Digital Soil Mapping of Soil Organic Matter with Deep Learning Algorithms

Abstract

Digital soil mapping has emerged as a new method to describe the spatial distribution of soils economically and efficiently. In this study, a lightweight soil organic matter (SOM) mapping method based on a deep residual network, which we call LSM-ResNet, is proposed to make accurate predictions with background covariates. ResNet not only integrates spatial background information around the observed environmental covariates, but also reduces problems such as information loss, which undermines the integrity of information and reduces prediction uncertainty. To train the model, rectified linear units, mean squared error, and adaptive momentum estimation were used as the activation function, loss/cost function, and optimizer, respectively. The method was tested with Landsat5, the meteorological data from WorldClim, and the 1602 sampling points set from Xinxiang, China. The performance of the proposed LSM-ResNet was compared to a traditional machine learning algorithm, the random forest (RF) algorithm, and a training set (80%) and a test set (20%) were created to test both models. The results showed that the LSM-ResNet (RMSE = 6.40, R2 = 0.51) model outperformed the RF model in both the roots mean square error (RMSE) and coefficient of determination (R2), and the training accuracy was significantly improved compared to RF (RMSE = 6.81, R2 = 0.46). The trained LSM-ResNet model was used for SOM prediction in Xinxiang, a district of plain terrain in China. The prediction maps can be deemed an accurate reflection of the spatial variability of the SOM distribution.

Assessing Soil Prediction Distributions for Forest Management Using Digital Soil Mapping

“…In the validation analysis, the SOM map showed reliable results, with R 2 = 0.61, MAPE = 41.18%, and RMSE = 2.03%. When compared to previous studies, we can observe that the general level of validation accuracy for SOM predictions ranges between R 2 = 0.45 at 10 m using topographic attributes only [70], R 2 = 0.51 at 30 m [71], and R 2 = 0.53 using a QRF at 30 m resolution [34]. In our study, the observed standard deviation of the prediction distribution ranged from 1.7% to 7.7%.…”

A Systematic Review on Digital Soil Mapping Approaches in Lowland Areas

“…Furthermore, climate-related covariates which focus on climatic factors were found in 8% of the reviewed articles. These covariates were recognized for their significance in shaping soil properties, especially in lowlands with diverse climatic conditions [35,49,62]. They play a critical role in assessing soil resilience to climate change and its implications for sustainable land use and agriculture.…”

Research on Hyperspectral Modeling of Total Iron Content in Soil Applying LSSVR and CNN Based on Shannon Entropy Wavelet Packet Transform

“…Second, the residual block structure in the ResNet network enhanced the depth of the model network. This enabled the model to learn the characteristics of the spectral curve more deeply, while avoiding model instability and accuracy loss caused by increased network depth [66,67]. Third, in the establishment of the weighted layer, we used the MSE with the largest subset of 1600 times for modeling, removed frequency bands with smaller MSE coefficients, and used the normalized absolute value of the MSE coefficient as the initial weight.…”