global and regional applications to make forecasts of the sea state, which can be used for many applications such as ship routing and offshore activities, and for the validation and interpretation of satellite observations. The relationship between wave model and satellite remote sensing becomes closer with every passing day. Satellite data were used to run the model as well as to validate the model forecast. Data from QuikSCAT and Topex was used to derive and validate the WAM model.
The Sixteenth Workshop of OMISAR Satellite Data Application for Wave Modelling over the EAST Sea MSc MT.Luong Van Viet, Eng Bui Chi Nam, BSc Le Anh Tuan Sub-institute of Hydro-Meteorology and Environment of South Vietnam E-mail Address: sihymete@hcm.fpt.vn ABSTRACT Satellite observations (TOPEX Altimeter, QuikSCAT Scatterometer and Synthetic Aperture Radar (SAR)) were used as input data for wave models and also to validate wave model forecasts In this study, we made runs of a WAM model for the whole of year 2001 in order to highlight the ability of the WAM model to make accurate hindcasts of wave patterns using remotely sensed data for the model input To derive the wave model for the above period, satellite wind data of QuikSCAT Scatterometer was processed and analyzed The comparison of hindcast model parameters was made with Topex radar altimeter derived parameters For the validation, in-situ data collected at Bach Ho Oil Rig in the East Sea (South China Sea) was utilized The study results show adequate agreement between the WAM model derived parameters and the validation data indicating that the WAM model is suitable for hindcasting/forecasting wave dynamics INTRODUCTION At present, the WAM model is used operationally in global and regional applications to make forecasts of the sea state, which can be used for many applications such as ship routing and offshore activities, and for the validation and interpretation of satellite observations coastlines by using wind data in numerical weather and wave-prediction models and to improve storm warning and monitoring The relationship between wave model and satellite remote sensing becomes closer with every passing day Satellite data were used to run the model as well as to validate the model forecast Data from QuikSCAT and Topex was used to derive and validate the WAM model The Ocean Topography Experiment (TOPEX/Poseidon)8 is a cooperative project between the United States and France to develop and operate an advanced satellite system to provide global sea level measurements with unprecedented accuracy On August 10, 1992, TOPEX/Poseidon was launched into its orbit 1336 kilometers above the sea surface The sea level data from TOPEX/Poseidon is being used to determine global ocean circulation and to increase the knowledge of the interaction between the oceans and the atmosphere SATELLITE DATA 2.1 Surface Wind Data from QuikSCAT NASA's Quick Scatterometer (QuikSCAT)7, was launched on 6/19/99 Its orbit is Sun-synchronous, 803 km height and has a 98.6° inclination The SeaWinds instrument (sensor) on the QuikSCAT satellite is a specialized microwave radar that measures sea surface wind speed and direction under all weather and cloud conditions The basic purpose of the scatterometer is to provide sea surface wind speed and direction data It measures the backscattered power (σo) of a signal transmitted by the radar onboard the satellite and returned from the ocean surface The surface winds are obtained from the σo values by making use of empirical relationships relating to wind and σo5 SeaWinds uses a rotating dish antenna with two spot beams that sweep in a circular pattern The antenna radiates microwave pulses at a frequency of 13.4 Gigahertz across broad regions on Earth's surface The instrument will collect data over ocean, land, and ice in a continuous, 1,800-kilometer-wide band, making approximately 400,000 measurements and covering 90% of Earth's surface in one day Its operational objectives are to improve weather forecasts near 2.1 Significant Wave Height from TOPEX Altimeter Sensor The TOPEX altimeter is a dual frequency radar instrument which draws upon a long heritage of singlefrequency altimeters extending back to SeaSat The primary channel for the altimeter is Ku-band (13.6 GHz), and the secondary channel is C-band (5.3 GHz) Inclusion of the secondary channel allows correction for propagation delays in the ionosphere, reducing a significant error source in the measurement The pulse repetition frequency is approximately 4500 Hz for the Ku-band and 1200 Hz for the C-band The antenna beam width is approximately 1.1 degrees for Ku-band and 2.6 degrees for C-band The primary objective of the radar altimeter is to provide sea surface topography and significant wave height With the analysis of the shape of the returned pulse and normalized radar backscattering crosssection, it is possible to derive significant wave height (Hs) and surface wind speed from altimeter data5 EXPERIMENTS 4-1 Satellite Data Application for Wave Modelling over the EAST Sea In these experiments, wind data from QuikSCAT scatterometer was used as the input for the WAM model The model output was tested using data derived from the Topex altimeter as well as in-situ wave height data taken from the Bach Ho Oil Rig in the East Sea Wind data from Bach Ho was also utilized to test the accuracy of the QuikSCAT satellite data prior to using it for the WAM model www.ngdc.noaa.gov) It is a worldwide bathymetry data set with x resolution The model was initialized with cold start, which means that it has been assumed that no waves are present at the first time step of the runs Due to the cold start runs, the model takes some time to get stabilized Hence, the derived wave parameters may not be quite accurate at the initial period The foundation of the WAM model is based on the wave spectrum1; it is expressed as the energy conservation equation ∂E + ∇.