GISdevelopment.net ---> AARS ---> ACRS 1989 ---> Agriculture & Soil

Interpretation and applications of the Slar imagery of Shanghai suburbs

Chen Jian, Ding Yi
Shanghal Jiao Tong Univwrsity
Shanghai 200030. China


Abstract
The Ka-band SLAR which was developed by Shanghai Jiao Tong University had Carried out remote sensing files over the suburb countries of Shanghai where are agriculture areas. The authors have chosen some typical state farms and private farm fields as studying object. They summarized various factors to list a detailed table for interpretation of the SLAR imagery of city suburbs. The paper presents a simple and practical vegetation model with the Leaves Area Index (LAI). The paper also describs the vegetation classification and the analysis of land use by the interpretation of SLAR images.

Introduction
The JD-08 Ka-band side looking airborne radar (SLAR) developed by SJTU Marhc and May 1987. The flying regions are indicated by frames in figure 1, including south bank of Chang Ming Island located at the mouth of Young-zi River (frame 1), Chang Xin Island and Heng Sha Island (Frame 2), the seacoast from Jin San Country to Fangxian Country of Hangzhou Bay (Frame 3 and 4). A group of clear SLAR pictures had beenacquired. We lay emphasis on pictures concerning frame 2 and 4. Our research purpose is to approach a general interpretation principle of radar remote sensing pictures over agriculture areas and to examine a vegetation scattering model with radar pictures, thus explaining the application prospect of SLAR images on agriculture.


Figure 1. Location of flying areas in Shanghai suburbs


Interpretation of Slar iamges
Shanghai area is located in front of deltaic plain of Yang-zi River. The lerrian is flat and ri-vers are very dense. Exploitation and use of this area are much earlier. The flying areas of various crops amall and intensively cultivated. The test region described by this paper is from the Shanghai Oil and Chemical Factory on the west to Nan Hui Mouth of Nan Hui Country on the east, covering a seacoast of about 70 Km long, the remote sensing area is about 300 Km2. The area interpreted in this paper is typically selected from xin Huo State Farm of Fen Xian Country including prawn-reising pools, fish pools, cow fields, farmer's houses, cct. The ratio of use is high and reflects comprehensive development is farming, stack raising, fishing and rural industry.

Table 1. The density change of slar imagery with ground influence factors
Land Mass Name Density Description Influence factor for Slar echo
Wheat fields 1.25 Growthing stage, H=29, 4cm, LA I=30 Roughness and dielectric constant of wheat
Green manure (Alfalfa) 1.10 Cover rate=64% for green manure, H30cm, LAI=1.8 other are weeds. Less roughness and dielectric constants
Rape fields 1.35 Cover rate=100%, H=37,. 45cm, LAI=4.6 Higher roughness and dielectric constants
Simple Constructed highway 1.30 Soild surface, with weeds along two sides, 4m width, along by rivers Rivers and roads are specular reflector with rough weeds bands and its dielectric
Farmer's houses 1.46 Dispersing and closing the roads The houses and ground act as a corner refectro
Play ground 1.02 Cement ground surface Specular reflector
Factory and Residential areas 1.59 Bigger Factory buildings and two to three floors houses. The houses and ground act as a corner refector and surrounding matter scatter radar signal
Rivers 0.77 20-40m width, stretching as lien, lush weeds on the banks the water surface act as specular reflector and seeds scatter radars signal
Sea 0.74 Very vast and calm Specular reflector
Fish polls 0.96 Rectangle form, rougher soil raod between two pools Water act as specular reflector and ground scatter radars signal
Swamp 1.54 Hard sand as rock and high moisture The roughness of swamp and higher humidity
Fruit grove 1.19 Not growthing leaves The roughness and dielectric constant of branch
Digging fish pools 1.33 The p[arts are water, the other are soil and breaking stone The refection and scatter of water, soil and stone
Iron Ship 1.58 Anchored on seabeach for breaking up The metal has strong reflection feature
Jetty 1.54 Extending toward to sea, piled with stone very rough The ground has larger roughness and acts as the corner reflector

We measured quantitatively the densities of various surface objects and scene on these SLAR pictures, then processed pictures with image enhancement, edge enhancement, smothing filtering and pseudocolour processing on computer system. In order to further analyse the variation and relationships of different earth image features, we referred to colour infrared photographs and latest maps, carried out practical survey simultaneously. We collected various data regarding land and crops, finally summerized the relationships among variation of radar image density and eart5h influence factors which has general significance listed in table 1.


Figure 2. A reclaimed area of Xin Huo Farm (March 1986)


Figure 3. Interpretation diagram of fig. 2 for land use


Figure 2. is the SLAR picture of reclaimed area of Xin Huo Farm which is acting as research object here. Its scale is 1:50,000. Through the proceduces of field inspection, interpretation and calibration, etc. a land use map is shown as figure 3. The following features are discovered in interpretation:
  1. The concrete roads, playground , sea surface, rivers, fish pools , etc., all of them have mirror reflection and their tones are black. Generally, fish pools are distributed in pieces or squares with clear outlines,. Rivers and canals stretch in lines and are interpreted easily.

  2. The image of roads extend in straight lines. Generally, they are built with soild dug from near Rivers. So they are often along rivers, both are parallel. Because of their narrow width and smooth surface, it is very difficult to interpret them directly from SLAR pictures. But they can be interpreted indirectly from strong return caused by wide grass fields and tree band along roads.

