본 연구에서는 인공지능기법을 이용하여 진동만의 용존산소량 예측을 하였다. 관측자료에 존재하는 결측 구간을 보간하기 위해 양방향재귀신경망(BRITS, Bidirectional Recurrent Imputation for Time Series) 딥러닝 알고리즘을 이용하였고, 대표적 시계열 예측 선형모델인 ARIMA(Auto-Regressive Integrated Moving Average)과 비선형모델 중 가장 많이 이용되고 있는 LSTM(Long Short-Term Memory) 모델을 이용 하여 진동만의 용존산소량을 예측하고 그 성능을 평가했다. 결측 구간 보정 실험은 표층에서 높은 정확도로 보정이 가능했으나, 저층에서는 그 정확도가 낮았으며, 중층에서는 실험조건에 따라 정확도가 불안정하게 나타났다. 실험조건에 따라 정확도가 불안정하게 나타났다. 결과로부터 LSTM 모델이 중층과 저층에서 ARIMA 모델보다 우세한 정확도를 보였으나, 표층에서는 ARIMA모델의 정확도가 약간 높은 것으로 나타났다.

The objectives of this study to determine seasonal changes in species composition and abundance of fish and decapod assemblage, and the relationships between environmental factors and their abundance in an eelgrass bed of Jindong Bay. A total of 26 fish species and 29 decapod species were collected by a small beam trawl from an eelgrass bed in Jindong Bay in 2002. The dominant fish species were Hexagrammos otakii, Pholis neulosa and P. fangi and these accounted for 48.4 % in the total number of individuals. Dominant decapod taxa were Palaemon macrodactylus, Charybdis japonica, Pagurus minutus and C. bimaculata. These were primarily small species or early juveniles of larger species. Species composition and abundance varied greatly showing a peak in the number of individuals in April and May, and peak biomass in fish in July and decapods in August. Catch rate was low in winter months both in fish and in decapod. Seasonal changes in the abundance of fishes and decapods corresponded with eelgrass biomass and abundance of food organisms indirectly.

본 연구에서는 미더덕양식장이 밀집되어 있는 남해안의 진동만 10개 정점에 대하여 미더덕과 주름미더덕의 양식순기에 맞추어 계절별(7월, 10월, 1월, 4월)로 실시하였으며, 어장 서식지 적합지수를 위한 어장평가인자로 수온, 클로로필 a, 부유입자물질, 해수유동, 염분, 용존산소 포화도를 이용하였다. 미더덕과 주름미더덕의 성장과 어장환경 상관성을 나타낼 수 있도록 각 인자들에 대한 그래프가 작성되었으며, 서식지 적합지수 산정결과 진동만의 서쪽해역(HSI 0.32~0.41)이 동쪽해역(HSI 0.19~0.27)보다 높은 적합지수를 보였다. 이러한 결과는 미더덕과 주름미더덕의 단위생산량과 유의한 상관성(R²=0.834, P<0.01)이 있음을 보였다. 본 연구는 과학적 어장적지평가를 통하여 지속적인 생산성을 위한 어장관리에 기여를 할 수 있을 것으로 판단된다.

Jindong bay at the northwestern part of Jinhae bay suffers from the occurrence of red tides in summer every year. In order to study the management methods of coastal environments, an ecological numerical model has been developed. The model experiments was forecasted that the load of nutrients from the land and field concentration will be cut down per 10% each. When we cut down 57.2% nitrogen load in the inner bay and 38.4% phosphorous load in the outer bay of bottom layer of the nutrients load from land and field concentration, the seawater quality standard levels up first grade. When we cut down 86.5% nutrients in the inner bay and 93.0% nutrients in the outer bay, the concentration of chlorophyll a decreases below 3.2(an individual concentration of phytoplankton : 10,000cell/ml), i.e. the red tides do not occur.

We studied the appearance and proliferation of red tide organisms from March, 1998, to February, 1999, in the Jindong Bay. There were two red tide events during investigation, and we found that the dominant species Ceratium furca in May and Gymnodinium sanguineum in September, 1998. At surface, temperature and salinity showed 18.3∼19.7℃ and 30.6∼30.9 psu in May and 25.6∼27.0℃, 28.0∼28.5 psu in September, respectively. When the red tide occurred, the water mass was stable. Dissolved inorganic nitrogen(DIN) and dissolved inorganic phosphate(DIP) in Jindong Bay showed seasonal variability. In May, the nitrogen was a limited nutrient in which the ratio of DIN versus DIP was less than 16 (Redfield ratio), while in September phosphate. During June to September, 1998, phosphate acted as a limited nutrient due to the increased river run-off from land. In May, chlorophyll-a, dissolved oxygen(DO) and chemical oxygen demand(COD) showed higher than 20㎍/ℓ, l0㎎/ℓ and 5.0㎎/ℓ, respectively. In September, they showed in turns 100㎍/ℓ, 10㎎/ℓ and 10㎎/ℓ, respectively.