SPECTRAL MEASUREMENT OF COMMON SOIL PHOSPHATES

I. Bogrekci, W. S. Lee

ABSTRACT. Compounds of Al, Fe, Ca, and Mg phosphates were particularly found in the Lake Okeechobee drainage basin
in Florida. Identification of the spectral characteristics of these compounds would improve sensing of phosphorus (P)
concentration in soil samples. This study investigated the effects of common soil P compounds on reflectance spectra of sandy
soils using ultraviolet (UV), visible (VIS), and near-infrared (NIR) reflectance spectroscopy. Pure sandy soil was leached to
remove all nutrients and organic matter. Effects of four different P compounds (CaPO4 , AlPO4 , FePO4 2H2 O, and
Mg3 (PO4 )2 2H2 O) at 0.0, 12.5, 62.5, 175.0, 375.0, 750.0, and 1000.0 mg kg −1 were investigated. Actual P concentrations of
the soil samples were analyzed. Reflectance of the samples was measured between 175 and 2550 nm with 1 nm intervals. The
highest absorbance peaks were found at 286, 2548, 2516, and 228 nm for FePO4 2H2 O, Mg3 (PO4 )2 2H2 O, CaPO4 , and AlPO4 ,
respectively. These wavelengths may be used for a calibration to detect or determine phosphates of soils. Correlation
coefficients, standard deviations, and first derivatives of absorbance spectra were computed. Wavelengths were selected using
a stepwise discriminant analysis to build calibration models for P prediction. Classification of P compounds was tested using
discriminant analysis. The results indicated that the Fe, Ca, Mg, and Al associated phosphates could be detected with a
classification error of 1.9%. Partial least squares (PLS) analysis results for the dry soil samples with four compounds yielded
R 2 of 0.48 to 0.75 and root mean square error (RMSE) of 27 to 43 mg/kg.
Keywords. Lake Okeechobee, NIR, P compounds, Phosphates, Phosphorus, Reflectance, Sensor, Spectroscopy, UV, VIS.

P
hosphorus (P) is commonly found in nature as phos- sensing different phosphate molecules. Janik et al. (1998)
phates. Corbridge (1990) stated that P chemistry is studied the diffuse reflectance analysis in comparison with
dominated by compounds containing phosphorus- soil extractions. They stated that recent developments in
oxygen linkages and these are termed as phosphates. infrared spectroscopy resulted in an increase in the potential
Nair et al. (1995) investigated the P forms in soil profiles for soil analysis. They also mentioned that infrared spectros-
from dairies in south Florida. They stated that most P derived copy in both the near- and mid-infrared ranges allowed rapid
from the Ca-Mg associated phosphates for the A horizon. acquisition of soil information at quantitative, qualitative, or
Brookes et al. (1982) measured P from microbial biomass indicator levels for use in agricultural and environmental
in soils. They found that most of the P released was in monitoring. Nyquist and Kagel (1971) investigated the
inorganic form, and the proportion increased with the infrared spectra of inorganic compounds. In particular, they
duration of fumigation. Tyler (2002) studied the P fractions listed characteristic frequencies for inorganic ions such as
in grassland soils. Total mineral (inorganic) P was calculated phosphide hypophosphite, orthophosphite, metaphosphite,
as the sum of P-Ca, P-Al, and P-Fe. In addition, total P was orthophosphates, pyrophosphate, phosphorothioate, phos-
calculated as the sum of mineral P and organic P. phorofluoridate, and phosphorodifluoridate.
Tyler (2002) studied the phosphorus fractions in grassland Dunn et al. (2002) investigated the potential of near-in-
soils. He mentioned that results of soil P fractionations are frared reflectance spectroscopy for soil analysis in southeast-
often related to general soil chemical properties, such as soil ern Australia. They suggested that near-infrared spectros−
acidity or weathering state. Calcium phosphate fraction was copy could successfully determine some soil properties in
high in soils of low acidity, and the concentration of iron (III) both the topsoil and subsoil. Russell et al. (2002) studied the
phosphate dihydrate was high in very acid soils. potential of near-infrared (NIR) spectroscopy to predict
Advances in spectroscopy have provided new methods to nitrogen mineralization in rice soils.
determine concentration of elements in chemistry. There- Lee et al. (2003) conducted research to estimate chemical
fore, it may be possible to use UV-VIS-NIR spectroscopy for properties of Florida soils using VIS and NIR spectroscopy.
They produced prediction models to estimate pH, organic
matter (OM), phosphorus (P), potassium (K), calcium (Ca),
and magnesium (Mg) concentrations of three representative
Article was submitted for review in October 2004; approved for soil orders such as Alfisol, Entisol, and Ultisol. Bogrekci et
publication by the Information & Electrical Technologies Division of al. (2003) studied the wet and dry reflectance spectra of soil
ASABE in September 2005. Presented at the 2004 ASAE Annual Meeting
as Paper No. 043114. samples collected from Okeechobee County in Florida and
The authors are Ismail Bogrekci, ASABE Member Engineer, correlated them with P concentrations of soil samples in order
Postdoctoral Research Associate, and Won Suk Lee, ASABE Member to produce a prediction model for P in soils.
Engineer, Assistant Professor, Department of Agricultural and Biological Currently, there are many chemical extraction methods in
Engineering, University of Florida, Gainesville, Florida. Corresponding
author: I. Bogrekci, Frazier Rogers Hall, Museum Rd., Gainesville, FL
determining the concentrations of soil P for various soil types
32611-0570; phone: 352-392-1864; fax: 352-392-4092; e-mail: (Bray and Kurtz, 1945; Nelson et al., 1953; Olsen et al., 1954;
bogrekci@ufl.edu. Carter, 1993; Mehlich, 1984; Pierzynski, 2000). The imple-

