Application of GOCI Satellite Data to Ocean Modeling

Transcription

1 GOCI Satellite Data and ROMS Ocean Model 1409 Application of GOCI Satellite Data to Ocean Modeling Chang S. Kim, Young-Je Park, Kwang Soon Park, Jae Seol Shim and Hak-Soo Lim Coastal Disaster Research Center, Korea Institute of Ocean Science & Technology, 787 Haeanro, Ansan , Korea Corresponding author Korea Ocean Satellite Center, Korea Institute of Ocean Science & Technology, 787 Haeanro, Ansan , Korea ABSTRACT Kim, C.S., Park, Y.J., Park, K.S., Shim, J.S. and Lim, H.S., Application of GOCI Satellite Data to Ocean Modeling In: Conley, D.C., Masselink, G., Russell, P.E. and O Hare, T.J. (eds.), Proceedings 12 th International Coastal Symposium (Plymouth, England), Journal of Coastal Research, Special Issue No. 65, pp , ISSN This study demonstrates possible application of satellite data to numerical modeling, and vice versa, to improve the quality of the processed information. A new geostationary satellite with the three missions of communication, ocean and meteorology was launched in June The ocean mission part, GOCI (The geostationary Ocean Color Imager), is now in data service of ocean color images at every hour during daytime with a spatial resolution of 500 m. The scanned data covers the north Pacific with focusing on Korean Peninsula and adjacent seas. Using the optical sensors, the GOCI images usually contain the cloud-blocked zone. We investigate a method to acquire the cloud-free images of environment information by using the operational 3-D ocean model, the ROMS. The clear-image data are used as boundary condition to the numerical modeling, and then the model results are used to recover the cloud-covered area. The hourly varying image data are also excellent data for the ocean numerical modeling in terms of temporal and spatial variation of the ocean environment. By close examination of hourly producing images from GOCI, the temporal variation is very useful for operational purpose along with the three-dimensional ocean model such as ROMS. Being in operational mode, the ROMS ocean model produces temporal and structural variation of coastal features by using the compatible GOCI data, or the GOCI images can be improved by using the model results for cloud-free images. This study shows some excellent test cases on various coastal phenomena, such as a river plume of highly turbid waters, coastal upwelling, transport of algal bloom, typhoon tracking and the distribution of surface suspended sediment concentration. A method for the convergence of GOCI data and ocean model results has been introduced. The clouding network system for the convergence of satellite data and ocean model data is a promising method that combines two different media, thus yielding dynamically validated products. ADDITIONAL INDEX WORDS: Geostationary Ocean Color Imager (GOCI), cloud-free image, hourly varying image data, ROMS operational mode, convergence of satellite data with model data. INTRODUCTION The GOCI (Geostationary Ocean Color Imager) has started the data service in May Being geostationary and operating at 36,000km above the Pacific equatorial latitudes, the GOCI provides images that cover the northern half (2,500 km x 2,500 km) of the Pacific Ocean and center on Korean peninsula. The horizontal resolution is approximately 500m. The GOCI scans its images at every hour during daylight (9 A.M. ~ 4 P.M., Korea local time GMT+9) for 8 hours a day. The very unique feature of consecutive images of 1-hour interval is a promising data set for the temporal variation study of oceanic phenomena. The Ocean color sensors have 8 bands (392 nm ~ 905 nm) with wave length ranging from visible to thermal infrared. Using single or multiple combinations of sensing data, the algorithm(gdps) for converting the original scanned data to oceanographic environmental parameters has been developed (Ryu et al., 2012). The application fields using GOCI data for oceanic phenomena are as follows; chlorophyll and biomass, turbidity, typhoon tracking, coastal upwelling, coastal flood inundation, oil spill, wake & vortex string DOI: /SI received 07 December 2012; accepted 06 March Coastal Education & Research Foundation 2013 in lee wave of island, surface currents, surf zone dynamics, temperature & salinity, tidal flat, exposure of sand bars/ridges, wave breaking zone, mixing of different water