• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• v.9 no.1
• /
• pp.397-402
• /
• 2005

In the Korean seas, Sea Surface Temperature (SST) and Thermal Fronts (TF) were analyzed temporally and spatially during 8 years from 1993 to 2000 using NOAA/AVHRR MCSST As the result of EOF method applying SST, the variance of the 1st mode was 97.6%. It is suitable to explain SST conditions in the whole Korean seas. Time coefficients were shown annual variations and spatial distributions were shown the closer to the continent the higher SST variations like as annual amplitudes. The 2nd mode presented higher time coefficients of 1993, 94, and 95 than those of other years. Although the influence is a little, that tan explain EININO effort to the Korean seas. TF were detected by Sobel Edge Detection Method using gradient of SST. Consequently, TF were divided into 4 fronts; the Subpolar Front (SPF) dividing into the north and south part of the East sea , the Kuroshio Front (KF) in the East China Sea (ESC), the South Sea Coastal Front (SSCF) in the South sea, and the Tidal Front in the West sea. TF located in steep slope of submarine topography. The distributions of 1st mode in SST were bounded in the same place, and these results should be considered to influence of seasonal variations. To discover temporal and spatial variations of TF, SST gradient values were analyzed by EOF. The time coefficients fo the 1st mode (variance : 64.55%) showed distinctive annual variations and SPF, KF, and SSCF was significantly appeared in March. the spatial distributions of the 2nd mode showed contrast distribution, as SPF and SSCF had strong'-'value, where KF had strong'+'value. The time of'+'and'-'value was May and October, respectively. Time coefficients of the 3rd mode had 2 peaks per year and showed definite seasonal variations. SPF represented striking'+'value which time was March and October. That was result reflected time of the 1st and 2nd mode. We can suggest specific temporal and spatial variations of TF using EOF.

• 한국해양학회지
• /
• v.30 no.6
• /
• pp.565-583
• /
• 1995

We have used a nonlinear quasi-geostrophic model to study effects of lateral friction and bottom topography on the motion of warm eddies. The two empirical orthogonal functions of the stream function, accounting for the vertical structure, represent the barotropic and first baroclinic dynamic modes. This model is integrated 360 days on a 1000 km ${\times}$ 1000 km domain with a resolution of 10 km ${\times}$ 10 km including both the thermocline and idealized topography of the East Sea. Prescribed inflow through the Korea Strait is compensated by outflow through the Tsugaru Strait. The balance between the nonlinear advection term and the planetary ${\beta}$-effect tends to make northward movement of warm eddy over a flat bottom. The motion of a warm eddy over a sloping topography can be dominated by the nonlinear advection, while nonlinearity plays a secondary role over a flat topography. For eddies dispersing over topography, the nonlinear tendency is a function of time. For a strong warm eddy, northward propagation can occur. For intermediate strength of eddies one might expect a balance between the nonlinear term and the topographic ${\beta}$-effect. As nonlinearity decreases with eddy dispersion, southward motion along the slope may occur by such as a topographic Rossby wave. Our numerical simulations have confirmed the importance of lateral friction on eddy motions, in such a way that the northward penetration of the warm eddy increases drastically by the decrease of the lateral friction. The northward motion of warm eddy can be prevented by reducing the Reynolds number sufficiently. We have also demonstrated the crucial role of topographic effects in the eddy motion process.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.27 no.4
• /
• pp.293-302
• /
• 1991

The anomalous sea level deviation or storm surge caused by the typhoon Thelma in 1987 are studied analysing tidal observation data at 7 stations in the south coast of Korean peninsula. The surges are calculated by subtracting the predicted tidal height from the observed tidal record. The tidal deviation at these stations along the coast are discussed in association with meteorological data. The sea level anomalies are studied by means of the empirical orthogonal function (EOF) analysis and the fast fourier transform (FFT) method. The results of analysis suggest that the peak value of surges are higher at the tidal stations in semi-enclosed bay and in long narrow channel than at the ones facing with the open sea. From the result of EOF analysis, the temporal and spatial fluctuations of storm surge can be described by the first EOF mode, which explains 63% of the total variances during the passage of typhoon Thelma. The deviation of storm surge in the studied areas indicates bi-modal peak during the passage of typhoon Thelma. From the results of FFT spectrum analysis, the peak of energy of autospectrum for surge, atmospheric pressure, and wind stress appeared at low frequency fluctuations band of 0.008-0.076 cph over the 4 stations. Auto-correlation function of surge showed periodicity, while that of atmospheric pressure and wind stress indicates no periodicity. The result of FFT analysis shows that the typhoon surges are related chiefly with the change of atmospheric pressure in an open bay (Cheju Harbor), but with the wind stress in a semi-enclosed bay (Yeosu Harbor).

