Knowledge of spatial variability of soil properties is important in precision agriculture as well as site specific nutrient management. This paper addressed the spatial distribution characteristics of organic matter (OM), pH, available nitrogen (AvN), available phosphorus (AvP), available potassium (AvK) and available sulphur (AvS) in Research farm of SKUAST-K, Shalimar, Srinagar. A total of seventy seven (77) soil samples were collected in a systematic grid design using geographical positioning system (GPS). Each grid was specified at a fixed distance of 50 × 50 m 2. The results showed that soil organic matter and S was distributed normally while as the three soil macronutrients (AvN, AvP and AvK) and soil pH followed log normal distribution. Soil available phosphorus had a highest coefficient of variation (56.87%) and the soil pH (7.06%) the lowest. All the soil macronutrients were found in medium range except sulphur which was found deficient in whole of the research farm. The experimental semivariogram of the log-transformed data of soil available phosphorus, potassium, sulphur, soil pH and normally distributed soil organic matter was fitted to exponential model. Gaussian model was found to be the best fit for experimental semivariogram of soil available nitrogen. Experimental semivariogram results indicated a moderate degree of spatial dependence for soil organic matter, available potassium and sulphur, soil pH and weak degree of spatial dependence for soil available nitrogen and phosphorus. Using such analyses, it is possible to plan appropriate soil management practices, including fertilization for agricultural production and environmental protection.
Soils are inherently heterogeneous in nature, diverse and dynamic system [
The availability of soil nutrients for plant growth and yield production is а function of different parameters, including soil pH, soil organic matter and texture, and soil biological activities [
The important way to gather knowledge in this respect is to prepare maps through spatial interpolation of point based measurements of soil properties using geostatistics. There have been growing interests in the study of spatial variation of soil properties using geostatistics since 1970s, as geostatistics techniques were well developed and successful in characterizing the spatial variations of soil properties [
The current study was undertaken in the Research farm of SKUAST-K, Shalimar for analyzing the spatial variability of soil properties and for identification of nutrient deficiency zones for site specific nutrient management.
The present investigation was carried out in a Research farm of SKUAST-K, Shalimar (34˚8'42" and 34˚9'3"N latitudes and 74˚39'5" and 74˚53'5.6"E longitude) Srinagar (
A total of seventy seven (77) samples were selected in a systematic grid design using Arc GIS. Each grid was specified at a fixed distance of 50 × 50 m2 grid from 0 - 22.5 cm depth. Samples were thoroughly mixed and ground to pass through 2 mm sieve, then stored in plastic bags prior to chemical analysis. Soil pH was determined in 1:2.5 soil: water suspension with digital glass electrode pH
meter [
Statistical parameters which are generally accepted as indicators of the central tendency and spread of the data, were analyzed. These include description of mean, minimum and maximum values, standard deviation and coefficient of variation. To decide whether or not the data followed the normal frequency distribution, the coefficients of skewness and kurtosis were examined [
Spatial analysis was carried out by the use of geostatistical method (Arc GIS) and mapping software (Surfer). Spatial variability in soil fertility parameters were calculated for 0 to 25 cm depth. Firstly variograms were applied to measure the spatial variability of sampled locations, which also provides the parameters that are necessary for interpolation of unsampled areas
Variograms and kriging interpolation were performed in ArcGIS 10.2. The formula applied to the variogram [
y ( h ) = 1 2 m ( h ) ∑ i = 1 m ( h ) ( Z ( X i + h ) − Z ( X i ) ) 2 (1)
where γ(h) is the experimental semivariogram value at a distance interval h, m(h) is number of sample value pairs within the distance interval h, Z(Xi), Z(Xi + h) are sample values at two points separated by the distance h. Several semivariogram functions were evaluated to choose the best fit with the data. Spherical, Exponential or Gaussian models were fitted to the empirical semivariograms. The stationary models, i.e., Gaussian (Equation (2)), Exponential (Equation (3)) and Spherical model (Equation (4)) that fitted to experimental semivariograms were defined in the following equations [
y ( h ) = C 0 + C 1 [ 1 − exp ( − h 2 / a 2 ) ] (2)
y ( h ) = C 0 + C 1 [ 1 − exp ( − h / a ) ] (3)
y ( h ) = C 0 + C 1 [ l .5 ( h / a ) 3 ] for h ≤ a (4)
where C0 is the nugget, C1 is the partial sill, and a is the range of spatial dependence to reach the sill (C0 + C1). The semivariance generally increases with sample separation distance before reaching an asymptote а (the range value). Samples separated by distances greater than range value are considered to be spatially independent where as, within the range, samples show greater similarity when they are nearer to each other [
The sill represented the amount of variation defined by the spatial correlation structure and it is the value of the semi-variogram at which the model first levels out (given as partial sill plus the nugget [
The nugget/sill ratio was used as a criterion for classifying the spatial dependence of soil properties. The variable has strong spatial dependence if the ratio is less than 25%; between 25% and 75%, the variable has moderate spatial dependence; and the variable shows only weak spatial dependence if the ratio is greater than 75% [
As the semivariogram models of the soil data were evaluated, they were used in the development of maps by ordinary kriging interpolation [
The descriptive statistics of the soil fertility parameters are given in
Parameter | Minimum | Maximum | Mean | Median | SD | CV (%) | Skewness | Kurtosis |
---|---|---|---|---|---|---|---|---|
OM | 0.52 | 5.68 | 2.88 | 2.79 | 1.09 | 37.91 | 0.19 | −0.45 |
N | 125.44 | 878.08 | 360.56 | 313.60 | 138.72 | 38.47 | 1.42 | 2.75 |
Log N | 2.10 | 2.94 | 2.53 | 2.50 | 0.16 | 6.23 | 0.03 | 0.90 |
P | 14.92 | 153.28 | 54.67 | 47.45 | 31.09 | 56.87 | 1.00 | 0.57 |
Log P | 1.17 | 2.19 | 1.67 | 1.68 | 0.25 | 14.85 | −0.01 | −0.89 |
K | 44.80 | 487.20 | 186.26 | 173.60 | 74.88 | 40.20 | 1.36 | 2.93 |
Log K | 1.65 | 2.69 | 2.24 | 2.24 | 0.17 | 7.59 | −0.26 | 1.54 |
S | 11.55 | 18.43 | 13.86 | 12.65 | 0.34 | 17.67 | 0.34 | 0.97 |
pH | 5.90 | 7.94 | 6.76 | 6.83 | 0.48 | 7.06 | 1.37 | −1.43 |
Log pH | 0.77 | 0.90 | 0.83 | 0.83 | 0.03 | 3.67 | 0.23 | −0.60 |
SD = standard deviation, CV = coefficient of variation, OM = organic matter (%), N = Available nitrogen (kg∙ha−1), P = Available phosphorus (kg∙ha−1), K = Available potassium (kg∙ha−1), S = Available sulphur (kg∙ha−1).
