38%, and the outperformed that of all the texture feature by 4.15%. (3) The classification accuracy of the DCNN method achieves better performance than other methods based on the whole validation image, with an overall accuracy of 84.64% and the Kappa coefficient of 0.80. (4) The developed DCNN model classification algorithm ensured the accuracy of all types is more balanced, and it also greatly improved the accuracy of tidal flat and farmland, while kept the classification accuracy of main types almost invariant compared to the shallow algorithms. The classification accuracy of tidal flat and farmland is up to 79.26% and 56.72% respectively based on the DCNN model. And it improves by about 2.51% and 10.6% compared with that of the other shallow classification methods.

To analyze the grain size and depositional environment of the foreshore sediments, a study was undertaken on wave refraction along the wide sandy beaches of central Tamil Nadu coast. The nearshore waves approach the coast at 45° during the northeast (NE) monsoon, at 135° during the southwest (SW) monsoon and at 90° during the non-monsoon or fair-weather period with a predominant wave period of 8 and 10 s. A computer based wave refraction pattern is constructed to evaluate the trajectories of shoreward propagating waves along the coast in different seasons. The convergent wave rays during NE monsoon, leads to high energy wave condition which conveys a continuous erosion at foreshore region while divergent and inept condition of rays during the SW and non-monsoon, leads to moderate and less energy waves that clearly demarcates the rebuilt beach sediments through littoral sediment transport. The role of wave refraction in foreshore deposits was understood by grain size and depositional environment analysis. The presence of fine grains with the mixed population, during the NE monsoon reveals that the high energy wave condition and sediments were derived from beach and river environment. Conversely, the presence of medium grains with uniform population, during SW and non-monsoon attested less turbulence and sediments were derived from prolong propagation of onshore-offshore wave process. These upshots are apparently correlated with the in situ beach condition. On the whole, from this study it is understood that beaches underwent erosion during the NE monsoon and restored its original condition during the SW and non-monsoon seasons that exposed the stability of the beach and nearshore condition.

Besides the Indonesian throughflow (ITF), the South China Sea throughflow (SCSTF) also contributes to the water transport from the Pacific to the Indian Ocean. However, this South China Sea (SCS) branch at the Karimata Strait is poorly observed until 2007, even though its importance has been suggested by numerical studies for decades. In this paper, we review the nearly 10-year field measurement in the Karimata Strait by the execution of the projects of “SCS-Indonesian Seas Transport/Exchange (SITE) and Impacts on Seasonal Fish Migration” and “The Transport, Internal Waves and Mixing in the Indonesian Throughflow regions (TIMIT) and Impacts on Marine Ecosystem”, which extend the observations from the western Indonesian seas to the east to include the main channels of the ITF, is introduced. Some major achievements from these projects are summarized.

This study examines a Kuroshio main path (KMP) cut-off event east of Taiwan Island occurred in fall-winter 2013–2014 and its impacts on the South China Sea (SCS) by analyzing satellite altimetry and mooring observations. Satellite altimeter sea level anomaly (SLA) images reveal a complete process that a huge cyclonic eddy (CE) from the Pacific collided with the Kuroshio and the western boundary from 15 October 2013 to 15 January 2014. Mooring observations evidenced that the Kuroshio upper ocean volume transport was cut off more than 82% from 17×10⁶ m³/s in September to 3×10⁶ m³/s in November 2013. The KMP cut-off event caused the Kuroshio branching and intruding into the SCS and strengthened the eddy kinetic energy in the northern SCS west of the Luzon Strait. Using the total momentum as a dynamic criterion to determine the role of eddy collision with the Kuroshio reasonably explains the KMP cut-off event.

Mesoscale eddy generation mechanisms in the eastern South China Sea (ESCS) are investigated using altimetry observations and solutions of a nonlinear, 1½-layer reduced-gravity model. We estimate the relative roles of the wind forcing in the interior South China Sea (SCS) and the remote forcing from the western tropical Pacific (WTP) in eddy generation in the ESCS. Model solutions show that the high-frequency wind in the interior SCS is the primary forcing for eddies, which explains about 54% of the mesoscale eddies generated in the ESCS. Signals from the WTP also play an important role. Wind-driven equatorial signals reach the west coast of Luzon Island through the Sibutu Passage and Mindoro Strait. The reflected Rossby waves from the west coast of Luzon Island propagate westward, become unstable, and turn into eddies. The signals driven by high-frequency wind from the WTP explain about 40% of the mesoscale eddies generated in the ESCS. The high-frequency wind forcing in both the SCS and the WTP is important for eddy generation in the ESCS.

