Temporal Comparisons of Apparent Electrical Conductivity: A Case Study on Clay and Loam Soils in Mississippi ()
1. Introduction
Proximal or on-the-go sensors measuring soil electrical conductivity have shown good relationships with soil properties affecting crop production, including texture [1] [2], cation exchange capacity [3] [4], soil water content [3] [5], drainage condition [6], salinity [7] [8] [9], and subsoil characteristics [8] [10] [11]. The systems have shown promise in improving and developing soil boundaries because of their links to the global positioning system [2] [12]. Their popularity is increasing because of the various configurations available for integrating them with farm and commercial equipment [8] [9]. They can acquire large amounts of data cheaply and more quickly than manual surveys, thus, providing more detail on the spatial variability of an area [13]. Other systems such as ground penetrating radar [14], multispectral and hyperspectral imagery [6] [14], and time domain reflectometry [15] have shown potential to map soil spatial variability. Nevertheless, apparent electrical conductivity (ECa) has been evaluated more than any other technique for studying soil spatial patterns [9] and has been recognized as a valuable tool to study soil spatial variability in agricultural settings [7] [16]. At the current rate, apparent electrical conductivity (ECa) systems will be the staple equipment for mapping soil spatial variability now and in the immediate future.
Yet, a better understanding of the stability of ECa measurements is needed so that users of the equipment have insight into the temporal stability of the readings. Using electromagnetic induction or the Veris mobile platform [17] [18] [19] [20] reported ECa stability ranging from 2 months to 4 years. Also, more information is needed on comparing ECa maps derived from over a period of time because the information would be valuable to farmers, consultants, and researchers interested in using the equipment to make management decisions.
Mississippi, USA, consists of various soils with different characteristics. Research is lacking in this area on the stability of apparent electrical conductivity measurements over time on soils used for agricultural production. The objective of this study was to compare spatial patterns of ECa data collected at two different periods to determine the temporal stability of map products derived from the data. The study focused on data collected in 2016 and 2021 from a field plot consisting of clay and loam soils in Mississippi.
2. Materials and Methods
2.1. Study Area
The study was conducted at the United States Department of Agriculture, Agriculture Research Service Farm (−90.872157 Longitude, 33.446486 Latitude), near Stoneville, Mississippi, USA. The average precipitation and temperature were approximately 133 cm and 17.5˚C, respectively [21]. The field plot was 4.6 ha and consisted of the following soil types: Commerce silty clay loam, 0% to 2% slopes (Ch), Commerce very fine sandy loam, 0% to 2% slopes, Sharkey clay, 0.5% to 2% slopes, and Tunica clay, 0% to 2% slopes [22]. The field was in a continuous soybean (Glycine max L.) and corn (Zea mays L.) rotation. The plot was subjected to the standard agricultural practices of the area related to irrigation, weed treatment, and fertilization.
2.2. Data Collection
Apparent electrical conductivity readings were collected with the Veris MSP 3 (Veris Technologies, Salina, KS, USA, Figure 1) system. It collected shallow (0 - 30 cm) and deep (0 - 90 cm) measurements, representing the topsoil and subsoil. The ECa system was moved through the field by a tractor. It used six coulters to penetrate the soil surface to a 6 cm depth. The coulters work in pairs with distinct functions. Coulters two and five injected the electrical current into the soil; coulters three and four recorded the EC shallow readings; coulters one and six recorded the deep readings. The sensor’s output was in millisiemens (mS) per meter. A Garmin global positioning system recorded each measurement’s latitude and longitude coordinates (WGS84). It recorded the location information when receiving differential global positioning data. A laptop computer inside the tractor’s cab served as the data logger for the system. On March 29, 2016, and April 22, 2021, data were collected from 19 transects separated by 8 m within the field. The data were collected from bare soil.
