Predication of the post mining land use based on random forest and DBSCAN

Area of Science:

• Environmental Science
• Soil Science
• Sustainable Development

Background:

• Mine reclamation is crucial for sustainable mining.
• Evaluating post-mining land use is essential for ecological restoration.
• The Hongliulin mining area in China faces land use challenges.

Purpose of the Study:

• To evaluate different post-mining land uses in the Hongliulin mining area.
• To classify soil samples based on environmental factors for reclamation planning.
• To provide recommendations for land use and idle land utilization strategies.

Main Methods:

• Collected 145 soil samples (0-20 cm depth) from existing land uses and areas for reclamation.
• Analyzed 14 environmental factors including soil organic matter (SOM), total nitrogen (TN), available phosphate (AP), available potassium (AK), K, slope, aspect, and Topographic Wetness Index (TWI).
• Utilized a random forest classifier, achieving 95% accuracy in land use classification.

Main Results:

• The random forest model accurately classified land use types with 95% accuracy.
• Soil samples classified as C1 exhibited significantly higher total nitrogen (TN) and K values, indicating smaller soil particles and sandy loam texture.
• Among 45 samples for reclamation, 15 were classified into C1, 23 into C2, 5 into C3, and 2 into C4, with C1-C showing maximum soil nutrients.

Conclusions:

• Soil nutrient content and environmental factors are key indicators for post-mining land use classification.
• Specific crop suggestions (soybean, murphy) and strategies (ground subsidence filling, soil nutrient enhancement) are proposed based on soil classification.
• The findings support informed land use planning and strategies for utilizing idle post-mining lands.