Jump to the main content block
  1. Home
  2. SDG 14

14.5.3

SDG 14.5.3

  1. The ecosystem dynamics associated with the environmental change surrounding the upwelling zone of Taiwan Bank resources in the coastal water off Taiwan maculate in the southwestern waters of Taiwan台灣淺灘(Taiwan Bank)湧昇區暨周邊水域生態系動態特性影響之研究

The knowledge regarding the feeding ecology of the greater amberjack remains limited despite its ecological and economic importance. Therefore, this study investigated the feeding dynamics of the greater amberjack and its key prey species in Taiwanese waters. Samples were collected from the Taiwan Bank (22.5° N–24.5° N, 118.5° E–121.5° E) and Northern Taiwan waters (25° N–26° N, 121° E–123.5° E). Analysis of samples obtained between June 2020 and June 2022 indicates that environmental factors influenced the availability of prey species, thereby affecting dietary preferences. The prey-specific index of relative importance indicated that the predominant prey species were pelagic species (52.81%), followed by demersal species and unidentified teleosts (18.74% and 16.58%, respectively); squid and crustaceans were the least frequently consumed prey species (6.11% and 5.76%, respectively). Although no difference was discovered between males and females in terms of diet, seasonal and size-related variations were noted in the feeding patterns, as evident from the substantial ontogenetic shift observed in the dietary composition of the samples during the study period. The findings of this study improve the understanding of the feeding dynamics of the greater amberjack and the association between oceanographic factors, such as sea surface temperature (SST), sea surface salinity (SSS), sea surface height (SSH), mixed-layer depth (MLD), and eddy kinetic energy (EKE), and prey availability in the coastal waters of Taiwan.

Sustainable Impact: The results of this study contribute to further exploring the ecological role of the greater amberjack in Taiwan's coastal waters and its relationship with environmental changes.

儘管杜氏鰤具有生態和經濟重要性，但有關其攝食生態的知識仍然有限。因此，本研究調查了台灣水域中杜氏鰤及其主要餌料生物的攝食動態。樣本採集自台灣淺灘(22.5°N-24.5°N, 118.5°E–121.5°E)及台灣北部水域(25°N-26°N, 121°E-123.5°E)。由2020年6月至2022年6月的樣本分析顯示，環境因素影響了餌料生物物種的可用性，從而影響了攝食偏好。餌料生物特異性相對重要性指數顯示，主要餌料生物種類為中上層種類(52.81%)，其次是底層種類和不明硬骨魚(分別為18.74%和16.58%)；魷魚和甲殼類動物是最不常被食用的餌料生物物種(分別為 6.11% 和 5.76%)。儘管在攝食方面沒有發現雄性和雌性之間的差異，但在攝食模式中註意到季節性和與體型相關的變化，這一點從研究期間觀察到的樣本飲食成分中觀察到的實質性個體發育變化中可以明顯看出。這項研究的結果提高了對杜氏鰤攝食動態以及海洋因素之間關聯的理解，例如海面溫度(SST)、海面鹽度(SSS)、海面高度(SSH)）、混合層深度(MLD)、渦動能(EKE)以及台灣沿海水域的餌料生物可用性。

永續影響力: 本研究結果有助於進一步探索杜氏鰤在台灣沿海水域的生態角色及其與環境變化的關聯。

Evidence: https://r088.ntou.edu.tw/var/file/103/1103/img/1564/757150331.pdf

  1. 112th ‘‘Advancing Satellite Monitoring Marine Environment Application Products’’ outsourcing case 112年度「精進衛星監測海洋環境應用產品」委外案

This project will make use of the continuous observation characteristics of Sunflower satellite data in high spatial and temporal resolution and panoramic range to refine sea surface chlorophyll concentration and sea surface temperature calculation algorithms, strengthen the display function of the satellite product fishery information platform, and explore the surrounding waters and coastal areas of Taiwan. , for the collection and analysis of disaster-causing weather and fishery (including aquaculture) information, it is expected to establish an early risk estimation model for satellite products in fishery disasters in the context of extreme weather affecting fish habitats, and reduce economic and social losses. risk. The main work of this year (112) is to

  1. improve the sea surface chlorophyll concentration algorithm in the panoramic range;
  2. develop artificial intelligence methods to establish all-weather hourly sea surface temperature products in target sea areas;
  3. apply artificial intelligence to develop PM2 in East Asia .5 Concentration ground estimation field production technology;
  4. Improve satellite fishery platform information and operational function optimization;
  5. Explore the status of disaster-causing weather and satellite products and fishery (including breeding) information.

