Two-Stage Soil Parameter Estimation From Hyperspectral Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for estimating soil parameters, particularly organic carbon content, struggle to efficiently combine historical data with modern satellite imagery and require a single type of machine learning, leading to inaccuracies in dynamic parameter estimation.
Innovation Solution
A two-stage architecture system that includes a first stage estimator using a deep neural network for low-dynamic parameters and a second stage estimator using classical machine learning algorithms, connected by a module for latent feature reduction, effectively combining historical and real-time data to estimate both low and highly dynamic soil parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single-stage architecture using only one type of machine learning is used, then the system is simpler to implement, but the accuracy of dynamic parameter estimation deteriorates
Solution Approach 1:
The system divides the estimation task into two separate stages: a first stage estimator for low-dynamic parameters (soil texture, elevation) and a second stage estimator for highly dynamic parameters (organic carbon content, moisture). Each stage uses machine learning algorithms optimized for its specific parameter type, thereby improving overall estimation accuracy while maintaining manageable system complexity through modular design.
2Reliability
If only modern satellite imagery is used for training, then the data is more relevant to current conditions, but the training dataset is insufficient for optimal system performance
Solution Approach 1:
The system merges historical soil parameter data with modern satellite imagery data to create a comprehensive training dataset. This combination allows the system to leverage the large volume of historical ground truth data while incorporating contemporary spectral information from satellite images, thereby achieving both sufficient training data quantity and high data relevance.
3Quantity of substance
If historical soil parameter data is combined with satellite imagery, then the training dataset becomes more comprehensive, but the system must handle data from different time periods with varying characteristics
Solution Approach 1:
The system segments the training data into distinct categories: historical soil parameter data and contemporary satellite imagery data. By keeping these data sources separate during the training process and using appropriate preprocessing techniques for each type, the system can effectively handle their different temporal characteristics and data formats, improving adaptability while maintaining comprehensive training coverage.
Data Source
AI summary
The invention proposes a new system and method for estimating dynamic soil parameters based on multispectral or hyperspectral images. The system 105 has two-stage estimation architecture and a fully connected layer 108 that connects the two estimators. The system estimates dynamic parameters such as organic carbon in the soil based on multispectral or hyperspectral images 102 and also uses geocoordinates 100 and soil elevation 101 as inputs. Multispectral or hyperspectral images 102 are further processed in the pre-processing module 103 and the data 104 is transmitted to the training neural network 107 of the 1st degree estimation module 106. It further processes them in submodule 108 for the extraction of latent features, estimates the low-dynamic parameters of the soil and forwards them to the output. Also, the output data of submodule 108 is the input data of module 2 for the 2nd degree estimation which estimates the highly dynamic parameters. The low-dynamic parameters 112 are e.g. soil texture, and highly dynamic parameters 113 are e.g. the concentration of organic carbon in the soil. Submodule 108 connects said modules 106 and 110.