Page 5 - การประเมินปริมาณโพแทสเซียมในดิน ด้วยเทคนิคเนียร์อินฟราเรด (NIR) The evaluation of Potassium content in soil by Near Infrared.

ห้องสมุดกรมพัฒนาที่ดิน

5

Abstract

Potassium is one of the most important macronutrients for plant growth. Potassium-

deficient plants tend to wilt easily, become stunted, yellow lower leaves, and burn along the leaf

margins affects to the growth and productivity of plants. Potassium content analysis in soil is
essential for recommending fertilizer rates to plants. The effective fertilizing of plants depends

on the accurately evaluation of soil analysis results and timely fertilization management. At the
present, the soil analysis method involves several steps in sample preparation and take a long

time in chemical processes cause spend more times for all process. Therefore, this study aims to

develop the soil analysis technique to predict potassium content by near infrared spectroscopy
technique. The 1,070 soil samples were collected and acquired for NIR spectra in the range 780-

2,560 nm with diffuse reflectance mode. Partial Least Square Regression models were developed
to predict available Potassium in the soil samples. Cross validation comparison of the predicted

and the actual values (using standard methods) was performed. The created models consist of

model was developed from the all sample, model of the samples with ≤20, 21-40, 41-60 and
>60% clay content, respectively. It was found that the predictive coefficients (R ) of the models
2
were 0.54, 0.71, 0.87, 0.90 and 0.97, respectively. The accuracy of the equations (RPD) were

developed from the samples containing 21-40%, 41-60%, and >60% clay content were 3.15, 3.22
and 5.99, respectively. They were evident that the models of the samples with more than 20%

2
clay contents size (R >0.87) could be effectively predicted soil potassium content. Especially, the
2
equations developed from samples with >60% clay contents (R = 0.97) showed the highest R
2
values and the best tendency to predict soil potassium content (RPD=5.99). However, further

studies by using soil samples with specific land use are needed to obtain high accurately
equations and can be applied widely in the future

Key words (EN): Plant nutrient, Near Infrared spectroscopy, Clay Particle