masses, coastal filament of coastal current/mixing, effluent from coastal areas (power plant/sewage), etc. In this study, we demonstrate a few good applications of GOCI data and ocean numerical modeling to oceanic features using operational three-dimensional ocean model ROMS(Kim et al., 2009; 2011). METHODS The GOCI (Geostationary Ocean Color Imager) The GOCI is an ocean sensing package installed on the COMS satellite of multiple purposes for communication, ocean and meteorology. The first geostationary ocean color imager, GOCI scans the images covering northeastern Asia centered at 130 E and 36 N with horizontal spatial resolution of 500m at the center area. The spectral range from visible to thermal infrared and general configuration of optical sensors are similar to the pre-launched CSCZ, SeaWIFS and MODIS etc., but the GOCI is geostationary one differing from polar orbiting, yielding high resolution of

2 1410 Kim, et al. 2500km. The GOCI scans the images at 1-hour interval for 8 hours in daylight time. Considering the return period of polar orbiting satellite as 2-3days, the GOCI provides the 8 images in time series with sampling time of 1 hour gives a great advantage to study fast varying phenomena in coastal water such as dispersal of turbid water algal bloom, chlorophyll patching etc. The GDPS (GOCI Data Processing System) is a software package produces various levels of ocean analysis data. The conversion algorithm have been developed and calibrated/ validated during last decade (Ryu et al., 2012). Among the useful ocean environmental data, the standard package provides the values for TSS (total suspended solids), chlorophyll concentration, DOM (dissolved organic matter) etc. The optional package provides also the daily products of the surface current vector, underwater visibility. The ocean environmental parameters produced from the GOCI images are fundamentally useful as input conditions to the ROMS modeling. The modeling results are also used for calibration and validation of Satellite-observed data by comparing with field observed data (Choi et al., 2012). Figure 1 shows GOCI-produced hourly varying SSC on 26 April 2012 and monthly mean SSC in April and July of The algorithm for SSC production of GOCI data has been a center state for calibration and validation. Figure 1. Distribution of SSC in the Yellow Sea produced by GOCI band 5 based on the SSC algorithm. The hourly SSC (top) was produced from GOCI images acquired from 09:00 to 16:00 local time (GMT+9) on 26 April The monthly mean SSC distribution (bottom) was derived from GOCI image averaging for whole images in April and July of Yellow Sea ROMS, SWAN, WRF FTP FTP server - U/V/T/S boundary data - Wave boundary data - Atmospheric data Operational system start FTP Data conversion Model simulation web-gis service (05:00, 17:00) (06:00, 18:00) Figure 2. Process of the sediment transport modeling system for the coastal waters of Korea using wave coupled model ROMS- SWAN. The open boundary condition for the hydrodynamics and wave is nested from the Yellow Sea operational model ROMS and SWAN. Atmospheric surface forcing is nested from the Yellow Sea and part of East Sea operational model WRF. horizontal even the whole coverage 16-slots image in 2500km by The ROMS Ocean Model The ROMS model is a three-dimensional ocean model that solves the primitive equations of motion with proper initial and boundary conditions such as tides, winds and other driving forcing. The ROMS is a tool for predicting water temperature, salinity, currents and tides in hydrodynamic module (Haidvogel et al., 2008). It is a common implementation to couple the hydrodynamic module with meteorological module (WRF), wave module (SWAN), sediment transport module, water quality (ROMS-ICM, Kim et al., 2011), etc. There have been many attempts to use the ROMS to solve the environmental problems around Korean waters. Among them, dispersal of sediment in Kyunggi Bay (Kim et al., 2009), water quality prediction (Kim et al., 2011) and river plume effect (Chang et al., 2003) are good examples of well validated modeling efforts. The ROMS model adopts the curvilinear coordinate optimizing well the complex coastline in Yellow Sea. Due to the macro-tidal regime, the dry-wetting boundary scheme is essential for the ROMS. The ROMS model after fully verified and validated for Korean waters is now being