Evaluation of near-infrared spectroscopy for analysis of soil and plant in agriculture
Rajendram, G. (2006). Evaluation of near-infrared spectroscopy for analysis of soil and plant in agriculture (Thesis, Doctor of Philosophy (PhD)). The University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/12775
Permanent Research Commons link: https://hdl.handle.net/10289/12775
In this research, a critical scientific appraisal of Near Infrared (NIR) spectroscopy for the analysis of soil and plant is presented. Near Infrared measurements in the study were collected using a KESNIR spectrometer with a scanning range of 400-1700 nm and a large scanning area. NIR was shown as the instrument of choice because of its versatility, nondestruction of sample on analysis and rapid measurement times. The focus of this study was to derive and test correlations between NIR spectra of soil and plant samples and concentrations of nutrient-related variables in these samples (as measured using conventional techniques). These NIR correlations effectively constitute a series of calibrations. The concentrations of a number of elements ( or other soil or plant physicochemical properties) will then be able to be reliably estimated on the basis of a sample's NIR spectrum alone. Part of this project is devoted to deriving calibrations, and the rest will focus on assessing how well these NIR calibrations perform, by comparison ofNIRpredicted values to those obtained using conventional chemical extraction followed by analyte-specific instrumental analysis. In order to achieve the above for soil testing, calibrations included samples from all 15 New Zealand soil orders and subsoil types, geographical regions, and land use. Most of the soil and plant samples analysed have been taken from the pool of such samples routinely submitted to AgResearch for analysis. In addition, to date, no studies have been reported in which NIR spectra have been recorded for a full range of soil orders in a country. In the case of the NIR measurements, field-moist soil and plant samples could be used, rather than dried and ground samples. This represents a considerable saving of time and effort. One aim of this project was to investigate and validate measurements for soil and plant samples by NIR using field-moist samples. This would then enable the measurement of nutrient status in the field. Soil nutients and parameters looked at in this study were pH, Olsen P, sulphate-S, available K, Mg, Ca and Na, soil type, P retention, reserve K, soil total N (TN), soil total S (TS), soil total C (TC), available N and nitrate-N. The study also aimed to determine that if key nutrients could not be ultimately measured accurately and rapidly in the field using NIR, a rapid alternative method was to be investigated. It was found that NIR could not accurately predict pH, Olsen P, sulphate-S, available K, Mg, Ca and Na and nitrate-N from a direct soil scan for New Zealand soils. However, a rapid sample preparation method/complexing ('signal enhancement') of fieldmoist or dried soils for the measurement of Olsen P and available K prior to presenting to the NIR could enable accurate measurements. NIR via the 'signal enhancement' method could measure Olsen P with results obtained for New Zealand soils with accuracy comparable to Olsen P determined by the traditionalby method. Therefore Chapter 8 in this thesis is dedicated towards determining exactly how the 'base reference method' (Olsen P) behaved for New Zealand soils. Furthermore, pH could be rapidly and accurately measured after a 10-minute extraction on field-moist or dried soils using a pH meter. The accuracy of pH mesurements using NIR was improved when compared to direct soil scans when the soil solution was complexed by an indicator and presented to the NIR. Soil nitrate-N can be measured by complexing prior to presentation to NIR using the same pH water extract. Soil type, P retention, reserve K, soil total N (TN), soil total S (TS), soil total C (TC), and available N could be measured via a direct soil scan. A patented procedure using C02 to determine parameters in wet plant was developed. Key plant nutrients looked at in this study was nitrate-N, N, P, S, Mg, Ca, Kand plant moisture. Plant moisture could be accurately determined by NIR. Nitrogen, nitrate-N and Ca in dry and wet plant material could be measured with a high degree of accuracy while NIR did not have the desired accuracy for the other major elements. This study also showed that transfereable calibrations was possible with the KESNIR instruments for the measurement of plant N. Two major topics within the overall aim were also identified. These were the N status test and a soil Sulphur test which both have a large agronomic advice and significant NIR component to them. A pot trial study using soils from throughout New Zealand was used to develop a soil nitrogen test for New Zealand pastoral soils. Currently, soil tests for N availability have proven difficult to develop, and no single test has been universally adopted. A total of 52 soils comprising the major soil groups in New Zealand were collected for the pot trial. Soils collected were from Northland, Waikato, Bay of Plenty, Central Plateau, King Country, Taranaki and all major districts of the South Island. The study found that inorganic-N in the soil drives pasture production. The N content in grass can be used to accurately predict N responses to fertiliser application. There is always a response to N application. N status varies greatly across New Zealand, and therefore varies due to soil type and land use. Using multivariate analysis, 81 % of the variation was accounted for when dry matter yield was correlated against inorganic-N soil TN, and Anion Storage Capacity (ASC). Soil TN, ASC and soil type can be accurately and rapidly measured using NIR. Soil and grass N can be categorised into Low, Medium or High N status. Soil nitrate-N can be measured by complexing prior to presentation to NIR.. The aim of the work on a new Sulphur test was to determine if the Total S pool in soil is a better measure of the sulphur status than sulphate-S or easily mineralisable organic-S. Total S accounted for 71 % of the variation when compared with relative yield for 43 field trials, whereas mineralisable-S accounted for 58% and sulphate-S accounted for 59%. Sulphate-S is easily influenced by external sources - such as urine and dung from grazing animals, leaching, fertiliser and atmospheric inputs. Organic-S accounted on average for 97% of the TS, of which EOS consisted of on average 3%. The sulphate-S component consists on average 3% of the TS. The Total Sulphur pool, because of its magnitude, is not influenced to the extent that EOS and particularly sulphate-S are by external sources. It is therefore proposed that Total S is a better and more robust measure of the Sulphur status for New Zealand pastoral soils.
The University of Waikato
All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.
- Higher Degree Theses