Abstract: Abstract: In land reclamation project of western China, the sandy loam is very fragile for erosion in its ecosystem. The main purposes of the sandy loam soil remediation are the anti-desertification and the improvement of vegetation growth ability. Since water content of sandy loam soil and degree of compacting are two important physical parameters, which have great influences on the vegetation growth, a fast and accurate detection method is needed to determine these parameters.In the land quality evaluation, the sampling method is important. The remote sensing method is based on big surface elements, which is not suitable for the micro examinations. The method of taking soil samples in the field and then oven-drying in the laboratory can yield good precision in the tests, but very inefficient and destructive to soil ecosystem. The other methods as Time Domain Reflectometry (TDR), γ-ray, et al. have the common disadvantages of discontinuity and inaccuracy. The ground penetrating radar (GPR) technique with fast, convenient, precise and nondestructive characteristics is widely used in quality test of engineering projects. The GPR transmits and receives high-frequency electromagnetic waves through the antenna for the accomplishment of detections. This study focused on sandy loam soil collected from a research area in Inner Mongolia where the land consolidation project had just finished. The research started with sandy loam soil samples gathering from the research area. Then a wooden container was built, which was filled with sandy loam soil samples as a physical model. After calibrating water contents and degrees of compactness of the sample in the model, a GPR system with 900MHz antenna was then used for determining soil water content and compaction. The GPR hardware in this study was the 'GR' radar system which was designed and built by China University of Coal Mining technology, Beijing.The Cadzow analysis was used in the Auto-Regressive and Moving Average (ARMA) power spectral recognition to reduce the number of parameters which can cause complex calculations. Then, the original data gathered by the GPR from the physical model was processed by the optimized ARMA power spectrum recognition in order to get the power spectrum data in frequency domain. To find out the relation between power spectrum and water contents and degrees of compactness of sandy loam soil, the distribution of radar power spectral energy in frequency domain needed to be analyzed.The envelopes of power spectrum distributed in different ranges and center frequencies, which represented the sandy loam soil with different water contents and degrees of compactness of different samples in the physical model. The distribution of high and low frequency envelopes showed primarily that the variation of different frequency spectrum energy was related to water contents and degrees of compacting in soil samples. Therefore, to define a cut-off point that can divide power spectrum data in frequency of sandy loam soil was very critical. By analyzed the spectrum data of the tests, two cut-points have been found according to maximum correlations of power spectrum and water contents and degrees of compactness, respectively. It was found that the low frequency part of the power spectrum energy had maximum correlation (r = 0.952) with the water content when the cut-point was at 670 MHz. Similarly, there was another cut-point at 1 000 MHz which had the maximum correlation (r = 0.947) with the degrees of compactness of the sandy loam soils. Two correlation models have been built to describe the water contents and degree of compactness with radar ARMA power spectrum energy data, respectively. The correlation models have been used in the actual tests of water contents and degrees of compactness of sandy loam soils in the research area in Inner Mongolia of China. The results showed that the values inverted by the models were well related to the real values obtained from the conventional test methods for soil water content and degree of soil compaction. .Based on experimental and actual tests and analysis, it was possible to use GPR technique to detect water contents and degrees of compactness of sandy loam soil in similar conditions as the research area. Also, the speed and the precision of physical properties detection can be improved by using ARMA power spectrum estimation method. So, it has significant meaning for quality tests of land consolidation projects. However there still some insufficiencies in this research. The ranges of application of the relation models were limited to the sandy loam soils. And the soil moisture contents and degrees of compactness were the average values from the ground surface to the effective detection depth. Future research was needed to solve these issues.