Transactions of the ASAE

Vol. 48(6): 2371−2378 E 2005 American Society of Agricultural Engineers ISSN 0001−2351 2371
mentation of these extraction methods is laborious, costly, measurement, soil samples were oven-dried at 104°C for
and time consuming. Therefore, the success of this investiga- 24 h. The soil samples were sent to a laboratory for chemical
tion into using spectroscopy to determine different P analysis of P concentrations. All soil samples were analyzed
compounds would be useful in designing a P sensor for for total P using the acid digestion method and inductively
agricultural and environmental use. coupled plasma (ICP).

OBJECTIVE REFLECTANCE MEASUREMENT

This study proposed first to detect the existence of soil A spectrophotometer (Cary 500, Varian, Inc., Palo Alto,
phosphorus compounds (CaPO4, AlPO4, FePO42H2O, and Cal.) equipped with an integrating sphere (DRA-CA-5500,
Mg3(PO4)22H2O) commonly found in the Lake Okeechobee Labsphere, Inc., North Sutton, N.H.) was used to collect
basin with the ultimate aim of designing a P sensor for that spectral reflectance for each soil sample. The instrument had
region, and second to determine the concentrations of each the wavelength accuracy of 0.020 nm for data interval and
P compound. Specific objectives were to: 0.20 nm for spectral bandwidth with an averaging time of
S Identify significant absorption bands and spectral sig- 0.033 s. Photometric errors of the instrument were less than
nature for each compound. 0.0003% and 0.003% transmission in ultraviolet-visible and
S Classify each P compound in soils. near-infrared, respectively. A 50 mm diameter polytetrafluo-
S Quantify P concentration of each compound. roethylene (PTFE) disk (Spectralon, Labsphere, Inc., North
Sutton, N.H.) was used to obtain the optical standard of the
system before spectral measurements. A 27.5 g soil sample
was placed into a sample holder, and the spectral signature for
MATERIALS AND METHODS each soil sample was collected using baseline correction
SOIL SAMPLE PREPARATION
mode. Reflectance for each soil sample with different P
In order to study the effects of different P compounds on compounds was measured in 175-2550 nm with an increment
reflectance spectra of soils, pure sandy soil was donated by
of 1 nm. In addition, reflectance for each P compound
a company located in Edgar, Putnam County, Florida. Sandy
(CaPO4, AlPO4, FePO42H2O, and Mg3(PO4)22H2O) was
soil was graded into three particle sizes using a sieve shaker measured in 175-2550 nm with an increment of 1 nm. Since
(Ro-Tap, W. S. Tyler, Inc., Mentor, Ohio). Sandy soil particle
spectrum of each soil sample had noise in 175-225 nm and
sizes were 125, 250, and 600 m for fine, medium, and
2525-2550 nm, only the spectra of soil samples between 225
coarse, respectively. A total of 112 medium-size sandy soil and 2525 nm were used. Reflectance of each compound had
samples were used in this study (table 1). Four replications of
noise in 175-208 nm; therefore, reflectance data in 208-
seven P application rates and four different P compounds
2550 nm were used for P compounds’ signatures. Reflec-
generated 112 samples for this study. Soil samples were tance of the soil samples was measured before and after
leached using 0.1 molar HCl acid solutions and de-ionized
drying.
water in order to remove existing P. After leaching, pH and
P concentration of the sandy soils were analyzed. Soil pH was
DATA ANALYSIS
measured using a pH/temperature meter (HI 991000, Hanna
Data Processing
Instruments, Woonsocket, R.I.). The pH meter accuracy was
Reflectances of all samples were converted into absor-
±0.01. Soil P was determined using a soil test kit (Luster Leaf
bance before further analysis in order to find a relationship
Products, Inc., Atlanta, Fla.). Soils were leached with
between P concentrations and absorbance of the samples at
de-ionized water to obtain pH of 6, if pH was detected lower
different wavelengths using Beer-Lambert’s law (Williams
than 6. Further leaching was applied if P was detected in the
and Norris, 2001). Absorbance was calculated using follow-
soil samples.
ing formula:
Different amounts of compounds were mixed into the soil
thoroughly, and the soil samples were wetted to field capacity Abs = Log(1/Ref) (1)
level (8% w.b. moisture content for Okeechobee county soils
in Florida). These compounds were CaPO4 (calcium phos- where Abs is absorbance and Ref is reflectance.
phate), AlPO4 (aluminum phosphate), FePO42H2O (iron (III) The data were filtered using the Savitzky-Golay polyno-
phosphate dihydrate), Mg3(PO4)22H2O (magnesium phos- mial convolution filter to remove the noise in the signal
phate dihydrate). All compounds were purchased in powder (Savitzky and Golay, 1964). Smoothing was controlled by the
form (Sigma-Aldrich, Inc., Milwaukee, Wisc.). The com- degree of polynomial (2) and the number of points (50). In
pound rates are given in table 1. Soil samples were incubated order to find the significant bands for each compound,
for seven days with all P compounds. After reflectance correlation coefficient and standard deviation spectra were

