Abstract:

This study used principal component analysis combined with distance clustering, K-means dynamic clustering, and comprehensive soil nutrient evaluation methods to analyze and evaluate the soil nutrient status of cotton field soil in Xinjiang, China. The results showed that the available copper, manganese, and alkali hydrolyzable nitrogen played a major role principal component analysis: The available copper content was high (1.82 mg·kg^-1); the available manganese content was low (11.36 mg·kg^-1); and the alkali hydrolyzable nitrogen content was 122.07 mg·kg^-1. The available phosphorus, potassium, zinc and iron contents were low. Thus, the distribution of soil nutrients was uneven. In the distance clustering and K-means dynamic clustering, the organic matter, available nitrogen, and available manganese contents were lower, whereas the other nutrient contents were higher. In the distance cluster, the soil nutrients could be expressed as: class I > class V > class IV > class II > class III, whereas the soil nutrients could be expressed in the K-means dynamic cluster as: class III > class I > class V > class II > class IV. In the comprehensive evaluation of soil nutrients (IFI), the grades of 1 and 16 are higher. The second company, third company, fourth company, sixth company, 15th company, 19th company, and the second prison area had higher levels, whereas the 8th, 9th, 10th, 11th, 12th, 17th, 18th and 20th companies were in the middle level. The grades of companies 5, 7, and 1 supervision district and agricultural city station were lower. Thus, K-means dynamic clustering is better than distance clustering, as it can be more scientific, reasonable, accurate, and effective for the comprehensive evaluation of soil nutrients.

Key words: soil nutrients, distance clustering, K-means dynamic clustering, comprehensive evaluation value