Vacillation of Thermal Fronts in the Bay of Bengal Project oriented interim Report (2008-2009) MASTER OF TECHNOLOGY IN
EARTH SYSTEM SCIENCE AND TECHNOLOGY
By HARVIR SINGH (Roll No.: 07CL6008)
Under the guidance of Dr. MIHIR KUMAR DASH
Centre for Oceans, Rivers, Atmosphere & Land Sciences (CORAL)
CENTRE FOR OCEANS RIVERS ATMOSHPERE AND LAND SCIENCES INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR-721302 INDIA
Introduction The marine environment is characterized by complex interactions between the upper ocean and the atmospheric boundary layer. Oceanic fronts, particularly thermal fronts, and eddies are the most active areas for the air-sea interaction. They are the areas where two water masses, having different properties mix. Their spatial and temporal scales range from meters to hundred of kilometres and seconds to several days. There are several reasons for the atmosphere to be affected by ocean fronts and eddies. (i)
Firstly, as air is blown across an SST gradient, an air–sea temperature difference and air–sea humidity difference is generated. This leads to changes in near-surface stability and surface stress as well as latent and sensible heat.
(ii)
The turbulent fluctuations of heat, moisture and momentum may be transported deeper into the boundary layer by large fronts and eddies resulting the formation of active weather systems.
(iii)
The surface currents of ocean fronts or eddies will impact the relative motion of the air and ocean, acting to change the surface stress, thus affecting the atmosphere as well as feeding back onto the ocean (Kelly et al., 2001; Cornillon and Park, 2001).
Earlier studies showed that emphasized how the response over cool waters (less than about 27 ◦C) is markedly different to the response over warm waters Xie (2004). The warm fronts modulate the amount of deep convection. In contrast, over the cooler seas the response may be more confined to the boundary layer, where the latent and sensible turbulent flux changes at a front alter the boundary layer temperature, and the wind profile adjusts to the changes in mixing and the induced hydrostatic pressure gradients. Earlier studies of the fronts in the Korean sea has been reported by Yoon and Byung (2005). The Indian Ocean is the most dynamic of the world oceans. The circulation pattern reverses with the reversal of the tread wind (see figure 1 from Shenoi et.al.1999). Among different areas of the Indian Ocean, Bay of Bengal (BOB) is the most active and affected by the winds as well as the
remote forcing. It occupies an area of 2,172,000 km2 and bounded by land masses from the three sides and open to the Indian Ocean at its south. Different surface current systems in the BOB as well as the tropical Indian Ocean are shown in the figure 1 for the month of January and July.
Figure 1 : Schematic representation of the circulation (Shenoi et.al.1999) in the Indian Ocean during January (winter monsoon) and July Summer monsoon).The abbreviations are as follows: SC, Somali Current; EC, Equatorial Current; SMC, Summer Monsoon Current; WMC, Winter Monsoon Current; EICC, East India Coastal Current; WICC, West India Coastal Current; SECC, South Equatorial Counter Current; EACC, East African Coastal Current; SEC, South Equatorial Current; LH, Lakshadweep high; LL, Lakshadweep low and GW, Great Whirl
Earlier studies reported low sea surface salinities, particularly in the northern region of the Bay of Bengal as a result of the heavy monsoonal precipitation and large freshwater influxes from the Ganges, Brahmaputra and Irrawaddy Rivers (Shankar et al., 2002). The excess of precipitation over evaporation during the summer monsoon is a significant source of freshwater to the BOB, which together with the large riverine outflows generates highly stable stratification in the upper layers of the northern Bay of Bengal. This barrier persists throughout the late summer and postmonsoon periods, and the associated hydrographic characteristics will have a profound influence on the biological productivity. The main hindrance in the study of physical properties of the BOB is the lack of high frequency and high resolution in situ observation. In the above scenario the observation from space plays an active role in studying the physical properties of this basin. This interim report utilizes the sea surface temperature (SST) observed from NOAA – AVHRR for 2003 – 2005 to study the movement of frontal systems in the BOB.
