|Sudduth, Kenneth - Ken|
Submitted to: ASABE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: 6/11/2013
Publication Date: 7/21/2013
Citation: Hu, G., Sudduth, K.A., Myers, D.B., He, D., Nathan, M.V. 2013. Factors affecting soil phosphorus and potassium estimation by reflectance spectroscopy. ASABE Annual International Meeting. [available online]. Paper No. 131595956. Interpretive Summary: Rapid estimation of soil nutrients is needed to assess soil variability and to guide site-specific fertilization. One candidate technology for this purpose is diffuse reflectance spectroscopy, which measures light reflected from the soil in the visible and near infrared wavelength bands. Reflectance spectroscopy has potential to replace traditional laboratory testing, which is often slow and expensive. However, previous studies have reported spectroscopy estimates of plant available soil phosphorus (P) and potassium (K) to be of variable accuracy, with many results less accurate than needed to guide site-specific fertilization. Using a large soil database of over 1500 samples, we conducted this study to determine under what conditions accurate P and K estimates could be obtained. No clear trends in accuracy were observed for different soil regions, textures, or organic matter levels. However, we did identify a new data transformation that consistently improved estimation accuracy. Although this study will provide soil scientists and instrumentation engineers with some new insight into the factors affecting the accuracy of P and K estimations, additional research will be needed to determine if these findings can improve accuracy to the level needed for guiding variable-rate fertilization.
Technical Abstract: Visible and near infrared (VNIR) diffuse reflectance spectroscopy has potential in site-specific measurement of soil properties. However, previous studies have reported VNIR estimates of plant available soil phosphorus (P) and potassium (K) to be of variable accuracy. In this study, we used a database of over 1500 soil samples to investigate what factors influenced P and K estimation accuracy. Specifically, the effect of classifying soil samples by major land resource areas (MLRAs), cation exchange capacity (CEC) or organic matter (OM) was investigated. Additionally, calibrations using only those samples within the approximate range of interest for fertilizer application to field crops – P from 0 to 108 mg/kg and K from 0 to 768 mg/kg – were compared to calibrations using the full range of soil samples. Pretreatments of log(1/reflectance) plus mean normalization plus median filter smoothing with or without direct orthogonal signal correction (DOSC) were investigated. Results from partial least squares regression (PLSR), principal component regression (PCR) and support vector regression (SVR) were compared. Reasonable estimates of P and K were obtained for soil samples from two Missouri MLRAs (119 and 115B) out of the eight analyzed. Model estimates were poor when soil samples were grouped by CEC or OM; however, there was some indication that VNIR estimation of P and K might be possible for soils low in OM. Accuracy was maintained when analyzing a reduced wavelength range from 1100 to 2450 nm, suggesting this narrower sensing range might be used for on-the-go sensors. PLSR provided better accuracy than PCR or SVR for both P and K. The DOSC pretreatment significantly improved P and K estimation accuracy. The results of this research provided some insight into the factors affecting the accuracy of P and K estimation by VNIR models, but additional research is needed to determine if these findings can lead to P and K estimations sufficiently accurate to guide variable-rate fertilization.