Table 1. Experimental treatments.

Particle Size Moisture Content Spectral Range P Compounds Application Rates
P Compounds (Μm) (% w.b.) pH (nm) (mg kg−1) (four replications)
No compounds 0.0
Very low 12.5
CaPO4 Low 62.5
AlPO4 250 8 6 175-2550 Medium 175.0
FePO42H2O High 375.0
Mg3(PO4)22H2O Very high 750.0
Extremely high 1000.0

2372 TRANSACTIONS OF THE ASAE

computed. Correlation coefficients were computed between Table 2. Compound application rates and average
absorbance and actual total P concentration of the soil actual total P concentration of soil samples.
samples. Higher correlation coefficients indicate that P Compound Average Actual
Application Rate Total P Concentration
concentrations of the samples could be estimated indirectly Compound (mg kg−1) (mg kg−1)
by measuring absorbance without any chemical analysis. In
0.0 11
addition, in order to observe the relationships between
12.5 18
absorbance and P compound concentration, first derivatives
Magnesium 62.5 23
of absorbance were computed. Standard deviations of
phosphate 175.0 35
absorbance were computed for soil samples at four P dihydrate 375.0 73
compounds with different concentrations. Data were divided 750.0 108
into two sets as calibration and validation. The calibration 1000.0 174
and validation data sets were obtained using simple random
0.0 21
sampling. The calibration data set had 56 soil samples, and 12.5 15
the validation data set had 53 soil samples. Three samples 62.5 23
Iron (III)
were discarded due to experimental error and outliers. The phosphate 175.0 36
SAS STEPDISC procedure was used to select important dihydrate 375.0 48
wavelengths and to reduce the number of independent 750.0 103
variables. The SAS REG and SAS PRINCOMP procedures 1000.0 139
were used to select and reduce the number of independent 0.0 14
variables (SAS, 1999). The main aim in computing the 12.5 16
correlation, standard deviation, and first derivative, and 62.5 22
conducting stepwise discrimination, stepwise multiple linear Calcium
175.0 32
regression, and principal component analysis was to reduce phosphate
375.0 58
the number of wavelengths before model building. 750.0 112
Statistical Analysis and Model Building 1000.0 153
In order to build robust prediction models, both calibra- 0.0 11
tion and validation data sets should be sampled using 12.5 9
statistical sampling methods. Each data set should represent 62.5 20
Aluminum
the whole population. During model building, the SAS 175.0 42
phosphate
375.0 73
DISCRIM procedure was used to classify each compound
750.0 125
(SAS, 1999). Wavelengths selected after correlation, step-
1000.0 168
wise discrimination, stepwise multiple linear regression, and
principal component analysis were used as independent
variables in the calibration and validation data sets for
discriminant analysis. The percentage error rate was com- are shown in figure 1. Iron (III) phosphate dihydrate had the
puted as the number of misclassified samples over the total highest absorbance values in the UV and VIS ranges. The
number of samples, multiplied by 100. The SAS PLS highest absorbance was at 286 nm for iron (III) phosphate
procedure was used to calibrate and predict P concentrations dihydrate. Absorbance values for iron (III) phosphate
of soils using spectral data (SAS, 1999). The same number of dihydrate sharply decreased from 286 nm to 667 nm. Iron
calibration and validation samples was used as in the data (III) phosphate dihydrate had three recognizable peaks at
processing step. The number of extracted factors was 867, 1464, and 1944 nm from visible to NIR. Absorbance
determined by cross-validation, that is, fitting the model to values for iron (III) phosphate dihydrate increased steadily
part of the data and minimizing the prediction error for the from 2152 to 2550 nm.
unfitted part. The predicted residual sum of squares (PRESS) Magnesium phosphate hydrate had lower absorbance
was used to determine the number of factors. The NIPALS values in the UV range. It showed almost the same
algorithm was used. For cross-validation, the split-sample 1.0
Iron (III) phosphate dihydrate
validation method was used. Magnesium phosphate hydrate
Calcium phosphate
0.8 Aluminum phosphate

