Soil moisture estimation based on GPR power spectrum and envelope amplitude in sand loam

Abstract: The soil water is of great significance to the management and restoration of the ecological environment in western China. In order to realize an accurate, rapid and continuous detection of the distribution of water content in the shallow sandy loam by virtue of the ground penetrating radar (GPR) technology, 2 means were thus used including radar wave average envelope amplitude (AEA) and auto-regressive and moving average (ARMA) power spectrum recognition, which were also utilized to invert the water contents of the shallow sandy loam soils. AEA was used to analyze and establish the relationship between the radar early-time average amplitude envelope signal value and the dielectric constants of surface strata with depth less than 1 m. ARMA power spectrum recognition was used to build the relation between power spectrum and water contents of the sandy loam soils with depth range within radar's maximum detection depth. The sandy loam soils with different water contents could differentiate the distribution of the envelopes of power spectrum. The variation of different frequency spectrum energy was related to the water contents in soil samples, which could be shown by the distribution of the high and low frequency envelopes. After the comparison of the water contents inverted by AEA, ARMA and drilling sampling, it was found that the water contents of the shallow sandy loam soils inverted by ARMA was effectively in the depth of 0-10 m. But the results were inaccurate in the depth of 0-0.5 m. However, the AEA provided accurate water contents of the shallow sandy loam soils in the depth of 0-0.5 m. The correlation coefficients of volumetric water contents derived by drilling sampling and inverted by ARMA were 0.57, 0.62, 0.96 and 0.99 respectively in the depth of 0, 0.5, 1.0 and 10.0 m, with the corresponding mean relative errors of 47.37%, 30.32%, 11.20%, and 16.71%. The volumetric water contents derived by AEA almost equaled the values derived by drilling sampling in the depth of 0-0.5 m. The correlation coefficient of volumetric water contents derived by drilling sampling and AEA were 0.98, 0.99 in the depth of 0-0.5 m respectively, but the correlation coefficient was only 0.07 when the depth was 1.0 m. Based on the applicability of the 2 methods, a joint model was established to invert the water contents of the shallow sandy loam soils in the depth of 0-10 m in the typical arid area of western China. The actual detection showed that the average relative error between the water contents and the drying method of the joint model in the range of 0-0.5 m decreased by 16.81%-41.80% compared to that of ARMA. Compared with AEA, the joint model could invert the water contents of deeper strata. The joint method had the average relative error and root mean square error was 5.57%-16.71% and 0.62%-2.08%, respectively. It indicates that the joint model can quickly, accurately and continuously obtain the volumetric water contents of the shallow sandy soils in the depth range varying from 0-10 m in the arid area of western China.