Development of a Mid-Infrared Technique for Determination of Soil Nitrate Content
By
BERNARD REUBEN JAHN
Trang 2UMI Number: 3212859
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy submitted Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction
In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted Also, if unauthorized copyright material had to be removed, a note will indicate the deletion ® UMI UMI Microform 3212859 Copyright 2006 by ProQuest Information and Learning Company
All rights reserved This microform edition is protected against unauthorized copying under Title 17, United States Code
ProQuest Information and Learning Company 300 North Zeeb Road
Trang 3Copyright by
Bernard Reuben Jahn
Trang 4Abstract
Nitrogen is a major crop nutrient that is often applied in large amounts in production agriculture to achieve high yield From an agricultural standpoint, over- application of fertilizers not only represents wastage of resources, but can also lead to nitrate leaching into soil zones that are too deep for plant roots to utilize With sufficient water, the nitrate will leach into the groundwater leading to health and environmental problems, such as stomach cancer and algal bloom A rapid and accurate method is needed for determination of nitrate amounts in the soil in order to determine the amount of nitrate fertilizer to be applied without resulting in leaching problems
Mid Infrared (mid-IR) spectroscopy experiments were conducted to detect added nitrate in various soil types both in the laboratory and field Wavelet analysis was applied to soil Fourier transform infrared (FTIR) attenuated total reflectance (ATR) spectra data in order to predict nitrate contents Soil pastes from 14 different soils, including sandy loam, clay, and peat, were analyzed for soil nitrate contents using the FTIR/ATR technique Nitrate concentrations for the laboratory experiments varied from approximately 0 to 2400 ppm NO3;-N while concentrations for the field experiments varied from approximately 0 to 190 ppm NO3-N Three-dimensional plots were created by graphing the wavelet deconvoluted values at 32 scales for each sample From each plot, the volume of the nitrate peak was determined and correlated to nitrate
concentrations Results of the laboratory experiments indicated R?-values as high as 0.99 and standard errors as low as 24 ppm NO;-N for soil-specific calibrations Results of the field experiments gave R*-values as high as 0.98 and standard errors as low as 5 ppm NO;-N for soil-spectfic calibrations
Trang 5An alternate technique to determine nitrate content was developed in which wavelet analysis was used to identify a few wavenumbers at which interferences from other ions were minimal This method produced calibration equations that were soil independent and gave superior results to those obtained based on correlating wavelet deconvoluted volumes to nitrate concentrations A universal calibration equation was developed allowing for predicting nitrate concentrations in the 0 to 1000 ppm NO.-N range with accuracies of approximately 40 ppm NO3-N
Trang 6ACKNOWLEDGEMENTS
Trang 7TABLE OF CONTENTS
Page
LÝ XY›Š.\@Mttitdd 11
ACKNOWLEDGEMENTS 0.0 ccccccccccce cesses eeeeeeeveeasveevaeuanearass V TABLE OF CONTENTS Q0 nh nh nay vị LIST OF ABBREVIATIONS and NOTATIONS x
LIST OF TABLES 0 0 2 cece eee ce eee eeeeeevucueeecuevevueunerenenens XVii LIST OF FIGURES 0 0 ccc cece ce cence ect c ae ce eee e este eee eeeceeetnevieaeneaeess XViil LIST OF EQUATIONS 0.00 0 eee ce cece nce cece eee e ee tne esses bunbeeeeaeneeaeaeens XXII CHAPTER I NTRODUCTION Q.00 nh nhe l 1.1 Nitrogen Sources and Losses l 1.2 Precision Farming 3
1.3 Objectives cece e cece eee e eee ene ee neta been ee aes 5 I] LITERATURE REVIEW 0 ccc cc ccc eccececseecec see ee seus eee vrate 7 2.1 Need for Nitrate Sensor 7
Trang 82.3 Principles of SD€CfTOSCODY uc n nh nh nh nhe 2.3.1 Absorption Sp€ctrOSCODY co 2.3.2 Molecular Vibrations co 2.3.3 Beer-Lambert Law c co cie 2.4 Infrared SpeCtrOSCODV cuc nh nh kh 2.4.1 Near-Infrared Region 2.4.2 Mid-Infrared Region 2.4.2.1 Mid-Infrared Absorption Spectroscopy 2.4.2.2 Mid-Infrared Reflection Spectroscopy 2.4.3 Far-Infrared Region ccecceeeeeeneee eee eaee 2.5 Infrared Spectrometer Components 2.5.1 Sources of Radiation cv 2.5.2 Methods of Separating Radiation According to Wavelength nen nen hee
Trang 94.1 Overview of ExperIimental Mecthods
4.1.1 Laboratory-prepared California Soil Samples
4.1.2 Laboratory-prepared Israel Soi] Samples
4.1.3 Field-prepared California Soil Samples
4.1.3.1 Statistical Procedures used for Split Plot AnalySIS che 4.1.4 Field Soils Containing Nitrate Added by Grower