In the development of panoramic sea surface chlorophyll concentration algorithm project, ocean water color algorithms were established for three sea areas. The results show that the chlorophyll concentration value of the Sunflower satellite panorama in this study is consistent with MODIS/Aqua, both in value and spatial distribution. The results of the chlorophyll concentration values ​​of MODIS /Terra, SNPP-VIIRS, and JPSS1-NOAA20 are quite consistent. This means that the accuracy of the chlorophyll concentration values ​​and distribution maps calculated by this study on the Sunflower satellite panorama is at a certain level.

In the research and development of artificial intelligence methods to establish all-weather hourly sea surface temperature products in target sea areas, this year’s work expanded the time-spatial-spatial radial basis function neural network (TS-RBFNN) to hourly sea surface temperature products. Surface temperature reconstruction and continuous development of the Temporal Periodic Neural Network (TPNN) model module. Its principle is similar to the developed amplitude basis function neural network. The main difference is that it is modified into a periodic function ( Determine the number of neurons and the period of sea surface temperature changes) to reflect the periodic changes in sea surface temperature. In the actual case of hourly sea temperature reconstruction, the root mean square error (0.436°C) of the reconstructed sea temperature of TPNN is higher than that of Temporal Radial Based Function Neural Network (TRBFNN), and the calculation efficiency is It is about three times faster than TRBFNN (0.0064 seconds).

In the technical project of applying artificial intelligence to develop a ground-based estimation field for PM2.5 concentration in East Asia, a prototype model of a ground-based estimation field for artificial intelligence PM2.5 concentration in East Asia has been initially built. At this stage, the model can effectively grasp the real observed concentration. In the final stage, the 110-year DNN model design was improved. By combining the PM2.5 concentration observation difference mechanism at the measuring station to evaluate the impact on the concentration estimates in coastal and mountainous areas, the estimation results of sea grid concentration in Taiwan and East Asia were provided. .

In the project of upgrading the information and operational functions of the satellite fishery platform, we have completed the refinement of habitat suitability and species distribution model forecasting functions, and planned and constructed the important fishery economic species model for mullet, sea bream, black sea bream, pomfret (white seabream). There are a total of 10 species of economically important fish species along the coast, including pomfret, pomfret, pomfret, and kingfish. The platform information integrates meteorological satellite and hydrological environment data, mainly providing information related to fishermen's operations and fishing ground forecast needs and fishery resource monitoring. Satellite telemetry data can be used as reference and evidence for academic research or project implementation, and through satellite telemetry image analysis, A marine fishery service application platform can be built to provide more diversified and customized service products, improving the application level of satellite products and the convenience and promotion of the operational service system.

In the project to explore disaster-causing weather and satellite products and fishery (including aquaculture) information, this year's analysis mainly focused on Taiwan's important farmed fish species - milkfish. The main production areas are Tainan City, Kaohsiung City, Chiayi County and Yunnan Province. In Lin County, according to the production type, it can be mainly divided into deep water exclusive culture and shallow flat polyculture. Milkfish is a temperate narrow-temperature bony fish with poor cold tolerance. Generally, its activity will significantly decline when the water temperature is lower than 14°C, and individuals will begin to die below 13°C. This study collected more than 110,000 water temperature data from the Tainan Water Quality Monitoring Station (sensor water depth is 40 cm) and temperature data from four nearby weather stations. We used regression analysis and combined with the temperature hazard quotient to analyze the correlation between water temperature and air temperature. The results show that water temperature has a delay characteristic of at least 3 hours with air temperature. When the temperature is 10°C, the water temperature of the breeding pond is about 13°C. Under these conditions, milkfish begin to die, causing breeding losses.

Sustainable Impact: The outcomes of this project not only enhance the efficiency of fishery management but also provide scientific evidence for response strategies under extreme weather conditions, thereby promoting the sustainable development of Taiwan's fisheries.

