Abstract
Based on the voltage vs. time data of the sensor it appears as if both soil OMC and the nitrate level of the soil have an effect on the time required for the sensor to stabilize. On average, for all samples, the time for the sensor output voltage to fall within 5% of the stable value (at 30 sec) was 6.95 seconds. Only 49% of the readings were stable after 5 seconds. 74% of all samples stabilized within 10 seconds and 90% were stable after 15 seconds. The testing order and nitrate level of the previous sample had very little or no effect on the time required for the voltage to stabilize. It was not possible to determine whether OMC had an effect on the accuracy of the sensor because of a delay in receiving results from laboratory nitrate tests.
Introduction
With the expanding role of precision farming in agriculture, rapid soil analysis is becoming more and more important. Several sensors have been developed which allow various soil nutrients to be analyzed quickly. One of these sensors is the Horiba Cardy nitrate specific electrode. This sensor was successfully used to measure the nitrate-nitrogen concentration in petiole sap (Hartz 1993). However, using the Cardy meter to measure nitrate-nitrogen in soil may present additional problems. Soil nutrient analysis by way of ion specific electrodes is somewhat problematic due to the large number of variables associated with soils (i.e., texture, moisture, nutrient levels, organic matter). It has been difficult to obtain consistent results from ion-specific electrodes because of the effects of organic matter on nitrogen availability (Beaton 1985). This study was designed to evaluate the effect of soil organic matter content on the accuracy and response time of the Cardy nitrate meter.
Materials and Methods
Four soils were selected which varied in organic matter content but were of similar texture. After the soils were selected the organic matter content of the soils were determined using a previously developed reflectance-bases organic matter sensor (Shonk 1990). These four soils (soils A, B, C, and D) are listed in Table 1 along with their respective textures and organic matter content (OMC).
Table 1. Soils used for nitrate analysis
Soil Texture %Sand %Silt %Clay %OMC A Silty Clay 10.0 49.4 40.6 6.5 B Silty Clay Loam 11.1 55.3 33.6 4.1 C Loam 51.4 33.0 15.6 4.4 D Silty Clay Loam 9.3 54.2 36.5 5.3
After the soils were selected, the sensor was then calibrated using both the 450 ppm-N and a 20 ppm-N standard solutions that are shipped with the sensor. These standard solutions also contain aluminum sulfate to help reduce interference from other ions in the soil solution. Six nitrate levels were tested for each soil sample. The samples randomly were assigned numbers corresponding to the order in which they were to be tested. Table 2 lists the soils and their testing order. Prior to this study sample of soils B-D with varying nitrate levels had already been prepared by adding potassium nitrate. These samples were found in plastic bags which were labeled with the soil name and the nitrogen level of the soil.
Table 2. Soil Testing Order and Cardy Readings
Sample Label Cardy Reading Calibration Check
Number Soil ppm-N Trial 1 Trial 2 Trial 3 450 ppm 20 ppm Recalibration
1 C 50 39 43 44 450 22
2 B 30 37 37 38 470 23
3 D 10 39 41 41 460 21
4 B 40 43 45 44 500 20 *
5 C 0 20 21 23 440 22
6 A 30 59 58 58 450 22
7 D 20 48 49 51 480 21 *
8 B 20 27 30 31 430 22
9 D 0 29 30 30 440 21
10 C 30 34 35 33 450 21
11 D 40 56 58 59 550 22 *
12 A 40 63 66 68 530 21 *
13 D 30 47 48 47 460 20
14 A 20 43 47 48 490 20 *
15 D 50 51 50 51 430 21
16 C 40 35 36 35 460 21
17 B 10 26 28 29 510 20 *
18 C 20 27 28 27 430 20
19 A 0 22 23 24 470 21
20 B 0 22 24 24 460 22
21 C 10 25 26 27 470 21
22 A 50 75 76 78 500 20 *
23 B 50 38 41 42 460 21
24 A 10 33 34 35 470 22
It was necessary to prepare samples of soil A with a range of nitrogen levels using the following procedure: Samples of soil A were prepared by first weighing 10 g of soil into a small beaker. Then 25 mL of 20 ppm-N extractant was added to each sample. In addition a small amount of 450 ppm-N was added to the solution to increase the total nitrate level of the solution. The goal was to add enough 450 ppm-N standard to produce six solution with added nitrate levels of 0, 10, 20, 30, 40, and 50 ppm-N. The following volumes of 450 ppm-N were added using a micro-pipette: 0 mL, 0.57 mL, 1.16 mL, 1.79 mL, 2.44 mL, and 3.13 mL. Due to an error in calculation, rather than increasing the nitrate levels of the solutions by 0, 10, 20, 30, 40, and 50 ppm-N, the nitrate levels were increased by 0, 9.6, 19.1, 28.7, 38.2, and 47.8 ppm-N, respectively. After the both extractant and the necessary amount of 450 ppm-N standard had been added, the samples were stirred using a magnetic stirrer for approximately one minute. The samples were then allowed to stand for three or four minutes so that most of the soil particles would settle out of the solution. The samples were then filtered through number 2 filter paper. This filtrate was then used when making all nitrate measurements with the Cardy meter.
