| Soil salinization is one major type of desertification and land degradation in the world nowadays, because of irrational human activities and vulnerable ecological environment in arid and semi-arid areas. About 9.52×108 km2 (7.26%) of land suffer from salinization to some distant. Soil salinization has become a worldwide problem. Today, land resources are extremely scarce, food security problem sweeping all the corners of the world, and the security of agriculture land has become an extremely important issue. In China, there are a large number of agriculture land and their back-up resources subject to different degrees of impact of salinization. Agriculture land and their back-up resources are threat seriously. So, the protection of arable land and its resources are urgent. Therefore, the understanding of the status of soil salinization has become an inevitable problem. With a large number of saline-alkali soil, it is difficult to get relevant information of saline-alkali soil fleely and accurately for the use of traditional techniques. Because of real-time, rapid, and the macro characteristics, remote sensing can acquire the scope of the distribution, change and trends of saline-alkali soil, and has become an important means of extracting information. Because of certain limitations of visible light and infrared remote sensing, remote sensing technology become the new technologies and means to extract the information of saline-alkali rapidly and large-scale, which can work all the day, under any weather, as well as such characteristics of penetrating.In this paper, we use the latest radar data, Canadian RADARSAT-2 data; take Ugan-kuqa Oasis for example, to research how to extract the information of saline soil. First of all, we obtain quad-polarized backscatters coefficient image by processing the quad-polarization radar image acquire in September 14, 2008, and then acquire the radar backscatters coefficient of samples, via GPS and the quad-polarization backscatters coefficients image. And then analysis relations between the radar backscatters coefficients of the samples and the contents of salt in the surface layer of soil. Then we find that it is difficult to differentiate various soils with various salt contents, via the origin radar backscatters coefficients. Therefore, we analyzed polarization combination of four polarization data; find out one polarization combination for identifying different soil with different salt contents. Of any the combination of polarizations, we found that salt polarization index PSI (that is (HV2-HH2) / (HV2 + HH2)) can reflect the amount of the salt contents in the surface layer of the soil. That is the more serious of salinization of the soil, larger the SPI will be. So it can be use as automatic classification data by computer. And other combinations of polarizations can be use to differ some, but not all kinds of salt lick more or less. Therefore, we should make use of these combinations of polarization data to differ different salt lick.And then, via processing the dual-polarization image, we get the dual-polarization backscatters coefficients image, and calculate the SPI based on this image, and then, we classify the originate image, the combination image of the originate image and one polarization combination image and the combination image of the originate image, one polarization combination image and the SPI image. Via evaluating the accurate of these classifying, we find that the information extracting of salt lick can satisfy the purpose by the SPI image, though it is difficulty to extract the information of salt lick, and making use of various polarization data can also attain the purpose partly. Therefore, we can obtain the salt lick information by: (1) calculate the SPI, and then automatic classify by computer;(2) make use of various polarization combination data, classify the image based on the different characteristics of different salt lick in some polarization combination image.And in this paper, we also classify the very high resolution, 3.125m quad-polarization radar images, and find that the information extracting of salt lick is not effective, maybe because of large number of noise in the image. Wherever, after resampling the very high resolution radar image from 3.125m resolution to 12.5m resolution, the accurate of information extracting of salt lick can improve highly. So, it can be seen that very high resolution radar data does not necessarily fit the information extracting of salt lick, only the appropriate resolution of radar data can be beneficial to the extraction of salt lick.To sum up, using RADARSAT-2 data can effectively extract information of salt lick in arid areas. Although the use of both SPI and multiple polarization combination data can acquire the salt lick information, the use of SPI to obtain the information can be more effective. And very high resolution radar image does not necessarily fit the information extracting of salt lick, and appropriate resolution are in favor of extracting salt lick information, such as resolution of 12.5 meters. |
PDF Full Text Request