  3. The farmer's houses, factory buildings, bridges, storehouse, etc., have strong return. Their image patterns are clear and cab be easily identified. We discovered that factory buildings and town houses are high and dense, corresponding to bright tones and vulgar outlines.

  4. Plantations are divided into clear strips and has different tones according to different crops, so we can classify different crops or analyse different growing state of same kind of crop.

  5. The boundaries between sea and land are very clearly and the reclaimed lands from the sea are directly perceived. The seabeach is made of sand which is hard like stone with scales caused by seawave and with high humidity. So it appears bright tone in SLAR image.

  6. Targets on the sea surface can be distinguished easier because of the black tone corresponding to the sea surface. In figure 4, at the left, the " " pattern along the coast is the image of ships ancho- red on the seabeach. At the middle, on the bottom, six bright short lines strenching into the sea are jetting, it is a boat hese features are not reflected in colour infared photograph taked in 1984 and the latest maps. By examining, the above interpretation is proved true.

    So it has been fully illustrated that radar images truly reflect variable development and the land use of Lu Chao Harbour on time.

Figure 4. Slar image of Nanhui Estuary in Nanhui country (march 1986)


Expression of vegetation model with leaves area index ( LAI)
Crop output is the function of many variates: Obsorbed nutrition, soid kinds, moisture content and weather condition in growing stage.

The basic energy source of photosynthesis - the sun energy bsorbed crop also is a important parameter for the estimation of output. The crop absorb the sum energy through the interaction between leaves an of crop, can be used to describe the photosynthesis. It can be used in serve as an important parameter.

The backscattering coefficient of vegetation canopies consists the sum of multiple scattering of vegetation and the backscattering soil as well as bare soil. Attema and Ulaby proposed that the vegetation canopy is represented by a " cloud" of water particles, the multiple scattering component between vegetation canopy and soil is neglected. The backscattering component of the canopy is given by

s°can (q) = s° KeJ (q) + s°s (q) ...................................(1) Where the s° KeJ is the backscattering component of the vegetation. The is the backscattering component of bare soil which is extinguhed vegetation. The is the incidence angle.

For Ka-band, the wavelength is very short. The ability of vegetation penetrtation is very poor. We can consider that the backscattering component of vegetation is mainly decided by the scattering component of the upper leaves of plant, therefore


Eq. 2 & 3.

Where A1, B1 and B2, mg are constant at a given frequency, h is the plant height, mg is the volumetric water content of the leaves canopy. is view of this, the formula (2) only depend on two physical quantities and my. It represents the plant moistrure content in a pillar with unit levwel erossection as bottom , in relation to LAI. If the my h is repleased by L(LAI) in formula (3), the result is obtained that.

L(q) = exp (al L) secq ………………………………… (4) where (al is constant for a given plant. L is LAI.

For the particles of different size in equation (2), let T-B m cos. if the equvalent particles density is small, T is increase linearly. If equivalent particles density is bigger, my has saturated, T altend to saturation. This feature can be shown by following function:

T= Al [ 1-exp (-Bl. L/h) ] cos(q …………………………………(5) where Al, Bl, are two constants at a given frequency, incidence angel polarization and the vegetation kinds. L/h correspond with the plant moisture content. Finally, integrating above equation (3) to (5), leads.


Eq. 6 & 7.

Where include two plant parameters LAI and h. Then the coefficients Al, Bl and al can be determined using regrassion analyses.

Crop classification
On the bases of previous research work, we have classified vegetation for figure 2. The table 2 summerized the growing parameters of the three kinds of crops, corresponding image densities ( negative( at interpretation results of them.

Table 2. Partial vegetation parameters, densities of negative and interpretation of three main vegetation
Crap Name LAI   H (cm) Interpreting result Density of negative
Rape 4.6   73.45 Light 1.35
Wheal 3.0   29.40 Grey 1.25
Alifalfa 1.8   30.00 Dark 1.10

Naturally, there are other kinds of vegetations in the test area such as small field of garlic, peas, etc. Because their back scattering are as same as or close to those of the three main vegetations, they appear much less frequently and have much smaller areas, so we put them into three main vegetations.

Conclusions
These years the suburbs of big cities developed rapidlly. The status of land and reclaimation from seacoast, the changes of city boundary and residential areas should be often supervised. Our tests and research work prove that SLR is a kind of convenient, economical and quick instrument for remote sensing. It can be one of general supervisory methods on land use, seabeach variation and river system variation. For the classification of crops, a few main vegetations can be classified on SLAR photographs with one dimension operating way. Its different tones represent the different growing status of the vegetations. If classification of many vegetations or more classification accuracy is needed. The SLAR system must be able to operate on multidimension way, including different incidence angel, different polarization and different waveband. The SLAR must be calibrated. We are undertaking this tasks now. At the same, the remote sensing flying must be divided into many times in different.

References
  • Dign Yi, "The Imaging Machanism of SLAR and its SOme application to the Agriculture Engineering", Master's thesis, Shanghai Jiao Tong Univ. 1988.
  • F.T.Ulaby, R.K. MOore and A.K. Fung, "Microwave Remote Sensing", vol. 2, Radar Remote Sensing and Surface Scattering and Emission Theory, Addison-Wasley Publishing Company, 1982
  • E.P.W. Attema, F.T.Ulaby, " Vegetation Modeled As A Water Cloud", Radio Science, Vol. 13, No. 2, 978, p. 357 - 364.