|Abstract: ||海洋與大氣皆有潮汐現象，參數分別是水位、水溫及氣壓、氣溫四項。水體的前兩項在相同的潮分量上高度相關，氣體的後兩項也一樣，但水體與氣體交叉互動將如何。為了確定在潮汐頻率上「海－氣交互影響」的成分，需有海上與附近陸上的觀測來比較，故選擇中央氣象局在馬祖陸上及海域浮標的每小時水溫、氣溫及氣壓數據，水位資料來自港內的潮位儀。因此共有陸上的氣溫、氣壓，海上的水位、水溫，浮標的氣溫、氣壓等六種參數自2011年6月1日至2013年12月31日的2.5年時序。 從觀測資料中，精確濾出各參數的各主要潮分量時序，即全日潮S1（水位只有K1）及半日潮S2、M2。四種當今最常用的高效能濾波方法：IFFT、Butterworth、EMD及Multi-running Mean，均加以仔細測試檢驗，結果顯示傳統的IFFT，對細膩拆解各潮分量而言，是最佳選擇。於是獲得六種參數共17項全日潮與半日潮的潮分量時序（氣溫沒有M2分量）。 這些分量的每日24小時系集（ensemble）圖，呈現出「波導管」（waveguide）狀態。在外觀上，因每天相位偏移，M2（及K1）分量皆無波導現象；其他分量均呈正弦曲線但多樣的波導圖案；氣壓場的波導性質超強，其波導管很細。高峰時間上，S1均非正午，因為陽光熱能累積要時間；低谷亦非半夜，熱輻射消散也要時間。季節演變顯示在波導的振幅上，外觀較複雜者，分季節處理繪製系集圖，即可見分別各季波導在時間上的轉換過程。 17種潮分量間，彼此兩兩交相關函數計算得知：要考慮太陽作用的S1分量，氣溫與氣壓反相互動，在海上是立即的，陸地則是氣溫領先氣壓達3小時；水溫則領先（加溫或冷卻了）氣溫約只需1小時。不考慮太陽作用的S2分量，海、陸氣壓完全相關，但氣溫相關偏低；陸上氣溫、氣壓的交叉相關不錯，海上則否。M2分量上，海、陸氣壓相關稍低；海上氣壓與水位高度相關，反應馬祖海域大潮差影響了浮標上量測的氣壓之高度差；水溫與水位呈高度正相關，亦即漲潮時水溫增高，不是次表層冷水垂直帶上海面而降溫的結果，故從海面溫度梯度受水平潮流驅動來探討，成為重要的關鍵議題。 探討東海至南海北端近岸環境，中國沿岸流CCC（China Coastal Current）和湧升流CCU（China Coastal Upwelling）是兩個主要現象。長江和錢塘江排出的水大幅影響東北季風下CCC的特性，西南季風則為CCU的推動機制。零散的海洋科學調查，甚至從過去NOAA的34年衛星海表溫（SST）觀測，都不能在時間和空間上描繪該對偶現象。利用一個截向和順向海表溫度梯度的簡單方法，竟然可以從SST資料上清楚顯示出CCC和CCU的外觀及其季節性的交替變化，亦即CCC盛行於寒冷季節，CCU在溫暖季節。大陸沿岸海域終年冷水盛行的水溫場，受不同季風有強度差異的影響，顯示CCC的力道明顯大於CCU。經由台灣海洋科技研究中心（TORI）的臺灣多重尺度社群海洋模式TaiCOM（Taiwan Community Ocean Model）在東海的潮流分析，證實了馬祖海域水位潮與水溫潮在最強成分的水位M2分量上有正相關，是西南西漲潮流帶來離岸較暖海水的結果。 1月至2月東北季風最盛時，黑潮入侵台灣北部的東海造成套流，會與CCC交互作用，把CCC的水引到台灣北邊的湧升區域來。這個34年系集統計現象，在逐年分析上並非年年可見。然而這種機制可將CCC在臺灣海峽北部截斷為北邊的浙閩沿岸流和南邊的廣東沿岸流。3月後，完整的CCC再度恢復。台灣北部伴隨黑潮套流而生的湧升，亦被海表溫度梯度解析看見季節性移動。CCC截向的海表溫度梯度之典型平均値約為0.2±0.07℃/10km，在一月和二月時，極端值達到0.6±0.2℃/10km，比黑潮的西邊海表溫度梯度大了數倍。概念上，CCC的寬度約在35到140公里之間。|
Tides are found in atmosphere as in the oceans and can be parameterized by wa-ter level and water temperature together with air pressure and air temperature. For the sea, the former two are of course highly correlated in the same tide constituent, and also for the latter two for the air. However, what could be happened between the sea and the air. In order to characterize the air-sea interaction in tide frequencies, observed data including sea water and air on the adjacent land are required. For that, hourly weather data at Matzu and a nearby buoy including water temperature, air temperature and air pressure provided by the Central Weather Bureau (CWB) are employed. CWB also measures water level from a tide gauge in harbor. In total, there are six parameters as air temperature and air pressure on land, water level and water temperature at sea, air temperature and air pressure at buoy, for a time series of 2.5 years from June 1 in 2011 to December 31 in 2013. Therefore, it is necessary to accurately filter out the principle tidal constituents from the time series of the six parameters, namely the diurnal S1 (but K1 for water level) and semi-diurnal S2 and M2. Four powerful filters in common use at present, the IFFT, the Butterworth, the EMD and the Multi-running Mean, are tested and carefully examined. The results show that the most traditional IFFT method is the best to exquisitely separate those constituents. And then, 17 time series of diurnal and semi-diurnal tides for the six parameters (no M2 for air temperature) are obtained. The 24-hours ensemble diagram of these tidal constituents reveals various pat-terns of waveguide. In exterior, there’s no waveguide for M2 (and K1) due to the daily phase shift. Others are shaped as sinusoidal curves but in variety for waveguide patterns. Air pressure is strongly characterized with an extremely narrow waveguide. For climax, they are not at noon-time for S1, because it needs time to accumulate the solar heat. For trough, they are not at mid-night, because it also needs time to relax the heat by radiation. For seasonal