10. Analysis of soil fertility and mapping using geostatistical information system
Main Article Content
Abstract
Physico- chemical characterization of an agriculturally important soil and its fertility mapping was conducted by collecting 72 soil samples at two depths (0-15 and 15-30 cm) from the Research Farm of Amir Muhammad Khan Campus, The University of Agriculture, Peshawar-Pakistan. These samples were collected at grid pattern with 100 m distances. The results indicated that all samples collected from the area were the total nitrogen content ranged from marginal to adequate level in some samples, ranges from deficient in 9.72 %, marginal in 33.33 % and 56.95 % sufficient nitrogen in the surface while in subsurface it was deficient 54.17 %, marginal in 30.83 % and 15 % sufficient. The AB-DTPA extractable phosphorus was deficient in 97 % surface and 100 % sub-surface soils while potassium was marginal to adequate levels in all samples with mean value of 150 mg kg-1. The surface soil sample was in adequate to the level of 58.33 % and subsurface it range from 86.11 % respectively. After analyzing the data though geostatistical techniques and GIS applications, fertility maps were developed though Kriging that delineated the status of soil properties at every sampled and non-sampled locations that could be used during planning for fertility management. Spatial trend and semivariogram were designed and spatial distribution of soil fertility status was further quantified and visualized. The kriging were used with three semivariogram models (circular, spherical and exponential). Mean Prediction Errors (MPE), Mean Standardized Prediction Errors (MSPE) and Root-Mean-Square Standardized Prediction Errors (RMSSPE) were used to evaluate the models. The results showed that the best model to generate soil fertility map was Kriging with all the three models on the best fitting formula, semivariogram model (MPE and MSPE close to 0, and RMSSPE close to 1).
Keywords: Soil; Fertility; Kriging; GIS; MPE; MSPE; RMSSPE