Data and Methodology
The 4 km AVHRR Pathfinder Version 5.0 SST developed by the University of Miami's Rosenstiel School of Marine and Atmospheric Science (RSMAS) and the NOAA National Oceanographic Data Center (NODC) have been used in this study. The temporal averaged data is available for 5-day, 7-day, 8-day, Monthly, and Yearly periods. These data have the improvement over the original 9 km Pathfinder SST data set as they include a more accurate, consistent land mask, higher spatial resolution, and inclusion of sea ice information. Additional improvements including better flagging of aerosol-contaminated retrievals and the provision of wind and aerosol ancillary data will be implemented in a future Version 6 reprocessing. This data is available from the http://data.nodc.noaa.gov/pathfinder/Version5.0_NOAA17 in the Hierarchical Data Format (HDF) format. The data file contains the seven parameters listed below. 1. "All-pixel" SST: All-pixel SST files contain values for each pixel location, including those contaminated with clouds or other sources of error. The Overall Quality Flag values
may be used to filter out these unwanted values. The SST value in each pixel location is an average of the highest quality AVHRR Global Area Coverage (GAC) observations available in each roughly 4 km bin. 2. First-guess SST: The Pathfinder algorithm uses a first guess SST based on the Reynolds Optimally Interpolated SST (OISST), Version 2 product. The OISST V2 is also used in the quality control procedures. 3. Number of Observations: This parameter indicates the number of AVHRR GAC observations falling in each approximately 4 km bin. 4. Standard Deviation: This is the standard deviation of the observations in each 4 km bin. 5. Overall Quality Flag: The overall quality flag is a relative assignment of SST quality based on a hierarchical suite of tests. The Quality Flag varies from 0 to 7, with 0 being the lowest quality and 7 the highest. 6. Mask 1 :These files contains a mask code, which along with Mask 2, can be to determine the tests in the hierarchical suite that were passed or failed, resulting in the Overall Quality Flag. 7. Mask 2: These files contains a mask code, which along with Mask 1, can be to determine the tests in the hierarchical suite that were passed or failed, resulting in the Overall Quality Flag. Additionally, in the Version 6.0 data planned for future production, wind speeds and aerosol content will also be provided. The data are all stored in Hierarchical Data Format Version 4 (HDF 4), using the Scientific Data Set (SDS) model. HDF-SDS is a self-describing format capable of storing multiple layers of data as well as metadata describing the file contents. Each of the seven parameter files listed above contains a mapped array with 8192 elements in longitude and 4096 in latitude, plus a vector of length 8192 identifying the longitudes and a vector with 4096 values indicating the latitudes. There are also global tags describing the entire contents as well as tags describing each of the 2 vectors and 1 array. The seven parameters are stored either as 8-bit or 16-bit unsigned integers which may be converted linearly (y = mx + c) to geophysical units using a scale (i.e., slope=m) and offset (i.e., intercept=c).
In this study uses the 5-day averaged optimally interpolated SST available in the HDF data file. The HDF file contains the data from the whole globe. The SST for the study region from 5oN to 25o N and 75o E to 100o E is extracted from the global data using the HDF library and the source code developed indigenously. Sample figures of SST for the year 2004 and 2005 are shown in the figure 2. From the SST image frontal edges have been generated using the image processing technique. The details of the technique are described above.
Edge Detection Techniques
An edge is defined as a local concept that based on a measure of Gray level discontinuity at a point. In image processing, the locations of the edges are identified by employing templates that respond to the first or second derivative of gray-scale intensity in the neighborhood of each pixel. There are six different methods for finding the edge. They are
1.
Sobel
2.
Prewitt
3.
Robert
4.
Laplacian of Gaussian
5.
Zero Crossing
6.
Canny
The Sobel method finds edges calculating the gradient in the gray level of the image. It returns edges at those points where the maximum gradient exist in the image. The details of this method are described below. The Prewitt method finds edges using the Prewitt approximation to the derivative, which is different from that of Sobel. The Laplacian of Gaussian method finds edges by looking at zero crossings after applying a Gaussian filter to the image. The Canny method finds edge by looking for local maxima in the gradient of gray level of the image. The gradient is calculated using the derivative of a Gaussian filter. The method uses two thresholds, to detect
Jan 06-10, 2004
Jan 21-25, 2004
Jan 31-Fab-04, 2004
Jan 06-10, 2005
Jan 21-25, 2005
Jan 31-Fab-04, 2004
Figure 2: Optimally interpolated sea surface temperature over the Bay of Bengal for 2004 and 2005 January as observed from the NOAA – AVHRR
strong and weak edges, and includes the weak edges in the output only if they are connected to strong edges.