used as an operational tool for 72-hours prediction of hydrodynamic and environmental prediction in Korea (Lim et al., 2011, 2013 in this issue). For the sediment transport modeling of the coastal waters of Korea, we uses four different domains with cell size of two domains for western coastal waters (W-1 and W-2) covering ~ E, 34.6 ~ 38.4 N and another two domains for southern coastal waters (S-3 and S-4) covering ~ E, 32.8 ~ 35.8 N (Figure 2). In vertical array, we use 20 levels with significant stretching near the surface and bottom to resolve surface suspended sediment transport and bottom sediment transport. For the tides at open boundary, we use eight major tidal constituents including semi-diurnal tidal constituents (M2, S2, N2, K2) and diurnal tidal constituents (K1, O1, P1, Q1) derived from the regional ocean tide model NAO.99jb with 5 resolution (Matsumoto et al., 2000). The initial hydrodynamic condition of the model is computed with 2 years simulation using monthly mean temperature and salinity data derived from the World Ocean Atlas 2005 (WOA2005) with monthly mean surface forcing derived from the Comprehensive Ocean-Atmosphere Data Set

3 GOCI Satellite Data and ROMS Ocean Model 1411 (COADS). The model system is driven on open boundary condition by the predicted results (05:00 and 17:00 Local Standard Time GMT+9) of another model ROMS with 9 km grid size, in which it has been operational using data assimilation method (Ensemble Kalman Filter) for the Yellow Sea providing wind-driven circulation, temperature and salinity in 72 hour-based prediction. Surface forcing such as wind speed and direction and the heat flux are derived from operational atmospheric model WRF for the yellow sea and part of the East Sea. The SSC or chlorophyll data provided by GOCI product for surface condition can be nudged into ROMS modeling system. Figure 2 shows the process of operational modeling system for the prediction of coastal waters in Korea using wavecurrent coupled model ROMS. The simulated sea surface elevation and currents derived from this sediment transport modeling system was also used for validating the hourly varying GOCI produced SSC images acquired at coastal waters of Mokpo, located at southwest coast of Korea showing temporal variation of sediment movements driven by the tidal cycle along the western coastal waters of Korea Peninsular (Choi et al., 2012). APPLICATION RESULTS Coastal Upwelling Coastal waters move according to the balance of forcing, such as surface winds, tides and other baroclinic processes. Among such movements, coastal upwelling is a phenomenon classically driven by winds that blow parallel and to the right of a coast line in the Northern Hemisphere. Persistent favored winds transport the surface water mass away from the coastline and then bottom waters rise up to the surface to compensate. This feature usually shows cold water temperature along the coastline and higher nutrient concentration along the coast. The southeastern coastal zone of Korea is a favorable area of coastal upwelling that has been reported widely in the literature. Figure 3 shows an example of coastal upwelling with higher concentration of chlorophyll images on 23 September 2011 by GOCI images implies a temporal variation of the higher chlorophyll band moving toward the north-east, following the persistent oceanic current in this area. These snapshots are a good indication of coastal upwelling and transport of coastal waters embedded into an oceanic current system. The chlorophyll distribution produced by GOCI image at 13:30 local time (GMT+9) on 25 September 2012 is also showing coastal upwelling compared with modelderived currents. The subject is a challenging study for precise 3-D numerical modeling using the ROMS modeling system. The simulation for upwelling test is initialized with 10 C temperature difference and forced by southerly wind speed of 10m/s. After 5 days simulation the surface distribution of temperature shows upwelling near the coastal waters (shown in figure 4). River Plume with High Turbidity There are several major rivers along the coast of Yellow Sea such as Yantze, Hwanghe, Han, Young-san rivers which show high discharge rates in summer season due to monsoon climate. The rivers plumes carry highly turbid water that is clearly visible by GOCI. The GOCI images only provide the phenomenon logical