Uncertainty Estimation Method for Soil Organic Carbon Prediction Based on Bootstrap Sampling

Abstract

The present invention proposes a soil organic carbon prediction uncertainty estimation method based on Bootstrap sampling, the steps of which are: collecting and preprocessing soil samples, obtaining organic carbon content data and soil hyperspectral data of soil samples and performing preprocessing ; Using the partial least squares regression method to establish the soil organic carbon content data of soil samples and the soil organic carbon prediction model of soil hyperspectral data; carry out random sampling with replacement on the original measured sample data, obtain a sub-sample for each sampling, and construct Calculate the estimated value of the parameters of each sub-sample by calculating the measured value and predicted value matrix; use Bootstrap re-sampling technology to extract a certain number of samples from the original sample; use the Bootstrap sampling method to extract the parameters of the sample to evaluate the soil organic carbon prediction model Uncertainty; Model Accuracy Evaluation. The invention reduces the problem of low prediction model accuracy caused by sampling representativeness and prediction model space variability.

Description

technical field [0001] The invention relates to the technical field of statistical analysis of uncertainty of soil organic carbon prediction models, in particular to a method for estimating uncertainty of soil organic carbon prediction based on Bootstrap sampling. Background technique [0002] Soil organic carbon is an important part of balancing the global carbon cycle and the main indicator for maintaining soil quality. However, due to the influence of natural factors and human factors, there are uncertainties in the content and spatial distribution of soil organic carbon. Errors caused by this uncertainty affect the precise prediction and mapping accuracy of soil organic carbon. [0003] Vis-NIR technology provides a fast and accurate near-ground remote sensing estimation method of soil organic carbon. The whole process from data acquisition to predictive modeling is quite simple, which saves the cost of laboratory analysis. At the same time, using the advantages of cros...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the technical problem of the lack of quantitative analysis of the uncertainty of the existing soil organic carbon prediction model, the present invention proposes a soil organic carbon prediction uncertainty estimation method based on Bootstrap sampling, which introduces the Bootstrap parameter estimation method in traditional statistics In soil science, the uncertainty assessment of the soil organic carbon prediction model is realized, which reduces the problem of low accuracy of the prediction model due to sampling representativeness and spatial variability of the prediction model (training samples and model parameters)

Images