RESULTS AND DISCUSSION

Absorbance

0.6
Compound application rates and average chemical analy-
sis results are given in table 2. The highest P concentration
for soil samples was 174 mg kg−1 and the lowest was 9 mg 0.4

kg−1. Although the samples were leached to remove P

concentration, reference method showed that soils still had P 0.2
in the concentration range from 10 to 21 mg kg−1. Since the
soil test kit showed no P while the reference method indicated
0.0
low concentration of P in the samples, these testing methods 500 1000 1500 2000 2500
seem to have different accuracies and sensitivities.
Wavelength (nm)
Average absorbance spectra for calcium phosphate,
aluminum phosphate, iron (III) phosphate dihydrate, and Figure 1. Average absorbance of four replications of the four P com-
magnesium phosphate hydrate in the UV-VIS-NIR regions pounds in 208-2550 nm.

Vol. 48(6): 2371−2378 2373

absorbance from 387 to 1122 nm. Since magnesium phos- extremely high P compound concentrations for UV-VIS-NIR
phate hydrate contains water molecules, two peak absor- (225-2525 nm). There was a positive linear relationship
bance values were detected in 1419 and 1938 nm. The highest between absorbance and P concentration of the wet sandy soil
absorbance values were observed between 1938 and samples with iron (III) phosphate dihydrate from 225 to
2550 nm. 365 nm. The lower the P concentration, the lower the
The calcium phosphate spectra showed three distinct absorbance was from 225 to 365 nm. The results from the
peaks at 283, 1439, and 1948 nm. The highest absorbance reflectance measurement of the wet sandy soil with iron (III)
was observed at 2516 nm for calcium phosphate. Aluminum phosphate dihydrate suggested that P concentrations of iron
phosphate showed lower absorbance values between 1374 (III) phosphate dihydrate could be determined using absor-
and 2550 nm. The highest absorbance values in 228 nm were bance in the UV region.
observed in the UV range for aluminum phosphate. Absorbance spectra for the wet sandy soils with calcium
The spectral measurement results for the wet sandy soil phosphate from zero to extremely high P compound con-
with magnesium phosphate hydrate at zero to extremely high centrations in 225-2525 nm are given in figure 2c. There were
P compound concentrations in the range of 225-2525 nm are three distinguishable peaks at 271, 1454, and 1936 nm. The
shown in figure 2a. The wet sandy soil with magnesium peaks at 1454 and 1936 nm indicate the water content of the
phosphate hydrate resulted in four recognizable peaks for all wet sandy soil with calcium phosphate.
P concentrations at 271, 1452, 1931, and 2206 nm. P The average absorbance spectra for the wet sandy soil with
concentration of the wet sandy soil with magnesium aluminum phosphate from zero to extremely high P com-
phosphate hydrate had a positive linear relationship with pound concentrations in 225-2525 nm are given in figure 2d.
absorbance in the UV range. In particular, high P concentra- The peaks at 1454 and 1936 nm indicate the water content of
tion of the wet sandy soil with magnesium phosphate hydrate the wet sandy soil with aluminum phosphate. For all
caused the dilution of absorbance spectra to almost medium wavelengths, extremely high concentration of aluminum
P concentrations in the NIR region. Water absorption bands phosphate in soil samples produced the highest absorbance
were also very distinct at 1452 and 1931 nm. values. On the other hand, zero concentration yielded lower
Figure 2b shows the average absorbance for the wet sandy absorbance values between 225 and 2525 nm.
soil samples with iron (III) phosphate dihydrate from zero to Figure 3a shows the average spectra for the dry sandy soil
with magnesium phosphate hydrate from zero to extremely

