Assessment of Moisture Retention of Soil with Biochar Using Linear Regression and Neural Network Pattern Recognition

Biochar is a carbon-rich solid material created by pyrolysis. It has special interactions with soil's engineering properties. The study investigated how the location, type, feedstock (namely, Wood, Rice Husk, and Corn), and rate of biochar affected porosity, bulk density, and moisture retention. The study aims to bridge the gap of limited studies pertaining to the statistical tools to showcase the relationship of biochar to soil properties. The data gathered for the study came from multiple related literatures studying biochar effects, for a total of 51 raw data to display the reliability and validity of the results. The researchers used Linear Regression as a statistical treatment to assess the relationships between biochar and the soil properties. Pearson Correlation was used to further verify the direction of the relationship of the variables. In addition, the significance of the independent variables is also statistically observed through the use of the SPSS app. Based on the results, the percent rate of Biochar has a direct relationship with Soil Porosity and Moisture Retention regardless of the location, type, feedstock of the Biochar. However, for the Bulk Density, it is found that there exists an inverse relationship with percent rate of Biochar. Furthermore, neural network pattern recognition was employed to determine whether soil moisture retention increased or decreased based on the identified variables. The output shows low values of error and cross-entropy, indicating that the classification model is performing well.