Automatic System for Classification of Precipitation Cells #

: This paper presents an automatic system for classification of the precipitations cells, conceived around a graphic interface. This interface is based on the fractal geometry and particularly on fractal dimension and the fractal lacunarity. We have initially analyzed these two parameters and we showed that they can be useful as discriminating parameters. Then, we developed a graphical interface which makes possible to identify in real time the type of cells. This tool was tested on different areas from the earth and showed its efficiency whatever the studied site. This system can be used in weather radar for the improvement of the precipitations estimations and in telecommunication for the correction of the signal for the microwave links.


Introduction
Precipitations are generally considered to be two clearly distinguishable types, stratiform and convective (A. Tokey  Stratiform precipitations fall generally from nimbostratus clouds, and are of low quantities but can persists for hours, leading to a significant rain accumulation. In the other hand, convective precipitations are caused by cumulus and cumulonimbus clouds (R. A. Houze, Jr.; 1993), (A. Tokey and D. A. Short; 1996), they are characterized by strong narrow updraft, i.e. convection, where raindrops can grow rapidly. The resulting convective rainfall is intense, relatively short in duration, and highly fluctuating (F. Tridon and al;2010). This paper aims to make a difference between convective and stratiform cells in radar images, using a graphical application programmed under MATLAB called interface "GUI". This approach is based on the fractal geometry and particularly on fractal dimension and fractal lacunarity.

Data Base
We considered in this study four different weather radar sites namely, Dakar (Senegal), Bordeaux (France), Melbourne (Florida, USA) and State College PA (Pennsylvania, USA).

Region of Dakar
Dakar is the capital of the Republic of Senegal. It is located at the western extremity of Africa, on the narrow peninsula of Cape Verde. Being in a tropical semi-desert, Dakar has a microclimate of coastal type, influenced by the monsoon trade winds and sea. The hot wet season extends from June to October with temperatures of about 27°C and a peak of precipitations in August (250mm). The radar of Dakar is installed on a tower of 30m above the sea level at Yoff Airport, in the east of Dakar. Its geographical coordinates are 14°44' North latitude and 17°29' West longitude. It is operated by both the National Meteorological Office of Senegal and ASECNA (Association for the Safety of Air Navigation). This radar works only during storm periods. It also has a chain SANAGA. The average intensity of rainfalls varies from 300mm to 1500mm, collected from Saint Louis to Cape Skirring (A. Nzeukou and H. Sauvageot;2002). The technical characteristics of the weather radar of Dakar are given in Table 1.

Region of Bordeaux
The city of Bordeaux is located in the south western of France. The Bordeaux radar is located in the Bordeaux-Merignac airport, whose geographical coordinates are 44°49'54'' North latitude and 0°41'30'' West longitude. It is one of 24 metropolitan France radars. (SANAGA) allows processing and data storage (H. Sauvageot, G. Despaux; 1990). The radar is part of the French network (ARAMIS) managed by Météo-France. The area is almost flat. Table 2 gives the technical characteristics of the radar of Bordeaux.

Region of Melbourne
Melbourne is a city of the United States of America located on the east coast of Florida. It is situated to the east of Orlando, and south of Cape Canaveral. It has a humid subtropical climate, typical of the Gulf and South Atlantic states. The radar used is Nexrad WSR-88D type, which works on the coherent Doppler principle. The geographical position of the radar of Melbourne is 28,109 ° N, 80,650 ° W. It allows to observe the precipitation in the oceanic, littoral and terrestrial part (D. B. Wolff and al;. The Melbourne climate is subtropical, with hot and humid summers and cool winters. The technical characteristics of the weather radar of Melbourne are given in Table 3.  Table 4.

Fractal Dimension
The

Fractal Lacunarity
Lacunarity is another concept introduced by Mandelbrot (B.B. Mandelbrot;1983), in an effort to quantize the nature of gaps in texture images. It is inspiring from the Latin (lacuna) origin of the english word ''lake'', defines a fractal as being ''lacunar'', when its gaps tend to be large. If a fractal has large gaps or holes, it has high lacunarity (K. I. Kilic Where: () sL = the average of masses per box, noted μ ; 2 s = variance of masses per box, noted σ 2 .

Result and Discussion
As an illustration, we present the results for the region of Dakar (Senegal).
( Figure.1) shows an image of convective and stratiform cells, taken from Dakar radar, in July 12, 2011, at 08h a.m. fractal dimension depending on the size. The type of cells is shown in the part "result and discussion-Fractal dimension" after having clicked on the button "cells type" Primary, we select the images to be treated. If the precipitation is composed of only one type of cells, the interface makes possible to identify the precipitation type without ambiguity. On the other hand, if the two types of cells are present, we must proceed by thresholding, by clicking on one of the choices given in the menu-bar "thresholding". The choice is made depending on the type of cell we want to highlight and on the studied site. After applying thresholding on the image presented in (Figure.1), we get the thresholded image as shown in (Figure.2). Proposed MethodologyData base In the present work, tasks like image acquisition, preprocessing, feature extraction, classification are carried out. The classification tree is given in (Figure.3). The detailed block diagram of adopted methodology is shown in (Figure.4). After choosing the mode of thresholding, we will select the type of treatment we want to apply to these images. We have two types of treatments, "fractal dimension" and "fractal lacunarity." For this, we go to the frame "type of processing" and we click on the appropriate button for our choice.

Fractal Dimension
( Figure.3 & 4) show that when we click on the button "Fractal Dimension", we obtain a straight curve which represents the fractal dimension depending on the size. The type of cells is shown in the part "result and discussion-Fractal dimension" after having clicked on the button "cells type".

Fractal Lacunarity
After clicking on the button "Fractal Lacunarity" in the part "type of processing", we obtain a power curve which represents the fractal lacunarity depending on the size. When we click on the button "cells type" in the part "result and discussion-Fractal Lacunarity", we see the type of the cells studied as it is shown in (Figure.5 & 6).  To classify the precipitations in convective and stratiform cells in real time, the GUI uses the concept of fractal lacunarity on comparing values of homogeneity factors "a" obtained in our process and those calculated in the previous work and inserted in our GUI. the algorithms used in this work are implemented using MATLAB 7.0. We now that the parameter "a" and the lacunarity value of convective cells are greater than that of stratiform ones, whatever the study region (N. Azzaz and B. Haddad;. This is due to the fact that convective cells are less homogeneous than stratiform cells.

Conclusion
In conclusion, our study shows the importance of our graphical interface GUI to facilitate anyone the classification of the precipitation in two types, convective and stratiform, in real time.
The GUI uses the fractal dimension and the fractal lacunarity as powerful discriminators between stratiform cells and convective cells.
This automatic system, tested on areas where prevail different climates, contributes to the improvement of the precipitations estimations through the relation Z-R for the weather radars, or to avoid the natural disasters caused especially by the violent storms. It can also be used in telecommunication for the correction of the signal attenuation for the microwave links. This interface can be inserted without any difficulty on any weather radar.