4.2 Sample Processing Procedures
4.2.1 Sample Extract Procedure for Analytical || oe) &: 10) 9 Q HH n HH HH nh né 4.2.2 Procedure for adding 50 ppm NO¿-N Booster
4.3 Spectroscopy Methods c co iằ 4.3.1 U.S Mid-IR Spectra obtained with FTIR Laboratory-based Spectrometer
4.3.2 U.S Mid-IR Spectra obtained with Filter-based SD€CITO€I€T nh nen 4.3.3 U.S NIR Spectra co à 4.3.4 Israel Mid-IR Spectra
4.4 Overview of Analysis Methods
4.4.1 Wavelet Analysis applied to Soil Spectra
4.4.2 Technique used to Select Limited Number of Absorbances nh, 4.4.3 Method of Indicator Variables
V.RESULTS AND DISCUSSION ccc
Trang 105.1.2 Soil Spectra from Israel 5.1.3 Field Experiments
5.1.3.1 NIR Results 5.1.3.2 Mid-IR Results
5.2 Absorbance Values at Selected Locations
Trang 11LIST OF ABBREVIATIONS and NOTATIONS
SYMBOL DEFINITION
a coefficient for Hamming and Hanning windows
A absorbance
A 1350 absorbance at 1350 cm’ of baseline- and water-corrected spectrum A 1500 absorbance at 1500 cm’! of baseline- and water-corrected spectrum Ad area of detector
ATR Attenuated Total Reflection
b coefficient for Hamming and Hanning windows C velocity of radiation in a vacuum
Cmol molar concentration of absorbing analyte
cm centimeter
cm’! wavenumber
C normalization factor
CaCO: calcium carbonate
CEC Cation Exchange Capacity
CO; carbon dioxide molecule
CcO;” carbonate molecule
CWT Continuous Wavelet Transform
dl intensity of light source absorbed by element of thickness, dz dz thickness of infinitesimally small element located in medium
Trang 12DO DTGS Cọ & E Eo E; E ctectronic E rotational E vibrational EC FIR FT FTIR GIS GLM GPS h
dideuterium oxide molecule Deuterated Triglycine Sulfate
molecular energy associated with vibrational ground state molecular energy associated with /“ vibrational state total energy of a molecule
molecular energy associated with 1° electronic state molecular energy associated with electronic ground state molecular energy associated with electronic states molecular energy associated with rotational states molecular energy associated with vibrational states Electrical Conductivity frequency Kubelka and Munk function center frequency frequency corresponding to scale, s Far-Infrared Fourier Transform Fourier Transform Infrared gram
Geographic Information System General Linear Model (used in SAS) Global Positioning System
Planck’s constant (6.6254x 107")
Trang 13HA HCOy H;CO; HPLC KBr KCI Io IR MCT humic acid bicarbonate molecule carbonic acid High Performance Liquid Chromatography sample number Kelvin potassium bromide potassium chloride
intensity of light source leaving medium intensity of light source contacting medium
intensity of light source at distance, =, through medium Infrared
liter
millisecond molar
Trang 14NEP NH; NH: NIR NOx ppb ppm PCR PLS SAS SS SSCM STFT Noise Equivalent Power ammonia molecule ammonium molecule near infrared nitrate molecule oxygen gas parts per billion parts per million
Principle Component Regression Partial Least Squares lead sulfide coefficient of determination ratio of reflected intensity of sample to that of nonabsorbing standard scale Statistical Analysis Software Sum of Squares
Trang 15VFA VRA WwW x(t) Xy XQ X(œ) Mijk Qj ar (œ8);
total thickness of medium Variable Filter Array Variable-Rate Application windowing function
concentration of absorbing medium signal as a function of time, ¢ indicator variable for soil set 1 indicator variable for soil set 2 indicator variable for soil set 3 signal as a function of frequency, w
slope of calibration line of wavelet deconvoluted nitrate peak volume vs nitrate concentration for i” soil type," nitrate source, and k” plot distance through medium
absorptivity constant effect of i” soil type
reciprocal of standard deviation
effect of interaction between i” soil type and /” nitrate source overall intercept
intercept parameter for soil set 1 intercept parameter for soil set 2 intercept parameter for soil set 3 effect of /” nitrate source whole plot error
Trang 16+ Evk um V V3 Và OK CO; Os frequency resolution of measuring system molar aborptivity subplot error refractive index slope of calibration line for indicator variables method wavelength (nm) overall mean slope micrometer (micron) wavenumber (cm’')
velocity of radiation at frequency, /
fundamental vibrational frequency of nitrate molecule due to symmetric stretch fundamental vibrational frequency of nitrate molecule due to out-of-plane bending fundamental vibrational frequency of nitrate molecule due to asymmetric stretch fundamental vibrational frequency of nitrate molecule due to in-plane bending scattering coefficient effect of k” block variance of plots