1.0 1.0
No Compound
Very low
Low
0.8 Medium 0.8
High
Very high
Absorbance

Absorbance

Extremely High
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
500 1000 1500 2000 2500 500 1000 1500 2000 2500
Wavelength (nm) Wavelength (nm)

(a) (b)
1.0 1.0

0.8 0.8
Absorbance

Absorbance

0.6 0.6

0.4 0.4
Absorbance

0.2 0.2

0.0 0.0
500 1000 1500 2000 2500 500 1000 1500 2000 2500

Wavelength (nm) Wavelength (nm)

(c) (d)

Figure 2. Average absorbance of wet soil samples with different concentrations of: (a) magnesium phosphate hydrate, (b) iron (III) phosphate dihy-
drate, (c) calcium phosphate, and (d) aluminum phosphate in 225-2525 nm. Each spectrum is an average of four samples.

2374 TRANSACTIONS OF THE ASAE

high P compound concentrations in 225-2525 nm. The centration of the Fe-associated P could be determined by
highest absorbance values occurred between the UV and VIS measuring absorbance in the UV region. Apart from the most
regions. Low peaks at 1414 and 1934 nm indicate traces of significant peak at 281 nm, four small peaks occurred at 871,
water in the dried soil sample. The other distinguishable 1413, 1934, and 2213 nm.
peaks were at 269 and 2211 nm. The results point out that Average dry sandy soil spectra with calcium phosphate
overall absorbance values of the dry sandy soil with from zero to extremely high P compound concentrations in
magnesium phosphate hydrate are lower than those of the wet 225-2525 nm are presented in figure 3c. Eight small peaks
sandy soil with magnesium phosphate hydrate. Comparing were observed for the dry soil with calcium phosphate at 268,
the average absorbance spectra between the wet samples 875, 1415, 1932, 2212, 2314, 2382, and 2500 nm. The highest
(fig. 2) and the dry samples (fig. 3), the spectral difference absorbance values were seen in the UV range. Absorbance
mainly comes from the water in the wet samples. Water is a values decreased sharply from 268 to 375 nm. Decreasing of
strong absorber and increases the overall absorbance. The absorbance values continued steadily from VIS to NIR.
most dominant water absorption bands are at 1450 and Figure 3d shows the average dry soil spectra with
1940 nm (Hruschka, 2001). The wet spectra in figure 2 show aluminum phosphate from zero to extremely high P com-
higher absorbance than those of the dry samples in fig. 3. The pound concentrations in 225-2525 nm. Dry soil sample with
wet spectra in figure 2 also show two main water absorption aluminum phosphate absorbed more light energy in the UV
bands at 1450 and 1940 nm, which do not exist in the dry and VIS ranges than in the NIR. The highest absorbance
spectra in figure 3. However, the spectral characteristics in (Abs = 0.44) was observed in the UV range. Seven low-am-
the UV region remain the same between the wet and dry plitude peaks in the NIR range and one high-amplitude peak
samples. in the UV range occurred.
The average spectra for the dry sandy soil with iron (III) Correlation coefficient spectra for the wet sandy soils with
phosphate dihydrate from zero to extremely high P com- four compounds with regard to P concentrations are present-
pound concentrations in 225-2525 nm are given in figure 3b. ed in figure 4. The highest correlation coefficient (r = 0.84 at
A positive linear relationship exists between absorbance and 317 nm) appeared in the UV range for iron (III) phosphate
P compound concentrations of the dry sandy soil with iron dihydrate, but correlation coefficients decreased in the VIS
(III) phosphate dihydrate, particularly in 225-490 nm. The and NIR regions. Correlation coefficients for aluminum
results from the reflectance measurement of dry sandy soil phosphate, calcium phosphate, and magnesium phosphate
with iron (III) phosphate dihydrate suggest that the con- hydrate did not vary much about the average correlation
0.7
No compounds
Very low 0.6
0.6
Low
Medium
High 0.5
Very high
Absorbance
Absorbance

Extremely high
0.4 0.4

0.3

0.2 0.2

0.1

0.0 0.0
500 1000 1500 2000 2500 500 1000 1500 2000 2500

Wavelength (nm) Wavelength (nm)

(a) (b)
0.7 0.7

0.6 0.6

0.5 0.5
Absorbance

Absorbance

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1

0.0 0.0
500 1000 1500 2000 2500 500 1000 1500 2000 2500

Wavelength (nm) Wavelength (nm)

(c) (d)

Figure 3. Average absorbance of dry soil samples with different concentrations of: (a) magnesium phosphate hydrate, (b) iron (III) phosphate dihy-
drate, (c) calcium phosphate, and (d) aluminum phosphate in 225-2525 nm. Each spectrum is an average of four samples.