feature or temporal variation. To supply at the internal Processes governing the vertical profile and further advection, we Figure 3. Chlorophyll distributions produced by GOCI band 5 based on the chlorophyll algorithm obtained from GOCI images at (a) 12:30 and (b) 13:30 local time on 23 September 2011 and at (c) 13:30 local time (GMT+9) on 25 September (d) Model simulated currents at the same time as for (c). Figure 4.Test run of coastal upwelling driven by a steady southerly wind of 10 m/s. (a) Initial condition of surface temperature and southerly wind of 10 m/s. (b) Initial vertical distribution of temperature with a 10 difference at the latitude of 35.3 N (dotted line). (c) Model-derived surface temperature after 5 days of simulation. (d) Model-derived vertical section of temperature at the latitude of 35.3 N after 5 days of simulation. need the three-dimensional structure of suspended materials. The

4 1412 Kim, et al. many case studies on Sediment dispersal due to seabed mining (Kim et al., 2009), residual circulation in Kyunggi Bay system (Kim et al., 2009b), and river plume in Changjiang river (Chang et al., 2003). Figure 5 shows monthly mean SSC produced by GOCI band 5 based on the SS algorithm in August and September, 2012 and climatologically monthly mean SSS and SSC in August 2003 simulated by ROMS. Figure 6 shows comparison of SSC between produced by GOCI band 5 based on the SS algorithm and simulated by operational coastal modeling system ROMS-SWAN during from flood tide to high water on October, The simulated non-cohesive sediment concentration with grain size of mm is well matched with pattern of the SSC derived from the GOCI image during at flood tide and high water showing temporal variation of surface suspended sediment based on tidal cycle along the coast. Figure 5. Monthly mean SSC in the Changjiang River Estuary produced by GOCI band 5 based on the SS algorithm for August (a) and September (b) Simulated monthly mean surface distribution of salinity (c) and sediment (d) for August Figure 6. Comparison of SSC produced by GOCI band 5 based on the SS algorithm and that simulated by ROMS-SWAN operational coastal modeling system. (a) SSC distribution obtained from the GOCI image acquired at 10:30 (a) and 13:30 (b) local time on 26 October 2011 during flood tide and high water. Simulated SSC overlaid with model-derived current velocity from ROMS-SWAN wave and current coupled model at 11:00 (c), 12:00 (d), 13:00 (e) and 14:00 (f), respectively. model implementation is a tool to compensate the GOCI imagery with internal information in the water column. There have been Typhoon Tracking Figure 7 shows the numerically simulated typhoon track for Typhoon Bolaven (1215), which passed along the west coast of Korea, producing tremendous casualties and damage to property on August 28, The advantage of hourly produced images from GOCI is the possible tracking of typhoon s core moving with time. Correct tracking of typhoon s core will provide a more accurate wind field. Geomorphology of Tidal Flats Numerous tidal flats of wide area, where the water line moves back and forth through the tidal cycle, producing a varying intertidal zone, occur along the west coast of Korea. The exposure or inundation of the tidal flat is very important to the ecosystem, sedimentation of geomorphology change. The texture of sediment and porosity are directly related to the vegetation class and benthic species. Using the SAR data of turbulence intensity, there are many attempts to quantify the tidal flat area (Kim et al., 2011b). However the GOCI products are only for the optical property, hence the vegetation classification could be a tool for quantifying tidal flat dynamics. Using the dry/wet scheme of the ROMS model and the temporal variation of tidal flat can be estimated. The GOCI image and model results are very coherent when suspended sediment concentration around the water line can be identically obtained, an optimized approach would be possible. Convergence of GOCI and Model Data The convergence of GOCI provides various features of ocean environmental features, such as dispersal of turbid water, chlorophyll distributions, tidal flat mapping etc. The ROMS ocean model provides temporal and spatial features of ocean environment in high resolution, as well as in vertical structure. The important point is such that the surface images