The seasonal characteristics and formation mechanism of the thermohaline structure of mesoscale eddy in the South China Sea are investigated using the latest eddy dataset and ARMOR3D data. Eddy-centric composites reveal that the horizontal distribution of temperature anomaly associated with eddy in winter is more of a dipole pattern in upper 50 m and tends to be centrosymmetric below 50 m, while in summer the distribution pattern is centrosymmetric in the entire water column. The horizontal distribution of eddy-induced salinity anomaly exhibits similar seasonal characteristics, except that the asymmetry of the salinity anomaly is weaker. The vertical distribution of temperature anomaly associated with eddy shows a monolayer structure, while the salinity anomaly demonstrates a triple-layer structure. Further analysis indicates that the vertical distribution of the anomalies is related to the vertical structure of background temperature and salinity fields, and the asymmetry of the anomalies in upper 50 m is mainly caused by the horizontal advection of background temperature and salinity.

In this study, subsurface eddies near the Vietnam coast of the South China Sea were observed with in situ observations, including Argo, CTD, XBT and some processed and quality controlled data. Based on temperature profiles from four Argo floats near the coast of Vietnam, a subsurface warm eddy was identified in spring and summer. The multi-year Argo and Global Temperature and Salinity Profile Programme (GTSPP) data were merged on a seasonal basis based on the data interpolating variational analysis (DIVA) method to reconstruct the three-dimensional temperature structure. There is a warm eddy in the central subsurface at 12.5°N, 111°E below 300 m depth in spring, which does not exist in autumn and is weak in winter and summer. From CSIRO Atlas of Regional Seas (CARS) and Generalized Digital Environment Model (GDEM) reanalysis data, this subsurface warm eddy is also verified in spring.

Four trawl-resistant bottom mounts, with acoustic Doppler current profilers (ADCPs) embedded, were deployed in the Karimata Strait from November 2008 to June 2015 as part of the South China Sea-Indonesian Seas Transport/Exchange and Impact on Seasonal Fish Migration (SITE) Program, to estimate the volume and property transport between the South China Sea and Indonesian seas via the strait. The observed current data reveal that the volume transport through the Karimata Strait exhibits significant seasonal variation. The winter-averaged (from December to February) transport is –1.99 Sv (1 Sv=1×10⁶ m³/s), while in the boreal summer (from June to August), the average transport is 0.69 Sv. Moreover, the average transport from January 2009 to December 2014 is –0.74 Sv (the positive/negative value indicates northward/southward transport). May and September are the transition period. In May, the currents in the Karimata Strait turn northward, consistent with the local monsoon. In September, the southeasterly trade wind is still present over the strait, driving surface water northward, whereas the bottom flow reverses direction, possibly because of the pressure gradient across the strait from north to south.

So far, large uncertainties of the Indonesian throughflow (ITF) reside in the eastern Indonesian seas, such as the Maluku Sea and the Halmahera Sea. In this study, the water sources of the Maluku Sea and the Halmahera Sea are diagnosed at seasonal and interannual timescales and at different vertical layers, using the state-of-the-art simulations of the Ocean General Circulation Model (OGCM) for Earth Simulator (OFES). Asian monsoon leaves clear seasonal footprints on the eastern Indonesian seas. Consequently, the subsurface waters (around 24.5σ_θ and at ~150 m) in both the Maluku Sea and the Halmahera Sea stem from the South Pacific (SP) during winter monsoon, but during summer monsoon the Maluku Sea is from the North Pacific (NP), and the Halmahera Sea is a mixture of waters originating from the NP and the SP. The monsoon impact decreases with depth, so that in the Maluku Sea, the intermediate water (around 26.8σ_θ and at ~480 m) is always from the northern Banda Sea and the Halmahera Sea water is mainly from the SP in winter and the Banda Sea in summer. The deep waters (around 27.2σ_θ and at ~1 040 m) in both seas are from the SP, with weak seasonal variability. At the interannual timescale, the subsurface water in the Maluku Sea originates from the NP/SP during El Niño/La Niña, while the subsurface water in the Halmahera Sea always originates from the SP. Similar to the seasonal variability, the intermediate water in Maluku Sea mainly comes from the Banda Sea and the Halmahera Sea always originates from the SP. The deep waters in both seas are from the SP. Our findings are helpful for drawing a comprehensive picture of the water properties in the Indonesian seas and will contribute to a better understanding of the ocean-atmosphere interaction over the maritime continent.