Vol. 48(6): 2371−2378 2375

coefficients of 0.58, 0.21, and 0.52, respectively. The highest Close to the 2500 nm region, absorbance spectra (fig. 2)
correlation coefficients of the wet sandy soil samples were show positive linear relationship with P concentrations, with
0.62 for aluminum phosphate at 1213 nm, 0.26 for calcium one or two exceptions. These exceptions caused lower
phosphate at 1384 nm, and 0.58 for magnesium phosphate correlation coefficients for the correlation coefficient spectra
hydrate at 831 nm. Results of the correlation between the of the wet soils (fig. 4). When absorbance variation for iron
absorbance and P concentrations for the wet sandy soils with (III) phosphate dihydrate in the UV region (figs. 2b and 3b)
four different P compounds suggested that the concentrations is compared with the correlation coefficient spectra (figs. 4
of Fe, Mg, and Al associated P in pure sand soil can be and 5), the absorbance spectra have a positive linear
determined by developing calibration models for individual relationship with P concentrations without any exceptions.
compounds. Therefore, values for the correlation coefficient are observed
Figure 5 shows the correlation coefficient spectra for the to be higher.
dry sandy soils with four compounds with regard to P The standard deviation spectra for the wet and dry soils
concentrations. Dry sandy soil with iron (III) phosphate with Fe, Mg, Al, and Ca associated P compounds are given
dihydrate produced the highest correlation coefficient (r = in figure 6a and 6b. Standard deviations of absorbance for
0.93 at 343 nm) in the UV range. The highest correlation each sample with different concentrations and types of
coefficient between the wet and dry soils for iron (III) phosphates were computed in 225-2525 nm. Regions that
phosphate dihydrate noticeably shifted from 317 to 343 nm. have large positive peaks are regions where the spectra vary
This shift might be due to molecular structure change during the most. Higher standard deviations indicate wavelength
water loss. Mean correlation coefficient of calcium phos- regions, which can be utilized effectively for developing
phates for the dry soil increased from 0.21 to 0.41. The calibration models since they contain more variability than
highest correlation coefficients of the dry sandy soil samples other regions. The highest standard deviation was observed
were 0.68 for aluminum phosphate at 2278 nm, 0.54 for at 1922 nm for aluminum phosphate and at 274 nm for iron
calcium phosphate at 2500 nm, and 0.73 for magnesium (III) phosphate dihydrate for the wet and dry sandy soils,
phosphate hydrate at 2520 nm. respectively. There was a general trend of higher standard
deviation values with increasing wavelength for regions
1.0
more than 1000 nm in figure 6a.
0.84@317 Aluminum phosphate When absorbance variation for iron (III) phosphate
Calcium phosphate
Iron (III) phosphate dihydrate
dihydrate in the UV region (figs. 2b and 3b) is compared with
0.8
the standard deviation spectra (figs. 6a and 6b), figures 2b
Correlation coefficient (r)

Magnesium phosphate hydrate

0.62@1213
0.6
0.30
Aluminum phosphate
Calcium phosphate
0.4 Iron (III) phosphate dihydrate
0.25 Magnesium phosphate hydrate
0.58@831
Standard deviation

0.2 0.20

0.26@1384 0.15
0.0

0.10
−0.2
500 1000 1500 2000 2500
0.05
Wavelength (nm)

Figure 4. Correlation coefficient spectra between chemical results (P) and 0.00
their absorbance for wet sandy soil samples with four phosphorus com- 500 1000 1500 2000 2500
pounds. Wavelength (nm)

1.0
(a)
0.93@343 0.73@2520 0.10

0.8 Aluminum phosphate

Calcium phosphate
Correlation coefficient (r)

0.08 Iron (III) phosphate dihydrate

0.6
Standard deviation

Magnesium phosphate hydrate

0.4
0.06

0.2

0.04
0.0 0.54@2500
0.68@2278
−0.2 Aluminum phosphate 0.02
Calcium phosphate
−0.4 Iron (III) phosphate dihydrate
Magnesium phosphate hydrate
0.00
−0.6 500 1000 1500 2000 2500
500 1000 1500 2000 2500
Wavelength (nm)
Wavelength (nm)
(b)
Figure 5. Correlation coefficient spectra between chemical results (P) and
their absorbance for dry sandy soil samples with four phosphorus com- Figure 6. Standard deviation spectra for (a) wet and (b) dry sand soil sam-
pounds. ples with four phosphates.