The preparation of samples of soils B, C, and D was somewhat different. Because these samples had already been prepared at various nitrate levels for use in another experiment, it was not necessary to add the 450 ppm-N standard solution as it was with soil A. 10 g of soil was weighed into a small beaker and 25 mL of 20 ppm-N extractant was added. The slurry was then stirred for approximately one minute with a magnetic stirrer and then filtered as above. Again the filtrate was used for making all nitrate measurements.
After the samples had been filtered, the Cardy electrode was rinsed with de-ionized water for approximately 5 seconds. The electrode was then briefly rinsed with the filtrate to ensure that all DI water had been removed. After rinsing, more filtrate was placed on the electrode and the reading was allowed to stabilize for approximately 60 seconds. The sensor output tended to drift within a range of 8-10 ppm. The average value of this range was recorded. While waiting for the reading to stabilize, the voltage output of the sensor was logged using an IBM PSL unit. After the first reading was taken the sensor was again rinsed with DI water for 5 seconds and two more readings were taken. The calibration of the sensor was checked after the three readings had been taken. If the calibration had drifted considerably, a recalibration was preformed before testing the next sample. The data obtained for each sample is listed in Table 2. The values listed in Table 2 have not been corrected for dilution, nor has the 20 ppm-N present in the extractant been taken into account (refer to Table 3 for corrected values).
In order to determine the actual nitrate levels of the soil, samples were sent to the Agronomy Department of Purdue University to be analyzed with a Lachet analyzer. Samples of each soil and each nitrate level were sent with the exception of soil A. Only one sample of soil A was sent. This sample had no added nitrogen.
Results and Discussion
The readings obtained from the Cardy using the above method need to be adjusted for two reasons: the extractant contains 20 ppm-N and the solutions are diluted by 2/5. Table 3 contains the average reading from the Cardy for each soil tested, as well as the corrected reading. The corrected value was obtained by first subtracting 20 ppm-N from the reading and then multiplying the result by 2.5.
Due to a delay in testing in the Agronomy Department, actual nitrate levels for the samples were not obtained. Because of this it was not possible to determine if OMC has an effect on the nitrate readings as reported by the Cardy sensor. However, once the nitrate levels are obtained from the laboratory, the data can be analyzed in the following manner: A graph of the actual nitrate levels versus the nitrate levels obtained from the Cardy (Table 3) should be made for each soil. A line should then be fitted to the data. From the regression the slope and intercept of the line can be obtained. If the Cardy sensor were 100% accurate the line would have a slope of 1, and an y-intercept of 0. The slopes and intercepts of the lines for the four soils can be compared to determine if the organic matter content of the soil has an effect on the nitrate readings from the Cardy sensor.
Table 3. Corrected Nitrate Readings (Average).
