Analysis of spatial variability of physico-chemical properties of soil and their potential use in Precision Agriculture Bogusław Usowicz, Mieczysław Hajnos, Zofia Sokołowska, Grzegorz Józefaciuk Grzegorz Bowanko, Jan Kossowski Institute of Agrophysics, Polish Academy Sciences ul. Doświadczalna 4, 20-290 Lublin, http://www.ipan.lublin.pl/~usowicz/ Spatial variability of plant yield observed under natural conditions is a result of the spatial variability of soil and meteorological conditions. Spatial variability of soil can result from a natural variability of soil properties as well as from human activity. Both above components of variability are in many cases difficult to distinguish and so these are analyzed simultaneously. Soil texture and mineral composition as well as their spatial distributions are not significantly dependent on time. However some soil components having important effect on plant growth and development, characterized by marked variability not only in time but in space, may be regulated in field conditions. By improving of plant growth conditions one tends to optimize the yield from every field site involving minimum costs and causing minimum hazard for the environment. Recognizing the spatial and time variability of particular soil features together with an experience coming from a large number of soil experiments on the effect of these components on yield may be helpful for decisions on application of particular management practices on defined soil sites. Determination of soil spatial variability needs an adequate number of samples and application of spatial analysis methods. The number of sampling sites may be roughly determined from statistical analysis; however the optimal sample number and their localities needs additional geostatistic analysis that additionally allows to minimize costs of investigations. The aim of this work was to determine spatial variability of selected physicochemical soil properties important for plant growth, on agricultural fields of different areas thus collecting data usable for precision agriculture. EXPERIMENTAL OBJECT AND METHODS OF ANALYSIS The present paper shows results measured in surface layers (1-10 cm) of two cultivated fields located in Trzebieszów Commune (Lublin voivodship). The first field (A) had dimensions of 30 × 600 m, and the second one (B) of 50 × 200 m. The data discussed include: granulometric composition, pH, amount of organic matter and cation exchange capacity (CEC) all measured using standard soil science methods and soil moisture measured with TDR just after harvest. Experimental points were located in these fields in regular networks determined using a belt-meter and the coordinates of the selected reper points (with a precision of 1 to 5 m) were estimated using the Trimble's GPS GeoExplorer 3. Except of soil properties the spatial distribution of yield was analyzed: mixture of oat, wheat and barley (on field A), and rye (on field B). The yield distribution was determined basing on yield coming from 1 m2 squares that covered the given field in as much as possible uniform network. For field A 72 squares were selected and for field B 30 squares. The plants were harvested manually, threshed and the grains dried and weighted. Separately for both objects basic statistical parameters of the studied soil properties were calculated, including the average value, standard deviation, coefficient of variability (CV), maximum and minimum values, as well as estimates of the distribution of a given property as skewness and kurtosis. Spatial characteristics of the experimental data were performed using geostatistical methods. Spatial variability of each soil property on a given field was determined using semivariograms. Value of nugget, ranges and thresholds of spatial autocorrelations were determined and the semivarigram models were fitted to the empirical data along with estimation of the model fitting parameters. The statistics, semivariograms, estimates of the studied soil properties and mapping was performed using computer programs GeoEas, Statistica 6, Variowin 2.21, GS+5 Demo and Surfer 8 Demo. RESULTS Te average yield of oat-wheat-barley mixture from 1 m2 square on the field A (188 g) was over 100 g lower than the average yield of rye in field B (302 g), similarly as the range of the yield being over 100 g lower for the mixture. The yield of rye was between 117 to 556 g·m–2 and of the mixture from 55 to 325 g·m–2. The coefficient of variability for the yield was similar for both fields (23-28%). The distribution of the yield of the mixture was smoother than that for the rye. The content of particular granulometric fractions was markedly different in the field-scale. In the surface layer (0-10 cm) of the field A the sand fraction accounted for 48.8%, and on the field B 54.7%. The silt fraction was 34.5% (field B) and 37.2% (field A) and the clay fraction was 10.8% (field B) and 14% (field A). The organic matter content was small for both fields, around 0.8%. The minimum and the maximum content of organic matter was 0.014 and 1.8%, respectively. The soil was acidic with the average pH (H2O) value of 4.46 for the field A and 4.81 for the field B. Minimum pH for all samples was 3.93 and the maximum 6.59. The average CEC value was around 11 cmol·kg–1, however high differences were noted for particular samples, from 3.69 to 23.8 cmol·kg–1. The