(C g E ) = S in + S ds + S nl ∂t where E is the energy density spectrum, E = E(f,θ,ϕ,λ,t) with respect to frequency f and direction θ as a function of latitude ϕ, longitude λ and time t, ∇ (Cg E) is the divergence of energy flux S in, Sds, Snl is the source represented as a superposition of the wind input, white capping dissipation and non-linear transfer respectively Model runs in this study have been made for the area 00N-300N and 1000E-1300E, the spectral grid was chosen as 25 frequencies and 12 directions, the computation grid was chosen as 0.250 x 0.250, the time step for model integration is set to 20 and the model output is stored at every six hours The wind data input for the WAM models was derived from QuikSCAT level The Level data were obtained from the Direction Interval Retrieval with Threshold Nudging (DIRTH), wind vector solutions contained in the QuikSCAT and are being provided on an approximately 0.25° x 0.25° global grid The Scatterometer sensor does not cover the entire study region in one day Therefore gaps in the data must be interpolated A bi-cubic interpolation scheme was applied to generate wind data for each model grid point Wind field from QuikSCAT after filling data gaps is showed below Fig.2: Significant wave height of WAM model (12Z Dec 14, 2001) WAM model output includes wind speed and direction, friction velocity, significant wave height, wave directions, wave period, 2D wave spectrum, swell height, swell period, swell direction and swell spectrum RESULTS AND DISCUSSIONS In this study, various experiments with the wave model were conducted Comparisons of model derived parameters were made with the available in-situ data of Bach Ho Oil Rig and with the TOPEX altimeter derived wave parameters As wind is major input parameter in deriving the wave model, an initial comparison was made between the input wind data from QuikSCAT and the available in-situ data 3.1 Comparison of QuikSCAT and In-situ wind speeds The Bach Ho Oil Rig data is available for all months from January to December 2001 and is collected four times a day The correlation coefficient (R) between wind speed from QuikSCAT and oil rig data for all the months of 2001 is presented in table Table The correlation coefficient between wind speed from QuikSCAT and oil rig data, 2001 Fig 1: Wind vector from QuikSCAT during the typhoon Ling Ling The bathymetry data for the model was derived from the National Geographic Data Centre (NGDC, 4-2 Mont h Jan Feb Mar Apr May Jun R 0.88 0.91 0.85 0.83 0.79 0.81 Mont h Jul Aug Sep Oct Nov Dec R 0.77 0.80 0.80 0.86 0.89 0.92 The Sixteenth Workshop of OMISAR The results indicate a sufficiently high correlation coefficient between wind speed derived from QuikSCAT and the oil rig data The highest correlation occurs in the months of the winter monsoon, the lowest correlation is in the summer monsoon The reason for this may relate to the observed limit of QuikSCAT (Vmax = 30m/s) Tropical cyclones were very active in the summer monsoon of 2001 (13/15 tropical cyclones) with high wind speeds, especially in July where four typhoons and one tropical cyclone occurred on EAST SEA QuikSCAT was unable to observe these high wind speeds, thus resulting in less accurate data during the summer monsoonal months we can be sure that the wind data used in this study to input into the model is highly accurate Table The correlation coefficient between Hs from Model and TOPEX over EAST SEA, 2001 Month Jan Feb Mar Apr May Jun R 0.95 0.92 0.90 0.85 0.84 0.89 Month Jul Aug Sep Oct Nov Dec R 0.93 0.84 0.86 0.89 0.94 0.89 3.2 Comparison Hs from Model and TOPEX The significant wave height provided by Topexaltimeter is characterized by very high accuracy data and is ideal for checking the model output For comparison of Topex along track data for significant wave heights with the model, Topex data were averaged over the model area within the model grid size Spatial interval was kept at ±0.250 from the model grid point, the time interval was chosen at ±3 hrs and maximum number of points available in a grid is approximately Significant wave height data obtained with the WAM model was compared with the Topex-altimeter observations WAM model runs were made for the area from 00N-1000E and 300N-1300E This area was assumed to represent a closed sea In real situations waves may travel inside the model area from very long distances that may influence the wave hindcast near spatial boundaries Hence, the model hindcast near the spatial boundary may have errors To avoid this effect, the area for comparisons of the model was smaller than the total extent of the model area In summer, the area for comparison was chosen to be 10 0N-250N, 1000E1200E because monsoonal winds are south westerly and hence model output maybe