Sobel Edge Detection Method SST Fronts (SF) are the regions, where isotherm are denser than other regions, i.e. steep gradient appears in SST pixels. Sobel Filtering, a method to detect edge using SST gradient, can obtain clean results in noisy parts of images because weight of filter is high. Sobel Edge Detection Method used the linear differential equation. ⎛ ∂f ⎞ ∇f ( x , y ) = i ⎜ ⎟ + ⎝ ∂x ⎠
∇f ( x , y )
Where
=
⎡G x ⎤ ⎢G ⎥ ⎣ y⎦
⎛ ∂f ⎞ j ⎜⎜ ⎟⎟ ⎝ ∂y ⎠
=
(1)
⎡ ∂f ⎤ ⎢ ∂x ⎥ ⎢ ⎥ ⎢ ∂f ⎥ ⎢⎣ ∂y ⎥⎦
⎛ ∂f ⎞ Gx = ⎜ ⎟ ⎝ ∂x ⎠
Gradient magnitude:
(2)
and
⎛ ∂f ⎞ G y = ⎜⎜ ⎟⎟ ⎝ ∂y ⎠
∇f ( x, y ) = mag [∇f ( x, y )] =
⎛ Gy Gradient Orientation Angle: α ( x, y ) = tan −1 ⎜⎜ ⎝ Gx
[G
2 x
+ G y2
]
1
2
⎞ ⎟⎟ ⎠
A 3 X 3 matrix is selected starting from the (1,1) pixel of the image(see figure 3). Let the gray level values are given by Z1 to Z9 as shown in the figure 3. Gx = (Z7 + 2Z8 + Z9) – (Z1 +2 Z2 + Z3) Gy = (Z3 + 2Z6 + Z9) – (Z1 + 2Z4 + Z7) Where Z’s are gray level
Z1
Z2
Z3
Z4
Z5
Z6
Z7
Z8
Z9
Figure 3: showing the arrangement of the gray levels in the 3 x 3 matrix of the image.
Applying Sobel Mask, magnitude of the gradient and orientation angle was calculated and determined edge when gradient magnitude exceeds threshold. Edge Detection Sobel Filtering
-1
-2
-1
-1
0
1
0
0
0
-2
0
2
1
2
1
-1
0
1
X direction Mask
Y direction Mask
Figure 4: Sobel mask in x and y direction
is shown in the figure 4. Image processing tools from MATLAB and the ERDAS are used to generate the edges from the SST images. Sample of the edge images over the Bay of Bengal for 2004 and 2005 January are shown in figure 5.
Interim Results We have divided our study period into two parts.(i) The pre-monsoon period (March-April-May) (ii) the post-monsoon period (November-December-January-February). This has been done because the east India coastal current reverses its direction during pre-monsoon to post-monsoon period (see figure 1). During the winter season (December± February), the low SST is seen in the head Bay. During January head Bay shows a SST as low as 23°C. Earlier study showed lowest SST (25°C) at the head of the Bay in January (Murty et al., 1998). During both the periods the frontal vacillation is seen both along the west and east coast of the basin. As a sample we have shown the movement of fronts during January 2004 and 2005. During second week of January 2004 the concavity towards equator developed off Andhra Pradesh (see line AA’ in figure 5). During subsequent weeks the concavity increase and shifted in the south-east direction. The east side of the basin shows the movement of fronts in the north-east direction (see line BB’ in figure 5). The variation in the position of the fronts from year to year is clearly seen in the years 2004
A
A’
B’
A
A
B’
A
B’
B
A’
B’
A
B
B’
B
A’
B
A’
A’
A
B
A’
B
B’
Figure 5: Position of sea surface temperature fronts over the Bay of Bengal for 2004 and 2005
January generated using Sobel Method.
and 2005. Year 2005 shows more intrusion of warm equatorial water to the basin at its east side compared to year 2004. The central bay shows that the equator-ward motions of the fronts are more during January of the year 2005 compared to that of 2004. This may be due to the effect of the Kelvin wave traveling from the equator towards the pole. The pattern of the frontal movement matches well with that BOB surface flow pattern as shown by Murty et al. (1998).
Work to be Done in the Spring Semester
•
The movement of the fronts to be tracked using the image processing technique
•
The effect of winds on the frontal movement will be studied.
•
The remote effect on the frontal movement is to be studied.
References:
Cornillon, P., K. A. Park (2001); Warm core ring velocities inferred from NSCAT, Geophys. Res. Lett., 28, 575–578 Kelly, K.A., S. Dickinson, M.J. McPhaden, G.C. Johnson (2001); Ocean currents evident in satellite wind data; Geophys. Res. Lett.; 28, 2469–2472 Xie, S. P. (2004); Satellite observations of cool ocean–atmosphere interaction; Bull. Am. Meteor. Soc.; 85, 195–208 Shenoi. S, P. Saji and Almeida (1999); Near-surface circulation and kinetic energy in the tropical Indian Ocean derived from Lagrangian drifters. Journal of Marine Research, 57, 885–907 Shankar. D, P. N. Vinayachandran, A.S. Unnikrishnan (2002); The monsoon currents in the north Indian Ocean, Progress in Oceanography, 52, 63–70 Hong-Joo Yoon and Hye-Kyung Byung (2005), Temporal and spatial analysis of SST and Ocean Fronts in the Korean Seas by NOAA/AVHRR, Geosciences and Remote Sensing Society IEEE, 2632 – 2636 V. S. N. Murty, B. Subrahmayam, M, L. V. Gangadhara Rao and G. V. Reddy (1998), Seasonal variation of sea surface temperature in the Bay of Bengal during 1992 as derived from NOAAAVHRR SST data, Int. J. Remote Sensing, 19, 2361- 2372