scanned by GOCI is a dynamic feature strongly linked with internal processes in the water column. Therefore, we need to use the bi-lateral advantages of image of GOCI for surface layer and dynamic structural features from ROMS results. Linking data originated from two different media is not a straight-forward task. In a simple ideal word, the GOCI data in the surface layer can be used as surface boundary condition at time step, and then the implemented ROMS result can be ported to the masked part of GOCI image due to the blocking of clouds. The GOCI system consists of many parts of hardware and software components. The final products of GOCI image comes through GDPS (GOCI Data Production System) to yield the environmental data in practical units. However, the ROMS model runs on a parallel

5 GOCI Satellite Data and ROMS Ocean Model 1413 different systems. We adopt a VPN (virtual private network) architecture to link GOCI data with modeling, or vice versa. The advantage of using the VPN is the random accessibility to each data, as if all were on the same platform. This VPN process, called clouding, is a useful technique for connecting and using distributed data from GOCI and the ROMS model. Figure 7 shows a schematic diagram of the core architecture of a clouding system between multiple platforms. Figure 8 shows an example of convergence of GOCI data with model data. Using Google-Earth, we integrated the cloud cover of Typhoon Bolaven (1215) derived from the GOCI image of 13:30 (local time) on 28 August 2012, with model SST predicted by the operational coastal modeling system when Typhoon Bolaven approached the west coast of Korea. Convergence of the GOCI data and predicted storm surge height is shown in Figure 8. Figure 7. (a) Map of the Typhoon Bolaven (2012) derived from GOCI image at 13:30 on August 28, 2012 (local time). (b) Convergence of GOCI image of Typhoon Bolaven and model SST predicted by operational coastal modeling system when the Typhoon Bolaven approached at the west coast of Korea. (c) Convergence of GOCI Bolaven cloud and storm surge height predicted by operational model ROMS for (c) the Yellow Sea and East Sea at 14:00 on August 28, (d) Storm surge height with high-resolution coastal model nested with coarse grid. Figure 8. Schematic diagram of the core architecture of a clouding system for multiple platforms. processing platform with high performance. Here, we need to obtain GOCI and ROMS model interaction to dialog the data of CONCLUSIONS The convergence of GOCI provides various ocean environmental features, such as the dispersal of turbid water, chlorophyll distributions, and tidal flat mapping etc. GOCI is the first geostationary ocean color imager scanning ocean environmental features every hour for eight consecutive hours during daylight time at a horizontal resolution of 500m. The images are centered at 130 E and 36 N to provide more precise data in the northeastern Asia. The ocean environmental products of GOCI are very useful for monitoring oceanographic phenomena and temporal change by analyzing the consecutive images of hourly interval. However, the surface images produced from GOCI data are insufficient for further investigations of oceanographic processes controlled by the internal dynamic balance. For this reason, we have demonstrated the bilateral use of GOCI and ocean model data to complement the shortcomings of each surface image from GOCI and initial/boundary condition of modeling. The introduction of GOCI data into the ROMS ocean model improves the accuracy of the model results, while the model results can be used to replace the missing data in the cloudmasked area. Monthly surface mean SSS produced from GOCI images averaging for whole month images can also be used to validate the dynamics of sediment movement and simulated sediment transport data based on monthly climate change along the coasts of Korea, China and Japan. The data exchange, based on direct dialog, can be achieved by using a cloud computing system that adopts a VPN network system. Providing optical color imagery, GOCI does not provide sea surface temperature, neither the salinity. In three-dimensional hydrodynamic modeling, SST is an essential variable for understanding the ocean processes and variability of the parameters. To overcome this problem, SST data compiled by using multiple satellite data (e.g., MODIS, SeaWiFS) might