• The Sea:JOURNAL OF THE KOREAN SOCIETY OF OCEANOGRAPHY
• /
• v.5 no.2
• /
• pp.86-94
• /
• 2000

In order to extract the spatio-temporal characteristics of long-term variability of the net heat flux on the sea surface in the East Asian marginal seas, empirical orthogonal function (EOF) analysis was conducted using data set calculated every 12 hours interval during 1978-1995. Among the first three modes explaining 73% of the total variance, the first mode having high peak at 1 year period indicates high variability area around the Sandong Peninsula and central and northeastern part of the East Sea. In the second mode which has spatial distribution of dipole type at the north and south, the peaks appear at 3.6 year and 2.3 year cycles. Time coefficient of the second EOF is believed to have close relation with the E1 Nino and has out-of-phase variation with NINO3 SST. Lagged correlation between NINO3 SST and time coefficient of the second EOF indicates four month time delay in the NINO3 SST. In the third mode which has opposite sign at the east and west, the periodicity of 6-9 year cycle has relatively clear appearance compared to other two EOFs. Also, high heat loss exceeding 800 W/$m^{2}$ in winter time occured at the south part of the Sandong Peninsula and Vladivostok. It reveals more frequent occurrence of about two times at the Sandong Peninsula than Vladivostok. The event is concentrated in January at Vladivostok, but it occurs primarily in December and January at the Sandong Peninsula.

• Korean Journal of Remote Sensing
• /
• v.30 no.5
• /
• pp.687-701
• /
• 2014

Sea ice is generally accepted as an important factor to understand the process of earth climate changes and is the basis of earth system models for analysis and prediction of the climate changes. To continuously monitor sea ice changes at kilometer scale, it is demanded to create more accurate grid data from the current, limited sea ice data. In this paper we described a downscaling method for Advanced Microwave Scanning Radiometer 2 (AMSR2) Sea Ice Concentration (SIC) from 10 km to 1 km resolution using a weighting scheme of sea ice days ratio derived from Moderate Resolution Imaging Spectroradiometer (MODIS) sea ice cover product that has a high correlation with the SIC. In a case study for Okhotsk Sea, the sea ice areas of both data (before and after downscaling) were identical, and the monthly means and standard deviations of SIC exhibited almost the same values. Also, Empirical Orthogonal Function (EOF) analyses showed that three kinds of SIC data (ERA-Interim, original AMSR2, and downscaled AMSR2) had very similar principal components for spatial and temporal variations. Our method can apply to downscaling of other continuous variables in the form of ratio such as percentage and can contribute to monitoring small-scale changes of sea ice by providing finer SIC data.

• Journal of the Korean earth science society
• /
• v.43 no.1
• /
• pp.30-40
• /
• 2022

Wildfires, which occur sporadically and irregularly worldwide, are distinct natural disturbances in combustible vegetation areas, important parts of the global carbon cycle, and natural disasters that cause severe public emergencies. While many previous studies have investigated the variability and changes in wildfires globally based on fire emissions, burned areas, and fire weather indices, studies on East Asia are still limited. Here, we explore the characteristics of variability and changes in wildfire danger over East Asia by analyzing the fire weather index for the 40 years-1981-2020. The first empirical orthogonal function (EOF) mode of fire weather index variability represents an increasing trend in wildfire danger over most parts of East Asia over the last 40 years, accounting for 29% of the total variance. The major contributor is an increase in the surface temperature in East Asia associated with global warming and multidecadal ocean variations. The effect of temperature was slightly offset by the increase in soil moisture. The second EOF mode exhibits considerable interannual variability associated with the El Nino-Southern Oscillation, accounting for 17% of the total variance. The increase (decrease) in precipitation in East Asia during El Nino (La Nina) increases (decreases) soil moisture, which in turn reduces (increases) wildfire danger. This dominant soil moisture effect was slightly offset by the temperature increase (decrease) during El Nino (La Nina). Improving the understanding of variability and changes in wildfire danger will have important implications for reducing social, economic, and ecological losses associated with wildfire occurrences.