variation, it is seen from the amplitude of waveguide. Those rather complex waveguides can be graphed in seasons to display how the waveguide varies in time. The cross-correlations between any two of the 17 constituents disclose many understandings. For S1, requiring the consideration of solar effect, the variations of air temperature and pressure are out of phase, promptly at sea, but temperature leading pressure 3 hours on land. Water temperature, on the other hand, leads the air tem-perature 1 hour, meaning that the water needs 1 hour to warm or cool the air. For S2, no consideration for solar effect, the air pressures at sea and land are completely cor-related, but in low correlation for air temperature. The cross-correlation between temperature and pressure is good on land, but not at sea. For M2, the air pressures at sea and land are rather low correlated. But the air pressure at sea is highly correlated with water level, reflecting the effect of high tide range at Matzu to the measurement of air pressure at buoy with varying altitude. Water temperature and water level are in highly positive correlation, i.e., water temperature increases in flood. Apparently, it is not the case that the flood tide should lift subsurface cool water vertically to decrease the surface water temperature. Therefore, to study how the tidal current horizontally driving the water temperature gradient becomes an important key issue. From the East China Sea (ECS) to the north of the South China Sea, the two foremost features are the China Coastal Current (CCC) and the China Coastal Upwelling (CCU). The discharged water from the Yangtze River and the Qiantang River strongly characterize the water of the CCC in northeast monsoon, while the CCU is induced by the southwest monsoon. Sporadic surveys in oceanography, even with the NOAA 34 years satellite SST data, may not limn the coupled actions. Apply-ing a simple method of transverse and longitudinal temperature gradients from the archived SST, to our surprise, the outlook and their seasonal alternation of the CCC and CCU is clearly seen, the former prevails in cold seasons, while the latter in warm seasons. Such prevailed coastal cold water along the mainland China is influenced by the different intensity of monsoons, indicating that the strength of the CCC is much greater than the CCU. The TAIwan multi-scale Community Ocean Model (TaiCOM) by TORI (Taiwan Ocean Research Institute) shows the west southwest flood currents in the ECS, evidencing the fact that the currents carry the offshore warm water to the coast in flood to cause the positive cross-correlation of water temperature and water level in the most significant M2 tide. Nevertheless, amazing result is that in January to February, when the northeast monsoon prevails, Kuroshio intrudes the ECS to form a loop current at north of Tai-wan and cause an offshore upwelling, may draw the CCC detoured to the upwelling region. Such a 34 years ensemble result does not always exist in individual year anal-ysis. However, at north of the Taiwan Strait, this mechanism may cut off the CCC into the ZheMin Coastal Current and the Guangdong Coastal Current. After March, the complete CCC would be restored. At north of Taiwan, the upwelling accompanied with the Kuroshio’s loop current was also observed moving seasonally in east-west via the SST temperature gradients analysis. The typical mean transverse SST gradient in the CCC basin was approximately 0.2±0.07℃/10 km, while in January and February the extreme value could reach 0.6±0.2℃/10 km, several times greater than on the Kuroshio’s west boundary. Conceptually, the width of the CCC could be between 35 and 140 km.