| dc.description.abstract | Analysis and interpretation of spatial variability of land/soil properties is a key-stone in
precision agriculture. Spatial variability is a problem in soil testing because mixing soil
cores from high- and low-fertility areas creates a soil sample that does not represent
either area because each point in a field is unique. Precision agriculture needs to be
employed in solving this problem, which involves the use of Global Positioning System
(GPS) and Geographic Information Systems (GIS). However, reliable soil information
required for precision farming planning is scarce, particularly in Kenya.
Therefore a more detailed soil study of a part of Kenya Agricultural Research Institute
(KARl) Katumani Research Centre was conducted with the principal objectives of (a)
characterizing spatial variability of soil properties (b) determining relationships between
spatial variability of soil fertility and their determining factors and (c) evaluating Gridpoint
versus Grid-cell soil sampling schemes for precision farming. Spatial variability
and soil management zone maps were presented each at scale 1: 6,000 and the results of
soil properties and their determining factors relationship were presented in bar charts.
GPS was used to geo-reference the sampled locations, which were based on the gridpoint
and grid-cell soil sampling schemes. Physical and chemical properties of the soil
samples were analyzed using the standard procedures. Spatial analyst extension of Arcview
GIS software was used for the GIS analysis. Genstat was used for statistical
analysis, where one-way analysis of variance to test the significance of the studied
factors. Spatial analysis by calculating variogram statistics was used to test spatial
variability of soil properties. Generally soil fertility data showed variability where, soil
phosphorus and percentage clay content were highly variable with 888.0292 and
115.3625 general vanances respectively. Total nitrogen was the least variable with
general variance of 0.0022. Present land use, vegetation cover and soil texture were the
major factors influencing soil phosphorus, total nitrogen and soil pH distributions in the
study area respectively. This is because they were all significant at P<0.05. Micro-relief
had no effect on the soil fertility within the study area by depicting unusual relationship
with the exception of phosphorus, which did not show significance at P< 0.05. Organic
matter was influenced by vegetation cover and not by slope due to its unusual
relationship, although was significant at P< 0.05. Phosphorus was high in cultivated area
than in grazing area. This was probably due to application of phosphatic fertilizers. Soil
reaction (pH) was found to be low in loamy sand soils than in the sandy clay, due to high
rate of leaching of basic nutrients and low organic matter hence low buffering capacity in
loamy sand soils. Generally, the study area had relatively acid soils due to the presence of
gneiss parent material. Vegetation cover played a major role in the distribution of soil
nitrogen in the area hence high levels were found in areas with thick herbs and shrubs. By
observation, grid-point sampling scheme was a better strategy in characterizing spatial
variability of soil properties in the area than the Grid-cell sampling scheme. This is
because variability was precisely defined within very short distances than in Grid-cell
sampling scheme, which generalized variability in each cell. This was also supported
statistically where grid-point sampling scheme gave higher general variances for all
land/soil properties than Grid-cell sampling scheme. Soil management decisions would
therefore be based on the developed soil management zones for precision agriculture
where inputs should be matched with the site potential to maximize crop yields while
minimizing excess usage of crop inputs. | en |