be an alternative to allow the utilization of SST in the numerical modeling. In summary, the convergence of GOCI data and threedimensional ocean model results is promising for developing further understanding of complex coastal phenomena, such as river plume dispersal, sediment transport, coastal upwelling, transport of algal bloom and typhoon tracking. ACKNOWLEDGEMENT This research was a part of the project entitled Development of Korea Operational Oceanographic System funded by the Ministry of Land, Transport and Maritime Affairs of Korea. Partial support by the project Satellite Application Techniques for Coastal Ocean Environmental Monitoring (SATCOM) and

6 1414 Kim, et al. Estimation of Storm Surge Inundation and Hazard Mapping : Busan, Masan, Yeosu funded by Korea Institute of Ocean Science & Technology, Korea is greatly appreciated. LITERATURE CITED Booij, N., Ris, R.C. and Holthuijsen, L.J., A thirdgeneration wave model for coastal regions, Part Ⅰ, Model description and validation. Journal of Geophysical Research, 104(C4), pp Cerco, C.F. and Cole, T., Three-dimensional eutrophication model for Chesapeake Bay. Journal of Environmental Engineering, 119(6), pp Chang, P.H. and Isobe, A., A numerical study on the Changjiang diluted water in Yellow and East China Sea. Journal of Geophysical Research, 108(C9): Choi, J.K., Park, Y.J., Ahn, J.H., Lim, H.S., Eom, J. and Ryu, J.H., GOCI, the world s first geostationary ocean color observation satellite, for the monitoring of temporal variability in coastal water turbidity. Journal of Geophysical Research, 117(C09004). Haidvogel, D.B., Arango, H., Budgell, W.P., Cornuelle, B.D., Curchitser, E., Lorenzo, E.D., Fennel, K., Geyer, W.R., Hermann, A.J., Lanerolle, L., Levin, J., Mcwilliams, J.C., Miller, A.J., Moore, A.M., Powell, T.M., Shchepetikin, A.F., Sherwood, C.R., Signell, R.P., Warner, J.C., and Wilkin, J., Ocean forecasting in terrain-following coordinates: formulation and skill assessment of the Regional Ocean Modeling System. Journal of Computational Physics, 227, pp Kim, C.S. and Lim, H.S., Safety criteria on water depth, offshore distance and dredging volume in marine sand mining operation in Kyunggi-Bay, Korea. Journal of Coastal Research, SI50, pp Kim, C.S. and Lim, H.S., Sediment dispersal and deposition due to sand mining in the coastal waters of Korea. Continental Shelf Research, 29, pp Kim, C.S. and Lim, H.S., Kim, J.A. and Kim, S.J., Residual flow and its implication to macro-tidal flats in Kyunggi Bay estuary of Korea. Journal of Coastal Research, SI56, pp Kim, C.S., Lim, H.S. and Cerco, C.F., Three-dimensional water quality modeling for tidal lake and coastal waters with ROMS-ICM. Journal of Coastal Research, SI64, pp Kim, T., Park, J. and Lee, H., Spatial variation of coastal sand ridge features appeared on SAR. Journal of Coastal Research, SI64, pp Lim H.S., Kim, J.A., Kim, C.S. and Park, K.S., SOON: The Saemangeum operational oceanography networks. Journal of Coastal Research, SI64, pp Lim H.S., Chun, Insik, Kim, C.S., Park, K.S., Shim J.S. and Yoon, J.J., High-resolution operational coastal modeling system for the prediction of hydrodynamics in Korea using a wavecurrent coupled model. Journal of Coastal Research, SI65, (in this issue). Matsumoto, K., Takanezawa, T. and Ooe, M., Ocean tide models developed by assimilating TOPEX/POSEIDON altimeter data into hydrodynamical model: A global model and a regional model around Japan. Journal of Oceanography. 55, pp Park, K.S., Lee, J.C., Jun, K.C., Kim, S.I., and Kwon, J.I., Development of an operational storm surge prediction system for Korean coast. Ocean and Polar Research, 31(4), pp Ris. R.C., Booij, N and Holthuijsen, L.H., A thirdgeneration wave model for coastal regions, PartⅠ, Verification. Journal of Geophysical Research, 104(C4), pp, Ryu, J.H., Choi, J.K., Eom J. and Ahn, J.H., Temporal variation in Korean coastal waters using Geostationary Ocean Color Imager. Journal of Coastal Research, SI60, pp Ryu, J.H., Han, H.J., Cho, Seongick, Park, Y.J. and Ahn, Y.H., Overview of Geostationary Ocean Color Imager (GOCI) and GOCI Data Processing System (GDPS). Ocean Science Journal, 47 (3), pp Warner, J.C., Sherwood, C.R., Signell, R.P., Harris, C.K. and Arango, H.G., Development of a three-dimensional, regional, coupled wave, current, and sediment-transport model. Computers & Geosciences, 34, pp