2376 TRANSACTIONS OF THE ASAE

and 3b show the absorbance variation in the UV region, and procedure increased the number of selected wavelengths.
figures 6a and 6b agree with the absorbance variation by Although the classification error was reduced to 15.1%, a
yielding higher standard deviations. lower classification error (1.9%) was observed when the
Derivatives are often used to enhance the appearance of correlation coefficient was equal to 0.35 and stepwise
spectra (Williams and Norris, 2001). The first derivatives of discriminant analysis was used. The results in table 3
average absorbance of the dry soil samples with iron (III) indicated that when the number of independent variables
phosphate dihydrate in 225-2525 nm are shown in figure 7a. (number of selected wavelengths) was high, the classifica-
Figure 7b shows the first derivative of average absorbance of tion error increased, and when the number of independent
the dry soil samples with iron (III) phosphate dihydrate in variables was small, the classification error again increased.
225-450 nm. There were two higher-amplitude peaks at 257 Fifteen was the optimum number of wavelengths and
and 354 nm. At 257 nm, the amplitudes of absorbance produced lower classification errors.
presented a direct positive linear relationship with iron (III) Partial least squares analysis (PLS) results from the
phosphate dihydrate rates, whereas at 354 nm, the amplitudes calibration and validation data sets for soils with four P
of absorbance produced a direct negative linear relationship compounds are presented in table 4. The number of factors
with iron (III) phosphate dihydrate. was six. The results showed that PLS predicted P concentra-
Table 3 shows the classification results for the dry soil tion of soils with root mean square error (RMSE) ranging
samples with four compounds using absorbance in selected from 27 to 43 mg kg−1. Prediction performance of iron
wavelengths and considering the given criteria. Increasing phosphate seems to be reasonable, as the model validation
the threshold level of the correlation coefficient (r = 0.4) and produced R2 of 0.75 and RMSE of 27 mg kg−1. For the other
using the stepwise discriminant procedure reduced the three phosphates, there seems to be more room for improve-
number of selected wavelengths. However, this reduction in ment of the validation models, which yielded lower values of
the number of wavelengths caused a high classification error R2 and higher values of RMSE than those for iron phosphate.
(54.7%). Reducing the threshold level of the correlation Ultimately, the developed PLS models could be useful in
coefficient (r = 0.3) and using the stepwise discriminant designing a P sensor for different applications and could be
used to improve sensing of P concentrations in soil samples.

0.002
CONCLUSIONS
Absorbance spectra of the leached sandy soil samples
Absorbance first derivative

0.000
changed with added P compounds. The highest change in
absorbance values was observed in the UV region of the
electromagnetic spectrum for the sandy soil samples with
−0.002 No Compounds
iron (III) phosphate dihydrate. Iron (Fe) associated phos-
Very low phates showed absorption bands in UV regions of the
Low
Medium electromagnetic spectrum, which other investigated phos-
High
−0.004
Very high
phates in this study did not indicate.
Extremely high Maximum correlation coefficients of the dry sandy soil
samples were 0.93 for iron (III) phosphate dihydrate at
−0.006 343 nm, 0.68 for aluminum phosphate at 2278 nm, 0.54 for
500 1000 1500 2000 2500

Wavelength (nm)
Table 3. Classification results for the dry
(a) soil samples with four compounds.
Number of Selected
0.002 Wavelengths or Error
Criteria Principal Components (%)
Absorbance first derivative

Correlation coefficient >0.3 and

0.000 stepwise discriminant analysis 51 15.1
Correlation coefficient >0.35 and
stepwise discriminant analysis 15 1.9
−0.002 Correlation coefficient >0.4 and
stepwise discriminant analysis 1 54.7
Stepwise multiple linear regression
−0.004 (SMLR) 52 32.1
Principal component analysis 5 18.5

−0.006 Table 4. PLS results for the dry soil samples with four compounds
250 300 350 400 450 (RMSE is in mg kg−1, and the number of factors was six).
Wavelength (nm) Calibration Validation
(b) P Compound R2 RMSE R2 RMSE
CaPO4 0.72 29 0.63 31
Figure 7. First derivative of average absorbance of dry soil samples with FePO42H2O 0.76 23 0.75 27
different concentrations of iron (III) phosphate dihydrate in (a) 225-
AlPO4 0.68 38 0.48 40
2525 nm and (b) 225-450 nm. Each spectrum is an average of four sam-
ples. Mg3(PO4)22H2O 0.93 17 0.56 43

