Using nonlinear geostatistical models for soil salinity and yeild management
Date
2013
Journal Title
Journal ISSN
Volume Title
Abstract
Crop production losses associated with soil salinity on irrigated lands are significant. The genetic complexity of crops with regards to salt tolerance has limited the success of improving salt tolerance through conventional breeding programs. In the meantime, land reclamation and leaching can be expensive and sometimes impractical when fresh water sources are scarce or not readily available. This research introduces a geostatistical approach for the management of crop yield under current soil salinity conditions. It uses three nonlinear geostatistical models - disjunctive kriging (DK), indicator kriging (IK), and probability kriging (PK) - to manage soil salinity and crop yield. The nonlinear models were applied to selected irrigated fields in a study area located in the south eastern part of the Arkansas River Basin in Colorado where soil salinity is a problem in some areas. The overall objectives of this research are: 1) estimate soil salinity in irrigated fields using nonlinear gestatistical models; 2) develop conditional probability (CP) maps that divide each field into zones with different soil salinity levels; 3) estimate the expected yield potential (YP) for several crops at different zones in fields under multiple soil salinity thresholds; 4) evaluate the performance of the nonlinear geostatistical models in developing the interpolated and CP maps provide guidance to farmers and researchers by considering the output of this research as input for precision management of agriculture; and 5) provide guidance to farmers and decision makers in precision management of agriculture. The three nonlinear geostatistical models DK, IK, and PK were used to develop CP maps based on soil salinity thresholds for different crops. These CP maps were compared with actual yield data taken while conducting a soil salinity survey for two fields cultivated with alfalfa and corn. The CP maps divide each field of interest into zones with different probabilities to reach a specific YP for a given crop at a specific soil salinity threshold. Different crops were selected to represent the dominant crops grown in the study area: alfalfa, corn, sorghum, and wheat. Six fields were selected to represent the range of soil salinity levels in the area. Soil salinity data were collected in the fields using an EM-38 and the location of each soil salinity sample point was determined using a GPS unit. Datasets of soil salinity collected in irrigated fields were used to generate the CP maps and to evaluate different scenarios of the expected YP% of several crops at multiple soil salinity thresholds. These datasets were selected to represent a wide range of soil salinity conditions in order to be able to evaluate a wide variety of crops (larger set of crops than those grown in the study area) according to their soil salinity tolerances. Yield data were collected at the same fields to compare the actual data with that estimated by the models. The crops were used for evaluation were selected based on two criteria: dominant in the study area, and represent high, moderate, and low soil salinity tolerances. Different scenarios of crops and salinity levels were evaluated. Semivariograms were constructed for each scenario to represent the different classes of percent yield potential based on soil salinity thresholds of each crop. The results of this research show the nonlinear geostatistical models are efficient in assessing the impact of soil salinity on the spatial variability yield productivity. The comparison of the actual yield data with the estimated yield from the three models shows good agreement where most of the yield samples were located at the appropriate zones estimated with the models. The IK and PK models generated very similar estimates for each of the zones. However, the zones generated by both of these models are slightly different to the zones generated using the DK model. Wheat and sorghum show the highest expected yield potential based on the different soil salinity conditions that were evaluated. Expected net revenue for alfalfa and corn are the highest under the different soil salinity conditions that were evaluated. The CP maps generated using the DK technique give an accurate characterization and quantification of the different zones of the fields. Upon the knowledge of the YP% of different areas, a management decision action can be taken to manage the productivity of a field by selecting another crop or adjusting the inputs such as fertilizer, seeding rates and herbicides in low productivity areas. The information provided by the models about the variability and hotspots can be used for the precision management of agricultural resources. The IK model can be used to generate guidance maps that divide each field into areas of expected percent yield potential based on soil salinity thresholds for different crops. Zones of uncertainty can be quantified by IK and used for risk assessment of the percent yield potential.
Description
Rights Access
Subject
cost-benefit analysis
classified and probability maps
crop-yield management
indicator variograms
non-linear geostatistical models
soil salinity