本計畫將利用向日葵衛星資料在高時空解析度及全景範圍的連續觀測特性，精進海表面葉綠素濃度及海表溫演算法，強化衛星產品漁業資訊平臺展示功能，並探討臺灣周遭海域與沿岸區域，對於致災之天候與漁業(含養殖)資訊的樣態蒐集分析，期能為極端天候影響魚種棲地的情境，建立衛星產品在漁業災害的早期風險推估模式，降低經濟與社會損失的風險。本(112)年度主要工作為

  1. 提升全景範圍的海表面葉綠素濃度演算法；
  2. 研發人工智慧方法建立目標海域全天候逐時海表溫產品；
  3. 應用人工智慧開發東亞區域PM2.5濃度地面推估場產製技術；
  4. 提升衛星漁業平台資訊及操作功能優化；
  5. 探討致災天候及衛星產品與漁業(含養殖)資訊的樣態。

在發展全景範圍的海表面葉綠素濃度演算法工項，針對三種海域分別建立海洋水色演算法，結果顯示本研究向日葵衛星全景的葉綠素濃度值不論在值或空間的分布，都與MODIS/Aqua, MODIS/Terra, SNPP-VIIRS, JPSS1-NOAA20的葉綠素濃度值的結果相當一致。這表示本研究所推算向日葵衛星全景的葉綠素濃度值及其分布圖的準確度有一定的水準。

在研發人工智慧方法建立目標海域全天候逐時海表溫產品工項，本年度工作拓展時間–空間幅狀基底函數類神經網路(Temporal-Spatial Radial Basis Function Neural Network, TS-RBFNN)至逐時海表溫重建及持續開發時間週期性類神經網路模式模組(Temporal Periodic Neural Network, TPNN)，其原理與已開發之幅狀基底函數類神經網路相似，主要差異在改納入週期性函數(決定神經元個數及其海表溫變化之週期)以反應海表溫的週期性變化。在實際變化的逐時海溫重建案例中，TPNN的重建海溫之均方根誤差(0.436℃)高於時間幅狀基底類神經網路(Temporal Radial Based Function Neural Network, TRBFNN)，計算之效率較TRBFNN提升三倍左右(0.0064秒)。

在應用人工智慧開發東亞區域PM2.5濃度地面推估場產製技術工項，初步建置東亞區域人工智慧PM2.5濃度地面推估場模型雛形，現階段模型已可有效掌握真實觀測濃度。於期末階段改良110年之DNN模型設計，藉由結合測站PM2.5濃度觀測差值機制以評估對於沿海與山區濃度推估值的影響，提供臺灣區域與東亞區域海上格點濃度推估結果。

在提升衛星漁業平台資訊及操作功能優化工項，目前已完成精進棲地適合度與物種分布模式預報功能，在重要漁業經濟物種模式規劃建置烏魚、午仔、黑鯛、鯧科(白鯧、定盤)、鰺科與石首魚等，沿近海重要經濟魚種共計10種。平台資訊整合氣象衛星水文環境資料，提供漁民作業及漁場預測需求與漁業資源監控相關資訊為主；衛星遙測資料可作為學術研究或執行計畫時之參考與佐證，並透過衛星遙測影像分析，可建置海洋漁業服務應用平台，提供更多樣化與客製化服務之商品，提升衛星產品應用層面及作業服務系統便利性及推廣使用度。

在探討致災天候及衛星產品與漁業(含養殖)資訊的樣態工項，本年度主要針對臺灣重要養殖魚種-虱目魚進行分析，主要生產區為臺南市、高雄市、嘉義縣及雲林縣，依照生產型態主要可分為深水專養及淺坪混養。虱目魚為溫帶狹溫性硬骨魚類，耐寒性不佳，一般水溫低於14 ℃其活動力便明顯下滑，13 ℃以下便會開始有個體死亡。本研究蒐集台南水質監測站（感應器水深40公分）共逾11萬筆水溫資料與鄰近4個氣象站之氣溫資料利用迴歸分析並搭配氣溫危害商數，進行水溫與氣溫之相關性分析。結果顯示，水溫之於氣溫至少有3小時的延遲特性，而在氣溫10℃時，養殖池塘水溫約為13℃，在此條件下虱目魚便開始產生死亡，造成養殖損失。

永續影響力: 本計畫的成果不僅能提升漁業管理的效率，還能為極端天候下的漁業應對策略提供科學依據，進而促進臺灣漁業的可持續發展。

Evidence:

https://r088.ntou.edu.tw/var/file/103/1103/img/1564/804006726.pdf

Click Num:
Print