Soil Label Sample Corrected Reading
ppm-N Number ppm-N
A 0 19 7.5
A 10 24 -
A 20 14 -
A 30 6 -
A 40 12 -
A 50 22 -
B 0 20 8.3
B 10 17 19.2
B 20 8 23.3
B 30 2 43.3
B 40 4 60.0
B 50 23 50.8
C 0 5 3.3
C 10 21 15.0
C 20 18 18.3
C 30 10 35.0
C 40 16 38.3
C 50 1 55.0
D 0 9 24.2
D 10 3 50.8
D 20 7 73.3
D 30 13 68.3
D 40 11 94.2
D 50 15 76.7
In order to determine if organic matter has an effect on the time it takes for the sensor reading to stabilize, voltage versus time data was collected for 60 seconds for each sample tested. The time at which the voltage output was within 5% of the voltage output at 30 seconds was determined. The instructions for the Cardy sensor suggest that the reading be allowed to stabilize for 15-30 seconds, therefore, it was assumed that the voltage output at 30 seconds could be considered stable. The time for each sample to stabilize can be found in Table 4.
After the times in Table 4 had been determined four graphs were plotted. Figure 1 is a plot of time to stabilize vs. soil type. From this graph it appears as if soils A and D which have high OMC (Table 1) have longer stabilization times than do soils B and C with lower OMC. Table 5 contains the average time to stabilize for each soil. Soil A (OMC = 6.5%) on average
Table 4. Time for Nitrate Readings to Stabilize
Soil Label Sample Number Time to Stabilize (sec)
ppm-N Trial 1 Trial 2 Trial 3
A 0 19 0.5 13.6 7.5
A 10 24 0.5 10.25 14.0
A 20 14 9.7 12.0 15.0
A 30 6 4.0 5.75 8.0
A 40 12 2.1 14.5 18.4
A 50 22 14.0 21.2 11.0
B 0 20 5.5 0.5 1.5
B 10 17 3.3 9.0 0.9
B 20 8 15 2.8 5.8
B 30 2 6.0 6.0 6.8
B 40 4 3.0 4.3 3.0
B 50 23 15.6 3.1 3.5
C 0 5 8.7 1.7 0.9
C 10 21 4.7 3.1 2.1
C 20 18 1.4 0.7 1.2
C 30 10 9.0 4.9 3.3
C 40 16 2.9 1.5 0.8
C 50 1 15.4 4.0 4.6
D 0 9 8.4 10.0 12.6
D 10 3 9.3 4.7 1.0
D 20 7 12.1 6.0 5.3
D 30 13 2.4 15.7 16.8
D 40 11 3.7 2.9 14.1
D 50 15 5.6 16.3 5.0
took the longest time to stabilize while Soil C (OMC = 4.4%) took considerably less time to stabilize. However, because of the large variation in stabilization times from one trial to the next it is difficult to state with a large amount of certainty that the different organic matter levels of the soils are the cause of the variation in the time required for the readings to stabilize.
Table 5. Average time to stabilize for each soil type
Soil Type OMC Average time to
(%) Stabilize (sec)
A 6.5 10.11
B 4.1 5.31
C 4.4 3.94
D 5.3 8.44
A second graph, time to stabilize vs. testing order was also plotted to determine if the order in which the samples were tested had an effect on the time for the readings to stabilize. On the first day of testing samples 1-10 were tested, the second day 11-16 were tested and on the final day, 17-24 were tested. It does not appears from Figure 2 that the time required for the samples to stabilize varied significantly from day to day. Table 6 contains the average time to stabilize for each day.
Table 6. Time to stabilize for each day
Day Average time to Stabilize
(sec)
1 6.41
2 8.85
3 6.19
Figures 3 and 4 were plotted to determine if the nitrate level of the soil had any effect on the time it took for the reading to stabilize. Figure 3 is a plot of the time to stabilize versus the nitrate level which was written on the bags of soil. It would have been best to plot the time to stabilize vs. the actual nitrate levels of the soils, but the actual nitrate levels were not available. Figure 3 shows that there is perhaps a slight effect of nitrate levels on the time it takes for the samples to stabilize. Table 7 contains the average time to stabilize for each nitrate level. Table 7 shows that high nitrate level soils, on average, take longer to stabilize than to low nitrate level soils. Prior to the beginning of testing, it was observed that the nitrate level of the sample which was previously on the sensor had an effect on the output of the sensor. For example, if the previous nitrate level was high, the reading for the current sample would start high and gradually decline until it stabilized. The reverse is true if a low nitrate level sample was on the sensor prior to the current sample. Figure 4 is a plot of the time to stabilize vs. the nitrate level of the previous sample. There appears to be no relation between the time to stabilize and the nitrate level of the previous sample.