variance of subplot error variance of whole plot error
translation of window
Trang 17yo) @) mother wavelet Fourier transform of yt) angular frequency
potential absorption frequency associated with a transition from ground state to /" vibrational state
Trang 18LIST OF TABLES
Table Page
2.1: Fundamental, overtone, and combination peaks for the NO,
mỌ€CUÌ€ c2 Qnn n n HT TK tk tk nh ky 25 4.1: Characteristics of sotls from Israel 68
4.2: Range of pH values, CO,” contents, HCO; contents, and NO;
contents for clay and loam solÌS - 72 4.3: Example input data set using indicator variables for four distinct
groups of data Two samples for each soil set are shown 88 5.1: PROC GLM results for split plot model using slope of calibration
line of nitrate concentration vs wavelet deconvoluted nitrate peak
volume as dependent varliable cà sằ 98 5.2: PROC MIXED results for split plot model using slope of calibration line of nitrate concentration vs wavelet deconvoluted nitrate peak
volume as dependent variable cà eee ea eres 99 5.3: Results of using stepwise regression procedure on original field soil spectra obtained using the Wilks filter-based spectrometer 115
Trang 19LIST OF FIGURES
Figure Page
1.1: Sources and losses Of nItrOg€n con nen nhe 2 2.1: Number of subsamples required to adequately represent a soil sample vs level of accuracy for the purpose of nitrate determination 9 2.2: Absorption, reflection, and transmittance of incoming light 20 2.3: Partial energy level diagram of a molecule showing 3 electronic states, 3
vibrational states within each electronic state, and 3 rotational states within
=0 1x14191871 na - 22
2.4: Fundamental modes of vibration of NO¿y' molecule 24
2.5: Attenuation of radiation with initial intensity Io, absorbing medium of
concentration c, and path length b nhện 26 2.6: Incident radiation striking smooth surface resulting in specular reflection,
indicated by the dotted line§ -cnnnn nn HH nh nà kh 32 2.7: Incident radiation striking rough surface resulting in diffuse reflection
Co Coy 6 Core 8 1 <tc) nn n n nn n KH nen ng nh kh ki kh vn 32 2.8: Schematic of Attenuated Total Reflection (ATR) technique 34 2.9: Detectability (D”) of MCT (solid line) and pyroelectric (dotted line)
CELCCLOLS cece =ã EES 4I
2.10: 2-D ATR absorbance spectra of Yolo loam soil sample containing 59 ppm
1 OF) 42
2.11: 3-D wavelet deconvoluted graph of Yolo loam soil sample containing 59 ppm NO;-N showing peaks due to nitrate (indicated by arrow), carbonate
(dotted line), and other compOn€nIS co non nhe nh se 43 3.1: Time-frequency representation for Fourler analysIS 52 3.2: Time-frequency representation for short-time Fourier analysis 52 3.3: Time-frequency representation for wavelet analysis 53
Trang 203.4: Nonstationary sinusoid signal used to compare the signal processing capabilities of Founer analysis, short-time Fourier analysis and wavelet
ANALYSIS EEE EE EE Een Eres 57
3.5: Spectrogram resulting from Fourier analysis of sinusoid with 50 Hz and
200 Hz frequency components 0 cee cce cece eee ee eee ee eet en esata teat eae 58 3.6: Spectrogram resulting from short-time Fourter analysis of a nonstationary sinusoid with 50 Hz and 200 Hz frequency components using a 166 ms wide
Hamming Window c cece cee ne ene ence ce eee cence eee eee ene ee eset ee eeeeeege es 59 3.7: Spectrogram resulting from short-time Fourier analysis of a nonstationary sinusoid with 50 Hz and 200 Hz frequency components using a 10 ms wide
Hamming WInOW HH nh KT KT ng và 59
3.8: Spectrogram resulting from wavelet analysis of a nonstationary sinusoid with 50 Hz and 200 Hz frequency components using a Coiflet 3
mother wavelet 0 ccc ccc ccc e cece ee eee ene eee tet e eee te betes seas beeen ee ee engi es 61 3.9: 3-D spectrogram resulting from wavelet analysis of a nonstationary
sinusoid with 50 Hz and 200 Hz frequency components using a Coiflet 3
1910/00/4111 erVG a2 61
3.10: Absorbance spectra of nitrate (388 ppm NO;-N), carbonate (103 ppm), bicarbonate (357 ppm), and humic acid (HA, 0.98 g/L) in distilled water from
Brooksby (2002) Q0 0n ng nn TH E nee EEE eaten debe Ete 63 3.11: Percentage of carbonate specles vs pH In soIlls 63 3.12: Typical soil spectra showing problems of interfering molecules such as
carbonate and bicarbonate overlapping with the nitrate peak 64 4.1: Bucher filter-funnel setup used to leach soIl 66 4.2: Plot layout for the clay fields showing six strips with each receiving a
S221 -14;14)1):109)3:114x::: tai 70
4.3: Mattson FTIR Spectrometer with ATR accessory used for collecting mid-IR
0] 21018 0: 76
4.4: Raw ATR absorbance spectra of Yolo loam soil sample containing 59 ppm