not correct for regions below 100N Similarly, in the winter monsoon, the area chosen for comparison was 50N-200N, 1000E-1150E because winds are north easterly hence model output were discarded for regions north of 200N or east of 1150E Model runs for the whole year were made from cold start It was seen from earlier experiments that wave model needs approximately days spin-up time hence first three days hindcast of wave model were discarded for the comparison with in-situ and Topex altimeter data The results of comparisons are shown in Table 2, Fig.3 below The correlation coefficient between Hs from the model and TOPEX was significantly high, ranging from 0.84 to 0.95; the highest correlation is in the winter monsoon This comparison for all months of 2001 proves that the WAM model is suitable to hindcast/forecast wave height over the East Sea Hence Fig 3: Comparison Hs of WAM model and Topex (data of July, 2001) 3.3 Comparison of WAM model results with in-situ data Wave data from Bach Ho Oil Rig is observed four times a day, giving more than one hundred samples for model comparison per month Wave data of Bach Ho includes significant wave height and wave period Results of the comparisons are presented in Table and Table The correlation coefficient between wave height of WAM and Bach Ho data is significantly high, in the range of 0.82 to 0.94 However for wave period the correlation was less (R from 0.68 to 0.74) Table The correlation coefficient between Hs by model and oil rig, 2001 Month Jan Feb Mar Apr May Jun R 0.92 0.88 0.88 0.82 0.85 0.85 Month Jul Aug Sep Oct Nov Dec R 0.90 0.83 0.83 0.87 0.92 0.94 Table The correlation coefficient between wave period by model and oil rig, 2001 Month Jan Feb Mar Apr May Jun 4-3 Satellite Data Application for Wave Modelling over the EAST Sea R 0.74 0.73 0.72 0.70 0.70 0.70 Month Jul Aug Sep Oct Nov Dec R 0.74 0.68 0.70 0.71 0.74 0.71 enable the confirmation of the ability of wave models to forecast wave dynamics in the East Sea region in any weather conditions REFERENCES In addition to the scatter plot, time variation of model derived wave height and in-situ data is also displayed (Fig 4) Although the time series for comparison is short and there is only one station to access, the results confirm a very high correlation between the WAM model and in-situ data Heinz Gunther, Susannne Hasselmann and P.A.E.M Janssen,1988: The WAM model cyclone 4, user manual, ECMWF/GKSS, MPI f.Met, KNMI I.R Young, 1999: Wind generated ocean waves, Elsevier, pp.45-80, pp 208-224 Komen,1994: Dynamics and Modeling of ocean waves, Cambridge Univ press Raj Kumar, Abhijit Sarkar, V K Agarwal and Vihang Bhatt, B Prasad Kumar and S K Dube, 2000, Ocean wave model: Sensitivity experiments, Proceedings of PORSEC-2000, Fifth Pacific Ocean Remote Sensing Conference, Dec 5-8, 2000, Vol II, 801-803 Stanley Q.kidder and Thomass H.Vonder Haar, 1995: Satellite Meteorology, Academic Press, pp.87-141, pp.331-349 Fig.4: Comparison significant wave height of model and oil rig (data of December, 2001) Vihang Bhatt, A Sarkar, Raj Kumar, Sujit Basu and V K Agarwal, Impact of IRS-P4 MSMR data on analysed winds and ocean wave prediction in the Indian Seas, Proceeding METOC-2004, 199-205 CONCLUSIONS http//winds.jpl.nasa.gov/missions/quikscat/index.cfm The comparisons between wave model derived parameters, the Bach Ho Oil Rig data and the Topexaltimeter data confirm that the WAM model is ideal for hindcast of significant wave height in the East Sea Comparisons of model predicted significant wave height and in-situ observations are well in agreement The model is unable to reproduce small scale variations observed in significant wave height This is due to the frequency of wind input was once a day If more frequent wind data is available then it may be possible to observe small scale variations in the model wave field http://topex-www.jpl.nasa.gov/ mission/topex.htm The model is able to reproduce large variations in significant wave height observed by Topex altimeter for almost all months of 2001, although the correlation in the summer monsoon was less This is due to the characteristic of wind field in the summer (the active cyclone season), and the ability of QuikSCAT to be able to observe high speed and highly variable winds This may also be the reason for the low correlation coefficients between wind speed of QuikSCAT and insitu data and between Hs of the model and the in-situ data for the summer monsoon The analysis of WAM model in this study is limited For increased accuracy, wind data input for the model needs to be increased Model derived parameters can be validated using in-situ and remote sensing data products as detail presentation above Also analysis across a greater time scale (several years) will 4-4 .. .Satellite Data Application for Wave Modelling over the EAST Sea In these experiments, wind data from QuikSCAT scatterometer was used as the input for the WAM model The model output... 0.94 Table The correlation coefficient between wave period by model and oil rig, 2001 Month Jan Feb Mar Apr May Jun 4-3 Satellite Data Application for Wave Modelling over the EAST Sea R 0.74... using data derived from the Topex altimeter as well as in-situ wave height data taken from the Bach Ho Oil Rig in the East Sea Wind data from Bach Ho was also utilized to test the accuracy of the