Table 7. Average time to stabilize for each nitrate level
Soil Label Average Time to Stabilize
ppm-N (sec)
0 5.95
10 5.24
20 7.25
30 7.39
40 5.93
50 9.94
In order to help determine the minimum time necessary to get a reading from the sensor which is within 5% of the stabilized reading a histogram was plotted. Figure 5 is a plot of the distribution of the times required for stabilization for the 72 readings taken. Because the maximum time to stabilize was 21.2 seconds, the time to stabilize was divided into the following 5 ranges: 0-5 sec, 5-10 sec, 10-15 sec, 15-20 sec, and >20 sec. Table 8 contains the number of samples which stabilized within each range as well as the percentage of the total number of samples which stabilized in a time less than the maximum value of the range. Nearly 74% of the samples yielded a stable reading after 10 seconds, while the number increased to 90% after 15 seconds. Table 8 contains the ranges and number of sample in each range.
Table 8. Stabilization Time Distribution
Range (ppm-N) No. of Samples in Range % of Total below max. of range
0-5 35 48.6
5-10 18 73.6
10-15 12 90.3
15-20 6 98.6
>20 1 100.0
In the early phases of this study, many problems were encountered with the Cardy sensor. Initially, the soil samples were prepared using and de-ionized water extractant and were not filtered before testing. When testing these samples it was found that the sensor gave inconsistent results and would not maintain its calibration. Because of this problem, experimentation with different methods of sample preparation was performed to determine which method gave the most consistent results. A soil sample was selected and then prepared using a variety of different methods. The sensor was calibrated, and then checked once before placing the sample on the sensor. Three nitrate readings were recorded for each sample. After the nitrate readings had been taken, the calibration was again checked. The results of this experiment can be found in Table 9. It was found that the best results were obtained if soil samples were prepared with the 20 ppm-N Cardy extractant containing aluminum sulfate.
Table 9. Cardy Sample Preparation Analysis
Soil Preparation* Calibration** Nitrate Reading Calibration Check
440 44 440 44 Trial 1 Trial 2 Trial 3 440 44
ppm-N ppm-N ppm-N ppm-N ppm-N ppm-N
B30 D,U 400 57 400 57 54 110 140 560 80
B30 D,U,I 450 44 460 44 78 110 120 480 47
B30 D,F,I 480 47 510 51 52 50 51 300 31
D10 D,U 450 44 440 40 550 270 450 590 57
D10 D,F 450 46 460 45 36 44 44 5000 180
D10 C,U 460 22 500 22 64 67 75 590 27
D10 C,F 460 21 440 21 39 40 40 440 21
The nitrate readings from the sensor also tended to drift within a range of 8-10 ppm-N The cause of this drift was not detemined. The readings that were recorded were the average value of the range within which the sensor drifted. Similarly, it was found that if the sensor was touched on the side, near the LCD, or if a small downward pressure was applied to the top of the sensor, the readings would jump by nearly 100 ppm-N. Care was taken to avoid touching the sensor while readings were being obtained.
References
Beaton, James D., Werner L. Nelson and Samuel L. Tisdale. Soil Fertility and Fertilizers. New York: Macmillan, 1985.
Hartz, T.K., et. al. On Farm Monitoring of Soil and Crop Nitrogen Status by Nitrate-Selective Electrode. Commun. Soil Sci. Plant Anal. 24 (1993): 2607-15.
Shonk, Jason Lee. Development of a Real-Time Organic Matter Sensor For Agricultural Equipment. Diss. Purdue University, 1990.