4.5: Smoothed, baseline-corrected ATR absorbance spectra of Yolo loam soil
sample containing 59 ppm NO:-N LH nh ke 78
Trang 214.6: Wilks VFA Spectrometer with ATR crystal used for collecting mid-IR
spectra sized with a floppy điSK cu n HH nh nh kh nh net 78 4.7: NIR raw absorbance spectrum for field soil sample containing 123 ppm
DNC © 80
4.8: Baseline-corrected NIR absorbance spectrum for field soil sample
containing 123 ppm NO@-N 0 eee cece eect eden eee teat ene een nena ees 80 4.9: 3-D wavelet deconvoluted graph of a soil paste containing 845 ppm NO;-N showing peak due to nitrate (circled), carbonate, water, and other components 82 4.10: 3-D wavelet deconvoluted graph of a soil paste containing 59 ppm NOQ3-N showing peak due to nitrate (circled), carbonate, water, and other components 83 4.11: Contour map created using the Savitsky and Golay method to locate the
coordinates of the nitrate peak cece ccc c een eee eee ee seat eee e eaten een e en eas 84 4.12: Wavelet decomposed values at scale 2 possibly due to carbonate and at
Củ] Ns1-0ic8,)0irì 6é d4 .aẢ ees 85
5.1: Plot of nitrate concentration vs volume of nitrate peak for experiment conducted in laboratory involving adding large amounts of nitrate (850-2400
ppm NO3-N) to Yolo loam SOIÏ ng bà cha 90 5.2: Plot of nitrate concentration vs volume of nitrate peak for experiment
conducted in laboratory involving adding lower amounts of nitrate (0-580 ppm
NO:-N) to Yolo loam soIÏ 202 2n SH nhe 91
5.3: Plot of nitrate concentration vs volume of nitrate peak for experiment conducted in laboratory involving adding nitrate (0-800 ppm NO;-N),
bicarbonate, and humic acid to Yolo loam soIl 92 5.4: Plot of nitrate concentration vs volume of nitrate peak for all three
experiments conducted in laboratory involving adding nitrate (0-2400 ppm NO;-N), bicarbonate (0-3000 ppm HCO; ), and humic acid (0-2300 ppm) to
Yolo loam SOL cece cece eee eee een nen EEE EEE EEE EEE EEE EEE EEE EEE EE aaa 93 5.5: Plot of nitrate concentration vs volume of nitrate peak based on 80 soil
samples for five noncalcareous soils from Israel 94
5.6: Plot of nitrate concentration vs volume of nitrate peak based on 29 soil
samples for three calcareous soils from Israel 95 5.7: Baseline- and water-corrected absorbance spectra of calcareous soil with
872 ppm NO;-N showing large CO;” peak at 1450 cm’! and shoulder due to
Trang 22NOy at approximately 1350 crm) .Ố.Ố.Ố.Ố.ằỀằ 5.8: NIR raw absorbance spectrum for field soil sample containing 71 ppm
5.9: Plot of nitrate concentration vs volume of nitrate peak for Capay clay soil for two fertilizers pooled together 0.0 ecee cece ene eee ee ene et eee nena 5.10: Plot of nitrate concentration vs volume of nitrate peak for Yolo loam soil for two fertilizers pooled together 0 cece cece scene cnet ne ene eee ee end 5.11: Plot of nitrate concentration vs absorbance at 1350 cm’ for Capay clay field exXperImenIS HH nà nh nhiệt
5.12: Plot of nitrate concentration vs absorbance at 1350 cm"! for Yolo loam
field experimentS cQQQn n nn nnn n HH HH kh nh b nk Bở 5.13: Calibration plot of actual nitrate concentration vs absorbance at 1350 cm ` based on 40 soil samples randomly chosen from all field spectra pooled
together err er Err ener ee eae ees
5.14: Validation plot of actual nitrate concentration vs predicted based on 23 soil samples randomly chosen from all field spectra pooled together 5.15: Calibration plot for 14 soils pooled together using method of indicator
A729 F< SH HT nọ ki Ki Sa
5.16: Plot of y-intercept value vs absorbance at 1500 cm’! for all calcareous, noncalcareous, and field SOIÌS c2 nh ng 5.17: Calibration plot based on 124 soil samples for all soil sets pooled
tOg€th€F c2 ng ki ki Hà kh
Trang 235.23: Absorbance spectra of two soils containing approximately 75 ppm
5.24: Calibration plot of a one-term model based on the absorbance at 1399
cm” for 16 soil samples from four grower`s fields 116 5.25: Calibration plot of a two-term model based on the absorbances at 992
cm’! and 1399 cm’! for 16 soil samples from a grower`s fields 117
Trang 24LIST OF EQUATIONS
Equation Page