2.3. Data Analysis
Post-processing of the data included assigning each measurement an identification number, changing the longitude and latitude coordinate information to the UTM coordinate system (UTM 15N, WGS84), and cleaning the data (i.e., removal of negatives values, duplicated x-y coordinates, and outliers). Assigning point identification numbers, converting the latitude and longitude values, and
Figure 1. Veris MSP3 implement and tractor.
removing negative values and duplicate x-y coordinates were accomplished with QGIS (version 3.18.3-Zürich [23]).
Then, the data were transferred to the R software (R version 4.1.0, “Camp Pontanezen,” [24]) to calculate histograms, boxplots, and descriptive statistics. That information was used to identify outliers and better understand the datasets. After the initial cleaning process, 1998 and 1765 data points were assessed, respectively, for analysis of the 2016 and 2021 datasets. Descriptive statistics (i.e., mean, median, minimum, maximum, and coefficient of variation) and the paired t-test (P < 0.05) were used to compare the differences between the shallow and deep ECa measurements. Pearson correlation coefficients (P < 0.05) were tabulated to evaluate the relationship between ECa measurements and between the ECa measurements and the x and y location coordinates. The relationship between the x and y coordinates and ECa would give some insight into the directional trends in the dataset.
The cleaned data were clustered using the attributes clustering plug-in of QGIS. The following parameters were used for clustering: 1) method-k-means, 2) the number of times to repeat the classification-20, and 3) the threshold-0.00001. The clusters’ summary statistics (i.e., mean, median, minimum, and maximum values) were determined with the QGIS prepared initial statistical summary module. The cluster summary statistics were employed to assign a cluster to ECa zones ranging from low to high. Note: the low to high assignment was based on the data obtained from this field. The final maps displayed in the figures were created with the QGIS software.
3. Results
The descriptive statistics are summarized in Table 1 for both years. The ECa shallow mean (t = 117.85, df = 1997, P < 0.05, 2016; t = 106.01, df = 1764, P < 0.05, 2021), median, minimum, and maximum values were less than the ECa deep mean, median, minimum, and maximum values. The ECa shallow readings were more variable than the ECa deep readings according to the coefficient of variation values. Statistically significant positive correlation coefficients (Table 2)
Table 1. Descriptive statistics of apparent electrical conductivity shallow (ECas) and deep (ECad) readings of the study site collected in 2016 and 2021.
an = number of samples, Min = minimum, Max = maximum, and CV = coefficient of variation; for each year, mean values followed by a different letter represent statistical significance at P < 0.05 according to paired t-test.
Table 2. Pearson correlation analysis between apparent electrical conductivity shallow (ECas) and deep (ECad) measurements and x (x cor) and y (y cor) coordinates.
an = number of samples, *statistically significant at P < 0.05.
were observed between each measurement data’s ECa shallow and deep readings. Also, both ECa measures had a strong positive relationship with the y-coordinate, indicating a linear trend in the dataset based on direction.
Figure 2 and Figure 3 illustrate the map derived from clustering the ECa shallow and deep values collected in 2016 and 2021, respectively. The lowest values were observed in the southeastern corner of the field. In contrast, the highest values were detected in the northern section of the plot. Low-Medium, Medium, and Medium-High values occurred from south to north in the field. A noticeable trend was observed on the maps; the values were similar in the southwest to the northeast direction (the other way around) and more variable in the south to the north direction (the other way around).
Descriptive statistics derived from the ECa readings of the clusters are presented in Table 3. The clustering algorithm assigned the lowest number of points to the lowest zone for each dataset. These points were clustered in the southeastern section of the field (Figure 2 and Figure 3). The second-highest number of points was assigned to the Medium-High to High zones established by the clustering algorithm for this field. These areas were primarily in the northern part of the field (Figure 2 and Figure 3). The difference between the low and high clusters means was approximately 67.2 mS·m−1.
