A method for estimating total carbon content of soil surface from visible images taken by drones and soil water content

Abstract

Problems related to nitrogen fertilizers in recent years include the adverse effects on the environment due to the accumulation of applied fertilizers and the soaring international price of fertilizers. As a countermeasure, there is variable fertilization that changes the amount of fertilization according to the nitrogen state in the soil. To evaluate nitrogen content, satellite imagery has been used to estimate soil humus content, Soil humus content is known as an index closely related to nitrogen content. However, affected by shooting timing restrictions and clouds. To obtain high-resolution imagery without cloud effect, we used drone. To estimate soil humus content. we used total carbon content as index. On March 15 and 29, 2022, aerial imageries were taken at the National Agriculture and Food Research Organization (NARO). On March 14, 2022, we collected soil samples at 1-m mesh intervals and measured the total carbon content using NC analyzer. We proposed a flowchart to estimate the amount of carbon from imageries including the following features: 1. Strings and footprints on imagery are removed by determining a brightness threshold. The optimum mesh size is determined by evaluating the influence of noise.3. The relationship between color space, soil moisture content at the time of photographing, and total carbon content is obtained by linear regression. By applying the above flowchart to aerial imageries, we obtained an estimation formula with an adjusted R-square of 0.526. In future work, we plan to apply our ?owchart to more various kinds of soil to improve accuracy of estimation.