Vol. 48(6): 2371−2378 2377

calcium phosphate at 2500 nm, and 0.73 for magnesium Hruschka, W. R. 2001. Chapter 3: Data analysis: Wavelength
phosphate hydrate at 2520 nm. selection methods. In Near-Infrared Technology in the
Discriminant analysis results showed that the existence of Agricultural and Food Industries, 39-58. 2nd ed. P. Williams
Fe, Mg, Al, and Ca associated P in a sandy soil sample can and K. Norris, eds. St. Paul, Minn.: American Association of
Cereal Chemists.
be estimated with a classification error of 1.9% using
Janik, L. J., R. H. Merry, and J. O. Skjemstad. 1998. Can
reflectance spectroscopy. In addition, PLS predicted P mid-infrared diffuse reflectance analysis replace soil extractions?
concentration of soils with root mean square error (RMSE) Australian J. Exp. Agric. 38(7): 681-696.
ranging from 27 to 43 mg kg−1. Lee, W. S., J. F. Sanchez, R. S. Mylavarapu, and J. S. Choe. 2003.
Estimating chemical properties of Florida soils using spectral
ACKNOWLEDGEMENTS reflectance. Trans. ASAE 46(5): 1443-1453.
The authors would like to thank to Drs. Donald A. Graetz, Mehlich, A. 1984. Mehlich 3 soil test extractant: A modification of
Vimala D. Nair, and Willie G. Harris of the Department of Mehlich 2 extractant. Comm. Soil Sci. Plant Analysis 15(12):
Soil and Water Science at the University of Florida for their 1409-1416.
Nair, V. D., D. A. Graetz, and K. M. Portier. 1995. Forms of
invaluable suggestions about experimental design and soil
phosphorus in soil profiles from dairies of South Florida. SSSA
properties. This research was supported by the Florida J. 59: 1244-1249.
Agricultural Experiment Station and a grant from the Florida Nelson, W. L., A. Mehlich, and E. Winters. 1953. The development,
Department of Agriculture and Consumer Services, and evaluation, and use of soil tests for phosphorus availability.
approved for publication as Journal Series No. R-10447. Agronomy 4: 153-188.
Nyquist, R. A., and R. O. Kagel. 1971. Infrared Spectra of
Inorganic Compounds. New York, N.Y.: Academic Press.
Olsen, S. R., C. V. Cole, F. S. Watanabe, and L. A. Dean. 1954.
REFERENCES Estimation of available phosphorus in soils by extraction with
Bogrekci, I., W. S. Lee, and J. Herrera. 2003. Assessment of P sodium bicarbonate. Circular No. 939. Washington, D.C.:
concentrations in the Lake Okeechobee drainage basins with USDA.
spectroscopic reflectance of VIS and NIR. ASAE Paper No. Pierzynski, G. M. 2000. Methods for phosphorus analysis for soils,
031139. St. Joseph, Mich.: ASAE. sediments, residuals, and waters. Southern Cooperative Series
Bray, R. H., and L. T. Kurtz. 1945. Determination of total, organic, Bulletin No. 396. Southern Association of Agricultural
and available forms of phosphorus in soils. Soil Sci. 59(1): Experiment Station Directors (SAAESD).
39-45. Russell, C. A., J. F. Angus, G. D. Batten, B. W. Dunn, and R. L.
Brookes, P. C., D. S. Powlson, and D. S. Jenkinson. 1982. Williams. 2002. The potential of NIR spectroscopy to predict
Measurement of microbial biomass phosphorus in soil. Soil Biol. nitrogen mineralization in rice soils. Plant and Soil 247(2):
Biochem. 14(4): 319-329. 243-252.
Carter, M. R. 1993. Soil Sampling and Methods of Analysis. Boca SAS. 1999. SAS/STAT User’s Guide. Ver. 8. Cary, N.C.: SAS
Raton, Fla: Lewis Publishing. Institute, Inc.
Corbridge, D. E. C. 1990. Studies in Inorganic Chemistry: Savitzky, A., and M. J. E. Golay. 1964. Smoothing and
Phosphorus, An Outline of Its Chemistry, Biochemistry, and differentiation of data by simplified least squares procedure.
Technology. 4th ed. New York, N.Y.: Elsevier. Anal. Chem. 36(8): 1627-1638.
Dunn, B. W., H. G. Beecher, G. D. Batten, and S. Ciavarella. 2002. Tyler, G. 2002. Phosphorus fractions in grassland soils.
The potential of near-infrared reflectance spectroscopy for soil Chemosphere 48(3): 343-349.
analysis: A case study from the Riverine Plain of southeastern Williams, P., and K. Norris. 2001. Near-Infrared Technology in the
Australia. Australian J. Exp. Agric. 42(5): 607-614. Agricultural and Food Industries. 2nd ed. St. Paul, Minn.:
American Association of Cereal Chemists.

2378 TRANSACTIONS OF THE ASAE