4. Discussion
The descriptive statistics patterns were consistent between the ECa shallow and deep values measured in 2016 and 2021. That pattern was higher mean, median, minimum, and maximum values for the ECa deep readings. Others documented the same pattern for soil ECa shallow and deep readings in The Republic of Trinidad and Tobago [20], Belgium [25], Canada [26], and Spain [27]. Also, these findings agreed with the findings of [20] [28], who observed positive, statistically significant correlations between ECa shallow and deep measurements. The correlation values between ECa shallow and deep values were greater than 0.80,
Figure 2. 2016 apparent electrical conductivity (EC) (A) shallow and (B) deep maps. (C) Soil survey map. Ch = Commerce silty clay loam, Cn = Commerce very fine sandy loam, Sb = Sharkey clay, and Ta = Tunica clay.
Figure 3. 2021 apparent electrical conductivity (EC) (A) shallow and (B) deep maps. (C) Soil survey map. Ch = Commerce silty clay loam, Cn = Commerce very fine sandy loam, Sb = Sharkey clay, and Ta = Tunica clay.
Table 3. Descriptive statistics of clusters derived from the apparent electrical conductivity shallow (ECas) and deep (ECad) readings.
an = number of samples, Min = minimum, Max = maximum, LM = low-medium, M = medium, and MH = medium-high.
indicating good quality data [9]; the ratio between the ECa shallow and deep values were less than one signifying a regular soil profile, meaning for this field, soil properties that may be correlated with ECa increases with depth [9].
The relationship between ECa data and soil physical and chemical properties can be complex [8] [9]. Apparent electrical conductivity measurements have provided a general estimate of soil texture and have shown promise for mapping soil spatial variability [9]. On non-saline soils, increases in clay content were associated with increases in ECa shallow and deep readings [19]. The soil in this study’s field was not classified as being saline [22]. Hence, it was assumed that the northern portions of the field contained more clay than the southern sections of field. Points in the northern section of the field were grouped into the High ECa zone. Furthermore, similarities and differences were apparent in the spatial patterns of the ECa shallow and deep measurements.
Furrow irrigation was used to supply water to the crops grown in this field when needed. Yearly, the plot was irrigated from south to north. Some of the variability seen on the field maps could be attributed to irrigation. Furthermore, irrigation and other farm management practices would have affected the topsoil more than the subsoil, thus contributing to the topsoil having more variability than the subsoil (ECa deep readings).
The results supported the theory that soil maps derived from the ECa data should show the same pattern over time [19]. They are additional to the times reported in other studies reporting comparisons ranging from 2 months to 4 years [17] [18] [19] [20]. Others have reported positive results of using georeferenced ECa data to be reliable for developing soil sampling strategies for non-point source pollutants [29], soil quality [30], and variations in crop yields [31]. Sampling schemes can be easily developed from the maps produced in this study.
The soil spatial variability map derived from the EC data is totally different from the field’s USDA soil survey map. The ECa map shows smoother transitions compared with the hard breaks between the soil survey map units. It is essential to point out that those differences are not new for maps derived with ECa data. Nevertheless, the ECa and USDA soil survey map provides important information for agricultural production.
5. Conclusion
The findings of this case study indicated that ECa measurements have a longevity of at least five years, supporting backward compatibility between ECa data and other types of data collected in the past. For example, a producer may have yield monitor data collected in 2020. However, apparent electrical conductivity data were not collected by the producer, consultant, etc., of the field until 2021. Thus, if the patterns observed on the 2021 ECa maps are good for five years or more, then the producer, consultant, etc., should be able to evaluate the relationships between the different datasets. Future research will continue to focus on that point and others to determine the agronomic significance of ECa maps derived from mobile sensors in agricultural fields in Mississippi, USA, over various time periods.
Acknowledgements
The author thanks Milton Gaston, Jr., for his assistance in collecting the apparent electrical conductivity data. This research was supported in part by the US Department of Agriculture, Agricultural Research Service. The findings and conclusions in this publication are those of the author and should not be construed to represent any official USDA or US Government determination or policy.