2.1: ›{22¡i7-140)/-2)3.5:-iiadadđaiiiẳaẳaảăảẳảỶẳảỶẢ b nate beens 20 2.2: Total molecular €I€TBY cm kh khe 21 2.3: Potential absorption frequency among electronic stat€s 22 2.4: Potential absorption frequency among vibrational states 23 2.5: Intensity absorbed by Infinitesimally small element 26 2.6: Intensity of light leaving medium cuc cà, 27 2.7: Absorbance of light by medium ccà 27 2.8: Absorbance for solution containing multiple absorbing species 27 2.9: Kubelka and Munk function 0 ccc ccc ce cee ce ee ene e eee eee ent ee eens ne eanes 33 2.10: Relationship between wavenumber and wavelength 34 2.11: Stefan-Boltzmann equation eee nee sent et eee en eae 35 2.12: NoIse equIvalent DOWT HH n nen nh kg 39 2.13: Detectability of detector cece cee eect een ee eee nh kh Hà 39 3.1: Fourler transÍOrm LH HT nh TH ng tk ra 48 3.2: Euler”s Íorma CS nh kh tk nh 48 3.3: Short-time Founter transÍOrm con nh nh nh nen 49 3.4: Gaussian window function 0.0 ccece cence ee ec eee eee e TH nh nh nha 49 3.5: Hamming/Hanning window function e cece eee ee eee teen eentees 50 3.6: Continuous wavelet transÍOrm ence nett eee ee eee e tee tee 54 3.7: Relationship between scale and frequency 54
Trang 253.8: AdmissIb!]ity COndILION u21 2n nn n nh n nh nh nh ke 55
E020) 21 ad enn EE EEE EEE Ebte 55
3.10: Mexican hat wavelet cece cece cette reece eens eenet eee enneeeenes 56 3.11: Nonstationary signal used for anaÌyses c 57 3.12: Carbonate reactions 00 cece cece eee eee nee e enna e nh nh nh 62 4.1: Split plot model used for field experiments 69 4.2: Constraints on fixed and random ÍactOrS 70 4.3: Indicator variables model c.ccn nọ nh nh nhe ha §7 5.1: Calibration equation based on field spectra obtained with FTIR
Sy of 101 80) 66 ol Co) On ere 104
5.2: Indicator variable calibration equation 105 5.3: Calibration equation based on field and Israel spectra obtained with FTIR
SY 01104 0 0) 66 eo) HH Kì kh nh ch kg 109
Trang 26CHAPTER |
Introduction and Objectives 1.1 Nitrogen Sources and Losses
Nitrogen is an important nutrient for crop production Alongside water and sunlight, nitrogen is one of the key ingredients for plant growth The amount and inorganic form of nitrogen in the soi] depends on a number of processes that involve fixing and converting nitrogen to other forms and can be summarized in a nitrogen cycle diagram as shown in figure 1.1 The two main natural sources of nitrogen fixation are from biological and atmospheric sources Biological fixation occurs in nodules on the roots of legume plants where nitrogen fixing bacteria take nitrogen from the air and convert it to ammonia (Ophardt, 1997) Atmospheric nitrogen fixation takes place due to lightning strikes that break apart nitrogen molecules allowing them to bond with oxygen The nitrogen oxides then dissolve in rain resulting in nitrates that are carried to the ground
Trang 27Nạ in Atmosphere Manure Nitrogen fertilizer Fixation | Mineralization Organic nitrogen > Nitrification - NHạ » NO; we sa vn : ô`* * đ a v ` ` Crop 4° Leaching uptake Denitrification Figure 1.1: Sources and losses of nitrogen The losses are shown with dotted lines
previously stated Denitrification is the conversion of nitrate to a gaseous form that is lost to the environment This process occurs in soils with warm temperatures, moist
conditions, and a near neutral pH There are two other major sources of nitrogen in
agricultural fields: animal manure and inorganic fertilizers These two sources account for the majority of the nitrogen in fields and consequently lead to the leaching problems that occur Weather plays an important factor on timing of fertilizer application
Trang 28Crops such as corn require large amounts of nitrate during key growth periods Usually nitrate fertilizers like ammonium nitrate, calcium nitrate, and/or urea are applied across the field uniformly However, there can be considerable yield variation within a field due to a combination of factors including differences in moisture, nutrients, topography, etc, even when the field is managed uniformly This implies that not all locations within the field require the same amount of nitrogen Uniform application leads to excess nitrate where yield potential is low This excess nitrate may leach into the ground water since nitrate is a highly mobile ion Nitrate contamination of groundwater can lead to health problems such as “‘blue-baby” syndrome and stomach cancer as well as environmental issues like algal bloom and greenhouse effect due to N20 (Ehsani et al.,
1999)
1.2 Precision Farming
Precision farming, a technique which involves managing agricultural inputs and outputs on a site-specific basis, has received much attention over the last decade due to its potential to decrease inputs such as fertilizers, protect the environment, enhance product quality and/or increase yields The technique attempts to use all available soil
information across the field, such as nutrient levels, moisture contents, pH, texture, etc.,
Trang 29model, and a soil nitrate map, and all these maps could be pooled together using a geographic information system (GIS) package (Adamchuk, 2003)
