Stratified Data Reconstruction and Spatial Pattern Analyses of Soil Bulk Density in the Northern Grasslands of China

• Main Authors: Yuxin Qiao, Huazhong Zhu, Huaping Zhong, Yuzhe Li
• Format: Article
• Language: English
• Published: MDPI AG 2020-11-01
• Series: ISPRS International Journal of Geo-Information
• Subjects: Soil bulk density; grassland; spatial pattern analysis; pedotransfer function; stratified soil
• Online Access: https://www.mdpi.com/2220-9964/9/11/682

The spatial pattern of soil bulk density in the grasslands of northern China largely remains undefined, which raised uncertainty in understanding and modeling various soil processes in large spatial scale. Based on the measured data of soil bulk density available from soil survey reports from the grasslands of northern China, we constructed a soil Stratified Pedotransfer function (SPTF) from the surface soil bulk density. Accordingly, the stratified bulk density data of soil vertical profile was reconstructed, and the estimation of soil bulk density data in horizontal space was performed. The results demonstrated that the soil bulk density of the grasslands of northern China was typically high in the central and northwestern regions and low in the eastern and mountainous regions. Mean soil bulk density of the grasslands was 1.52 g·cm⁻³. According to geographical divisions, the highest soil bulk density was observed in the Tarim basin, with mean soil bulk density of 1.91 g·cm⁻³. Conversely, the lowest soil bulk density was observed in the Tianshan Mountain area, with mean soil bulk density of 1.01 g·cm⁻³. Based on data obtained on various types of grasslands, the soil bulk density of alpine meadow was the lowest, with a mean soil bulk density of 0.75 g·cm⁻³, whereas that of temperate desert was the highest, with mean soil bulk density of 1.80 g·cm⁻³. Mean prediction error, root mean square deviation, relative error, and multiple correlation coefficient of soil bulk density data pertaining to surface layer (0–10 cm) in the grasslands of northern China were 0.018, 0.223, 16.2%, and 0.5386, respectively. The approach of employing multiple data sources via soil transfer function improved the estimation accuracy of soil bulk density from stratified soils data at the large scale. Our study would promote the accurate assessment of grassland carbon storage and fine land characteristics mapping.