stable soil parameters [
The normal distribution of data was examined by Quantile-Quantile (QQ) plot. The quantile-quantile plot (QQ plot) is a simple graphical method for comparing two sets of sample quantiles [
Descriptive statistics in this study indicates moderate to high skewness. The value of skewness varies from −0.01 to 1.42 depicting moderate to high skewness (
For geostatistical analyses of soil parameters, an appropriate model was chosen. Best suited models for various parameters are presented in
When the distribution of soil properties is moderately or strongly spatially correlated, the average extent of these patches is given by the range of the semivаriogram [
Variation at microscales smaller than the sampling distances will appear as a part of the nugget effect [
Parameter | Model | C0 | C1 | C0 + C1 | Range | DSD (%) | SD | Estimated error | |||
---|---|---|---|---|---|---|---|---|---|---|---|
MSE | ASE | RMSE | RMSSE | ||||||||
OM | Exponential | 0.72 | 0.658 | 1.378 | 99.6 | 52.25 | Moderate | 0.18 | 0.19 | 1.14 | 0.96 |
Na | Gaussian | 0.13 | 0.02 | 0.15 | 763.2 | 86.67 | Weak | −0.04 | 138.18 | 145 | 1.06 |
Pa | Exponential | 0.28 | 0.06 | 0.34 | 338.35 | 82.35 | Weak | 0.015 | 35.91 | 30.84 | 0.84 |
Ka | Exponential | 0.1 | 0.07 | 0.17 | 99.61 | 58.82 | Moderate | −0.02 | 84.34 | 77.19 | 0.93 |
S | Exponential | 2.19 | 2.97 | 5.16 | 184.03 | 42.44 | Moderate | −0.02 | 2.08 | 2.06 | 0.99 |
pHa | Exponential | 0.002 | 0.004 | 0.006 | 99.62 | 33.33 | Moderate | 0.02 | 0.48 | 0.48 | 1 |
C0 = nugget effect; C1 = partial sill; C0 + C1 = sill; degree of spatial dependence (DSD) = C0 /(C0 + C1) DSD; strong DSD (<25%); moderate DSD (>25 to <75%); weak DSD (>75%). SD: Spatial dependence; MSE: Mean square error; ASE: Average standard error; RMS: Root-mean-square error; MSE: Mean standard error; RMSSE: Root-mean-square standardized error. alog transformed.
of macronutrients was probably because of high soil heterogeneity resulting in large spatial variability of these nutrients.
The spatial dependence can indicate the level of similarity or disturbance of the soil condition [
Based on the results of the present study we may conclude that moderate and weak spatial dependence of soil fertility parameters can be usually attributed to soil and crop management practices [
Cross-validation was used to estimate which of the semivariogram models could give the most accurate predictions of the unknown values of the study area. It was shown that the error terms ME and MSE were close to zero. Subsequently, with implementing these best fit theoretical models and corresponding semivariogram parameters, spatial variability maps of soil properties were created using the ordinary Krigging (
This study demonstrated that the classical statistics of the soil elements indicated a coefficient of variation up to 56.87%, which could not support to identify the sources of variability. This indicates that the classical statistical techniques were utilized to identify an overall variability of soil elements but lacked the necessary techniques to identify the kind of systematic spatial variability at farm scale. However, the geostatistical techniques offer alternative methods over the classical statistics for estimating the parameters spatial dependence and variability in the farm. According to the results, the semivariogram analyses show the presence of a moderate to weak spatial dependence of the selected soil properties within the study area. The Krigged maps of soil parameters can help the researchers to become familiar with the characteristics related to the analyzed soil properties and accordingly can plan appropriate agricultural strategies, including fertilization. Such analyses can save time and expenses, while being statistically of great precision and usability.
Ramzan, S., Wani, M.A. and Bhat, M.A. (2017) Assessment of Spatial Variability of Soil Fertility Parameters Using Geospatial Techniques in Temperate Himalayas. International Journal of Geosciences, 8, 1251-1263. https://doi.org/10.4236/ijg.2017.810072