Site-specific-crop-management (SSCM) 1s based on a system-engineering approach to crop production where inputs are applied on an "as needed basis," and was made possible by recent innovations in technologies such as microcomputers, geographic information systems, positioning technologies (Global Positioning System, GPS), and automatic control of farm machinery (Robert et al., 1994) SSCM combined with variable-rate application (VRA) allows one to apply the nght amount of fertilizer at the correct location in the field Experiments conducted at the University of Idaho showed that reductions in fertilizer application amounts of 25% were possible using variable rate application technology with no decrease in yield (Fisher et al., 1993) Farmers typically apply fertilizers uniformly in amounts excess of what the crops need in order to prevent potential yield loss due to nutrient deficiency Morgan and Ess (1997) stated, “For a typical Midwestern corn grower, fertilizer accounts for over one-fourth of total cash production expenses” Applying lower fertilizer amounts to areas with limited yield potential within a field can lead to savings in fertilizer costs as well as reduced leaching of nitrate into the groundwater
Trang 30intensive soil survey data (Franzen and Cihacek, 1998) Soil survey data allow for the most complete representation of a field However, this method requires significant amounts of time and labor Due to the interaction of soil properties, often extensive soil sampling is required in order to obtain an accurate representation of the field at that given time Nitrate is a highly mobile ion and easily leaches from the soil when water is
applied However, phosphorus and organic matter levels, as well as soil pH, tend to be more stable with time and do not require as extensive sampling Currently there are commercial sensors available for in-situ measurements of soil pH but none is available
for accurate determination of soil nitrate content
1.3 Objectives
The objectives of this study were:
1 To apply wavelet analysis to Fourier transform infrared/attenuated total reflection (FTIR/ATR) spectra of nitrate-spiked Yolo loam soil pastes to predict nitrate concentrations in a laboratory environment
2 To apply wavelet analysis to the FTIR/ATR mid-infrared (mid-IR) spectra of the soils used by Linker et al (2004), and compare the performances of the PLS and wavelet approaches
3 To investigate the applicability of wavelet analysis to deconvolute FTIR/ATR mid-IR spectra of several soil types treated with nitrate fertilizers in-situ The main features that distinguish objective 3 from 1 and 2 are- (a) working directly with field soils containing nitrate and (b) working only with soil pastes on a 1:1
Trang 314 To investigate the feasibility of using wavelet analysis of near infrared (NIR) soil spectra to predict soil nitrate concentrations The majority of the work presented involves using wavelet analysis with mid-infrared soil spectra This objective applies the same procedure to the near-infrared region
5 To investigate the feasibility of using a few selected wavenumbers rather than a continuum to predict nitrate contents The motivation for this objective was that currently there is a lack of reasonably-priced mid-infrared spectrometers available for detecting the low levels of nitrate typically found in agricultural soils and fixed filter-based systems might be less expensive
Trang 32CHAPTER Il
Literature Review 2.1 Need for Nitrate Sensor
As previously mentioned, nitrogen, specifically in nitrate form, is a very
important nutrient required for crop growth However, the negative charge of the nitrate molecule results in high susceptibility of the molecule leaching to depths where plant roots can’t utilize it for growth Even worse, the nitrate molecule can move all the way down into the groundwater leading to environmental and health problems There are two conditions that must exist for nitrate leaching to occur: abundance of nitrate in the soil and sufficient rainfall or irrigation for drainage to occur The amount of nitrate in the soil depends on the amount of nitrate applied, the amount derived from mineralization,
amount lost due to nitrification, and the amount the plant takes up for growth The drainage conditions of the soil depend on the hydraulic conductivity of the soil, which in turn is related to the soil texture
Soils with abundant nitrate amounts and shallow groundwater tables pose a high risk to nitrate leaching into drinking water supplies The greatest concer of nitrate in groundwater is for infants less than one year old and for pregnant animals (Killpack and
Buchholz, 1993) The nitrate consumed is converted to nitrite which combines with