average value of soil moisture differ less than 1% for both fields (0.14-0.15 m3 m-3). Also the variability coefficients were similar, around 35% for both fields. The investigated soil properties exhibited space variability, mostly of a spherical character. The exponential character was observed only for three variables (sand content in field A and silt and organic matter content in field B). Parameters of semivariograms showed that nugget effect is present which indicates less variability of the investigated properties in comparison to the assumed minimum distance of sampling. Values of sill are similar to the values of variance estimated using standard methods. From analysis of the above values one can suspect that within the studied fields no evident trends of changes in the studied soil properties occur. Values of sill were governed by the content of particular granulometric fractions with the highest values noted for sand, smaller for silt, and the smallest for clay and these were markedly higher for field A than B. Values of sill for other soil properties were also higher for field A. Spatial dependencies of physicochemical soil properties seemed to be related to the magnitude of the studied field. For field A the range was from 78 to 500 m and for field B - 12 to 310 m. The range of spatial dependence for yield was similar for both fields and equal to around 27 m, however for the field B of lower area the semivariance was about 1.8 times higher than in field A with mixed crop. The spatial dependence of soil moisture for both studied fields was recognized. For field A the exponential dependence occurred whereas spherical one for field B. The ranges of spatial dependencies were similar, around 100 m for both fields. Parameters and models of semivariograms estimated for particular soil properties and yield, as well as the data measured in particular points were applied – using kriging method – for mapping of the spatial distributions for given object (Figs 1 and 2) and for calculating the estimation error. The estimation errors for all the studied data did not exceeded 10% of the analyzed property. In the vicinity of the experimental points these errors were markedly lower (around 1-2%) and the highest values occurred on the boundaries of sampling networks. The yield distribution was more or less related to all investigated soil physical and chemical properties. Thus knowing the spatial distributions of physicochemical soil properties and their influence on plants, one may predict the approximate amount of yield in a given location on particular field. Simultaneously, the estimated maps of spatial distribution of soil physicochemical properties provide essential information which can be used for making a proper decision on agricultural management practices on the whole field or on its part. Marked differentiation of spatial distributions of all investigated parameters observed on both fields despite of their small areas certifies that having larger amount of sampling points is advantageous. Visualization of soil physical and chemical properties and their estimation errors allows for distinguishing areas on a given field wherein an increase (or a decrease) of the amount of the sampling points is required in further measurements to optimize the error values and adjust them to the experimenter needs. Plon Yield [g] a) Distance [m] Odległość Odległość Distance [m] b) pH [H2O] Distance [m] Odległość Odległość Distance [m] c) CEC [cmol kg-1] Distance [m] Odległość Odległość Distance [m] d) Wilgotność Distance (m) Water content (m3 m-3) Odległość Odległość - Distance (m) Fig. 1. Spatial distribution of grain yield (a), pH (b), CEC (c) and water content (d) in the cultivated field A. a) Plon Yield [g] Distance [m] Odległość Odległość Distance [m] b) pH [H2O] Distance [m] Odległość Odległość Distance [m] c) CEC [cmol kg-1] Distance [m] Odległość Odległość Distance [m] Wilgotność d) Water content (m3 m-3) Distance (m) Odległość Odległość - Distance (m) Fig. 2. Spatial distribution of grain yield (a), pH (b), CEC (c) and water content (d) in the cultivated field B. SUMMARY The studies performed may have not only theoretical but also practical importance. The general state of the basic soil chemical and physical properties in the variously used fields was recognized, the parameters describing their spatial variability were determined and the spatial dependences of soil features was mapped. The knowledge on spatial variability of soil properties and plant yield in agricultural fields allowed for estimation of the realistic conditions of plant development that may constitute a base for selection of the areas requesting changes in agricultural measures as liming or fertilization, as well as may serve for elaboration of detailed agrotechnical instructions for precise soil management. The studies described herein constitute a basis for precision agriculture that actually under increasing interest in developed countries. Particular attention should be placed on these soil features which are rather stable in time but rather difficult in estimation. However dynamic soil features, changing with external conditions are also useful and these should be collected in databases, as well.
Pages to are hidden for
"Analysis of spatial variability of physico-chemical properties of soil"Please download to view full document