The improvement of soil depth estimation using parent materials dataset


The respected Comrade Kim Jong Un said:

"In order to solve the problem of food in our country, which has limited land under cultivation, we should do farming in a scientific and technical way so as to increase the unit-area yield of crops."

The primary and most important problem arising in thoroughly adhering to the principle of the right crop on the right soil and the right crop at the right time as required by the Juche farming method is to assess quantitatively the state of the soil, the basic means of agricultural production. Among the soil physicochemical properties, soil depth is a very important attribute in the growth of crops and is a key factor for evaluating the potential of agricultural land.

Observations of soil depth have been carried out by field surveys that are time consuming and labor consuming for a long time. Since soil depth is closely related to topographic characteristics, digital terrain analysis technology has the potential to save time and labor without destroying solum compared to conventional methods based on field surveys. Some researchers have found that soil depth has a close relationship with parent material. The accuracy of soil depth estimation can be improved if detailed parent materials dataset are available.

We analyzed the linear and nonlinear relationships between soil depth and topographic characteristics according to the parent materials using digital terrain analysis techniques and statistical modeling methods, and have developed a model to evaluate soil depth with high accuracy in the study area.

Soil depth estimation models of different parent materials were made by linear regression analysis and nonlinear regression analysis respectively, and compared using coefficient of determination (R2) and root mean square error (RMSE).

Nonlinear regression analysis produced models with higher accuracy compared to linear regression analysis, and the use of parent material dataset developed reliable soil depth estimation models in the study area. This indicates that the relationship between soil depth and topographic characteristics is typical and varied with parent materials.

Table 1. Comparison parameters between linear and non-linear regression. R2 is the coefficient of determination; RMSE (cm) is the root mean square error

The approach is of great significance in precision agriculture for the optimization of agricultural land management by rapidly producing soil depth maps at low cost.

Our results of this study were published in the journal "Eurasian Soil Science" (2021, Vol. 54, No. 1, pp. 1-12) of the SPRINGER Publishing House under the title of "Digital Terrain Analysis Approach to Improve Soil Depth Prediction with Parent Material Dataset" (