Trang 33Oregon in which 28 of 82 drinking water wells tested exceeded this amount Studies conducted in the Salinas Valley Watershed in California found 35% of the wells tested contained nitrate concentrations greater than 10 ppm (ALBA, 2005) Kachanoski (1984) found that the nitrate amounts in groundwater increased pound for pound with the extra nitrate applied once the optimum nitrogen fertilizer rate for that given plant and field conditions was exceeded
From an agricultural standpoint, nitrate leaching results in loss of a necessary nutrient for crop growth and thus, leads to an increased cost and/or decreased yield for the grower In many situations the soil may already possess more than adequate nitrate for crop production and additional application would only result in residual nitrate in the soil that is susceptible to leaching Also, nitrate may be applied to the soil at a stage where the crop is not utilizing it at a high rate When the crop reaches a stage where nitrate is consumed at a high rate, it may not be available due to leaching resulting in lower crop yields Regarding corm, its greatest nitrogen needs start between 30 to 45 days after emergence (Francis and Piekielek, 2004) If insufficient nitrogen is available to the plant during growth periods, the farmer must apply additional fertilizer to account for nitrate leaching losses to achieve yield goals This in turn leads to added costs for the grower Fuel and oil only account for approximately 30% of a typical farmer’s energy costs, while the remaining 70% stems from pesticide and fertilizer costs (Francl et al,
1998)
Trang 34knowledge regarding the amount of nitrate in the soil Traditionally, soil nitrate analysis is accomplished by collecting soil samples in the field, performing the necessary soil processing steps such as oven drying and grinding, and submitting the samples to a laboratory for further analysis (typical laboratory-based methods are covered in a subsequent section) This method is very time-consuming, tedious, and expensive Often, as many as fifty samples may be required to obtain a crude representation of the nitrate variation for a typical quarter section (160 acre) field Each sample collected should be a composite of many subsamples Franzen and Cihacek (1998) suggest the following number of subsamples for nitrate analysis based on the desired level of
accuracy
100 6
0
0 +5 +10 215 220
Level of acouracy (% deviation from mean)
Figure 2.1: Number of subsamples required to adequately represent a soil sample vs level of accuracy for the purpose of nitrate determination from
Franzen and Cihacek (1998)
Trang 3510
samples multiplied by 20 subsamples (for accuracy of 15%) yielding 3200 total soil samples to be collected The samples must then be chilled until nitrate analysis is performed, since nitrate 1s a highly volatile molecule Nitrate analyses in laboratories typically cost approximately $10 per sample including the oven drying and grinding To obtain nitrate concentrations on these 160 samples for this one field would cost
approximately $1600 plus the labor costs to collect the samples The cost of this
sampling procedure alone tends to discourage growers from performing nitrate analyses Rather, they tend to over-apply nitrate in order to prevent yield limitations due to nitrate deficiencies A method to allow for rapid and inexpensive analyses of nitrate contents in agricultural fields is needed In addition to the laboratory-based methods, there have been numerous procedures developed in the last 20 years for the purpose of determining nitrate in agricultural fields and many of these are discussed in the next section
2.2 Nitrate Measurement Techniques
2.2.1 Soil vs Plant
The main reason for sensing nitrate in fields is to determine how much, if any, fertilizer to apply to meet the needs of the plants and to prevent over-applying which can lead to nitrate leaching One may reason that the best way to determine the plant’s needs is to obtain information from the plant directly Stresses due to moisture and nitrogen
deficiencies cause the plant leaves to change color Grinenko (1987) discovered that
Trang 3611
nitrogen and chlorophyll contents, under a controlled condition where stress was due to nitrogen deficiency alone There are many commercial chlorophyll meters available which basically measure the greenness of the plant However chlorophyll meters are subject to variability resulting from changes tn light intensity from shade, cloud cover, and sun intensity Due to these concerns Knuse et al., 2004 cautioned - ‘‘while the chlorophyll content can be related to the nitrogen status in the plant, you should be careful basing fertility programs on these readings.”