• Korean Journal of Remote Sensing
• /
• v.33 no.6_1
• /
• pp.917-930
• /
• 2017

To understand the temporal and spatial variations of surface chlorophyll-a concentration (Chl-a) distribution in the Indian Ocean ($30^{\circ}E{\sim}120^{\circ}E$, $30^{\circ}S{\sim}30^{\circ}N$) by the Indian Ocean Dipole (IOD), we conducted EOF and K means analyses of monthly satellite-derived Chl-a data in the region during 1998~2016 periods. Chl-a showed low values in the central region of the Indian Ocean and relatively high values in the upwelling region and around the marginal regions of the Indian Ocean. It also had a strong seasonal variation of Chl-a, showing the lowest value in the spring and the highest value in summer due to the change of the monsoon and current system. The EOF analysis showed that Chl-a variation in EOF mode 1 is related to ENSO (El $Ni{\tilde{n}}o$/Southern Oscillation) and that of mode 2 is linked to IOD. Both modes explained spatially opposite trends of Chl-a in the east and west Indian Ocean. From K means analysis, the Chl-a variation in the east and west Indian Ocean, and around India have relatively good relationship with IOD while that in the tropical and middle Indian Ocean closely associated with ENSO. The spatial and temporal distribution of Chl-a also showed distinct spatial and temporal variations depend on the different types of IOD events. IOD classifies two patterns, which occurred during the developing ENSO (First Type IOD) and the year following ENSO event (Second Type IOD). Chl-a variation in the First Type IOD started in summer and peaked in fall around the east and west Indian Ocean. Chl-a variation in the Second Type IOD occurred started in spring, peaked in summer and fall, and disappeared in winter. In the Chl-a variation related to IOD, developing process appearing in the Chl-a difference between the east and west Indian ocean was similar. Chl-a variation in the northern Indian Ocean were opposite trend with changing developing phase of IOD.

• Journal of the Korean earth science society
• /
• v.21 no.2
• /
• pp.137-158
• /
• 2000

The correlation and Empirical Orthogonal Function (EOF) analyses over the globe have been applied to intercompare lower-stratospheric (${\sim}$70hPa) temperature obtained from satellite data and two model reanalyses. The data is the19 years (1980-98) Microwave Sounding Unit (MSU) channel 4 (Ch4) brightness temperature, and the reanalyses are GCM (NCEP, 1980-97; GEOS, 1981-94) outputs. In MSU monthly climatological anomaly, the temperature substantially decreases by ${\sim}$21k in winter over southern polar regions, and its annual cycle over tropics is weak. In October the temperature and total ozone over the area south of Australia remarkably increase together. High correlations (r${\ge}$0.95) between MSU and reanalyses occur in most global areas, but they are lower (r${\sim}$O.75) over the 20-3ON latitudes, northern America and southern Andes mountains. The first mode of MSU and reanalyses for monthly-mean Ch4 temperature shows annual cycle, and the lower-stratospheric warming due to volcanic eruptions. The analyses near the Korean peninsula show that lower-stratospheric temperature, out of phase with that for troposphere, increases in winter and decreases in summer. In the first mode for anomaly over the tropical Pacific, MSU and reanalyses indicate lower-stratospheric warming due to volcanic eruptions. In the second mode MSU and GEOS present Quasi-Biennial Oscillation (QBO) while NCEP, El Ni${\tilde{n}}$o. Volcanic eruption and QBO have more impact on lower-stratospheric thermal state than El Ni${\tilde{n}}$o. The EOF over the tropical Atlantic is similar to that over the Pacific, except a negligible effect of El Ni${\tilde{n}}$o. This study suggests that intercomparison of satellite data with model reanalyses may estimate relative accuracy of both data.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.24 no.4
• /
• pp.215-228
• /
• 1991

In the East Sea of Korea the vertical structure functions of the temperature field were evaluated and the characteristic thermal zone was classified by the use of the empirical orthogonal function(EOF) method. The East Sea of Korea within the hydrographic lines of 10-107 of the Fisheries Research and Development Agency of Korea(FRDA) can be divided into three thermal regions by the characteristics of the vertical temperature variability. They are the North Korean Cold Current(NKCC) region near the coast which extends parallel to the north-south direction, the Warm-Core(WC) region which dominates almost all the hydrographic stations of the Line 104 of the FRDA and occupies a few stations of the Line-103 and -105 with its axis at the Line 104, and the East Korea Warm Current(EKWC) region which is bisected into the northern and the southern part by the WC region, respectively. Considering the two most important modes, $85.20-98.20\%$ of the total variance of temperature variation are explained in the NKCC region, $85.20-92.90\%$ in the EKWC region, and$85.50-91.70\%$ in the WC region. The first mode has its peak value at the surface with the annual cycle of variation. The spatial pattern of the first mode portrays a coherent vertical variation in the EKWC region and a clear anti-correlation both in the NKCC region and in the WC region where the zero-crossing depths are loom and 200m, respectively. The second mode of the NKCC region is particularly noticeable, haying its peak at loom with coherent vertical variation. To study the time dependency of the vertical structure functions, the extended EOF(EEOF) method was used. The persistence of the first mode is less than 4 months in the study area. The annual variation of the first mode in the NKCC region is different from those in the WC region and in the EKWC region.