CollectHomepage AdvertisementContact usMessage

Arid Land Geography ›› 2022, Vol. 45 ›› Issue (5): 1534-1546.doi: 10.12118/j.issn.1000-6060.2022.015

• Biology and Pedology • Previous Articles     Next Articles

Response analysis and modeling of microwave dielectric properties of typical saline soil in Xinjiang

ZHAO Shuang1,2,3(),DING Jianli1,2,3(),HAN Lijing1,2,3,HUANG Shuai4,GE Xiangyu1,2,3   

  1. 1. College of Geography and Remote Sensing Sciences, Xinjiang University, Urumqi 830046, Xinjiang, China
    2. Xinjiang Key Laboratory of Oasis Ecology, Xinjiang University, Urumqi 830046, Xinjiang, China
    3. Key Laboratory of Smart City and Environment Modelling of Higher Education Institute, Xinjiang University, Urumqi 830046, Xinjiang, China
    4. College of Geography and Environment, Liaocheng University, Liaocheng 252000, Shandong, China
  • Received:2022-01-10 Revised:2022-03-13 Online:2022-09-25 Published:2022-10-20
  • Contact: Jianli DING E-mail:zhaoshuang@stu.xju.edu.cn;watarid@xju.edu.cn

Abstract:

Soil salinization negatively affects regional economic and ecologically sustainable development. Due to the unique arid climate conditions in Xinjiang, China, there are many varieties of saline soils that are widely distributed across the region. Currently, researchers have proposed many models to describe the relationship between the dielectric constant of soil samples and the soil moisture content. However, it is still difficult to establish a model that describes the soil salt dielectric properties, which can also clarify the response relationship between dielectric constant and salt contents. In order to explore the influence of soil salinity on the dielectric properties of soil, two typical saline soil types (sulfate-chloride type, N a 2 S O 4 - N a C l; chloride-sulfate type, N a C l - N a 2 S O 4) were used in this study. The dielectric constant of 240 soil samples was obtained at 500 discrete frequencies from 0.3 GHz to 20.0 GHz to analyze the effects of water content, salt content, salt type, and soil texture on the soil dielectric constant of soil (including the real part, ε '; imaginary part, ε ; the real part of the modulus, M '; the imaginary part of the modulus, M ). Random forest was used to develop salt estimation methods; this work permits us to evaluate the effectiveness of using the dielectric constant of soil in estimating the soil salt content. The Random forest method has been widely used in the research related to soil property prediction. The algorithms have the capacity to provide the relative importance of the variables used; this allows us to further explore the relationship between soil salt and dielectric constant. The obtained results were as follows: (1) Salt content has an effect on both the real and imaginary part of the complex dielectric constant of wet soil and dry silty loam. (2) There is a more significant difference between ε at a frequency of 0.3 GHz for the two moist saline soils of the same class ( ε N a 2 S O 4 - N a C l> ε N a C l - N a 2 S O 4). (3) The results of machine learning indicate that the imaginary part of the modulus ( M ) is more closely related to the salt content and 0.3-5.0 GHz is the most important frequency range for such predictions. This research indicates that the soil salinity estimation method combined with soil dielectric characteristic analysis experiments and machine learning can effectively establish the most important dielectric constants and frequency ranges. In conclusion, the results presented here provide a scientific background for the microwave remote sensing monitoring of soil salinization on the complex underlying surface.

Key words: microwave dielectric constant, soil salinity, machine learning, mircrowawe remote sensing