There are two potential problems associated with using plant color to determine its nitrogen needs First, if a nitrogen deficiency is apparent, then it may be too late to apply fertilizer and expect a positive plant response resulting in increased yield The other major problem is that different plants exhibit different colors in response to similar nutrient deficiencies, 1.e there is a lack of nutrient deficiency-specific color Corn leaves become yellow-green in color if either nitrogen or moisture deficiencies exist This makes it difficult to distinguish the particular deficiency for a plant based solely on color Also, standards for multiple plants may be required as the same nutrient deficiencies manifest in varying colors
2.2.2 Laboratory-based Methods
Trang 37converting nitrate molecules to other ions, such as nitrite or ammonium, so that calibration procedures have to be used to relate the raw results back to nitrate concentrations
The analytical lab at UC-Davis utilizes the flow injection analyzer method (Wendt, 1999) for nitrate determination A copperized cadmium column is used to reduce nitrate to nitrite The nitnte is then determined by diazotizing with sulfanilamide and the product is coupled with N-(1-naphthyl) ethlyenediaminie dihydrochloride The absorbance at 520 nm is measured and related back to nitrate concentration through the use of a calibration equation This method has a detection limit of approximately 0.05 ppm NO;-N
A technique which involves determination of nitrate by ultraviolet detection at 214 nm is termed capillary electrophoresis This technique involves separation of compounds in a narrow tube by the use of electric current Since the velocity of an ion is a function of electrophoretic mobility and voltage applied by the source, ions are
separated by differences in solute velocity The mobility of an ion for a given medium is a constant and is determined by the force balance between the attraction of the ion to an oppositely charged electrode and the frictional forces in the medium trying to prevent the ion from moving (Yi, 1992) Almost all molecules can be separated using capillary electrophoresis Advantages of electrophoresis include small sample sizes needed, ability to separate hundreds of anions simultaneously, and can be easily automated This
Trang 3813
commonly used for analysis of liquid samples such as soil solutions As with capillary electrophoresis, the HPLC method 1s based on ion mobility The liquid sample is termed the mobile phase and is passed through a column containing a stationary phase The mobility differences among the ions in the mixture result in varying degrees of attraction to the stationary and mobile phases Due to mobility differences, separate bands are formed that contain distinct components Once these bands are formed, ultraviolet or infrared spectroscopy may be used to determine absorbances at specific wavelengths, which may then be related back to ion concentrations This method can easily be
automated and is applicable to a wide range of substances Detection limits of 0.05 ppm NO:-N are possible with HPLC
2.2.3 Electrochemical Methods
The use of nitrate-selective electrodes is based on electrochemistry The measurements are performed in an electrochemical cell containing a nitrate-selective electrode, a reference electrode, and the liquid sample to be analyzed The nitrate- selective electrode can consist of a membrane attached to an inert tube The membrane contains an ion exchanger which attracts nitrate ions For nitrate analysis, cell electrical potential depends on the nitrate concentration in the sample and increases as nitrate
concentration decreases To determine the nitrate content of a soil sample, an extract
must be obtained for analysis which is usually accomplished using deionized water as the extractant
Trang 3914
Problems with inadequate mixing of the soil and water were observed and the system did not produce repeatable results Also the calibration procedures were tedious and possibly inaccurate due to changing potentials on the electrodes Subsequently, Adsett et al (1999) redesigned the system and used a wood saw blade and conveyer to collect and
transport soil samples of known volume and density to the extraction unit However,
problems with unacceptable levels of signal noise and clogging up the extractor outlet due to plant residues were observed
Birrell and Hummel (1993) attempted to use ion-selective field effect transistors (ISFET) in a multi-ISFET sensor chip to measure soil nitrate The advantages of using ISFET chips include fast response times and low sample volumes needed Based on work done by Birrell and Hummel, Kim et al (2003) performed experiments using 1on- selective membranes for nitrate determination They successfully used nitrate-selective membranes to predict nitrate concentrations in manually obtained soil extracts with correlation coefficients of 0.9 and greater However at low nitrate concentrations (below 60 ppm NO}), the extracting solution significantly affected the selectivity and sensitivity of the membrane
Trang 4015
2.2.4 Near-Infrared Spectroscopy
The availability of inexpensive and durable NIR spectroscopic equipment has attracted many researchers to use this technique for the purpose of soil nitrate analysis The peaks due to nitrate that exist in the NIR range are overtones and combinations of the fundamental peaks found in the mid-IR region and tend to be much weaker However, using appropriate signal analysis techniques the nitrate peak may be separated from overlapping peaks due to carbonate and humic acid
Sudduth and Hummel (1993) designed a real-time spectrophotometer for sensing NIR reflectance of soils They used a circular variable spinning filter monochromator to obtain reflectance measurements in the 1650 nm to 2650 nm wavelength range Output from a broadband NIR source contacted the sample and reflectance was measured using a lead-sulfide detector Soil reflectance data obtained in a laboratory with this unit agreed well with data collected with a traditional research grade spectrophotometer Cho et al (1998) attempted to predict total, inorganic, and organic nitrogen contents in soil with NIR spectra They found that NIR spectra combined with multiple linear regression analysis has the potential to predict soil nitrogen content Shibusawa et al (2001) developed an on-line soil spectrophotometer for the purpose of correlating soil parameters such as organic matter and moisture content to NIR reflectance Optical fibers were used to connect the spectrophotometer sensor to the sampling interface located in the tillage shank of an implement The sampling interface was in contact with soil below the surface Although good correlation was found with electrical conductivity (R’=0.86), poor correlation was obtained with soil nitrate (R°=0.19) content Reeves et