Login
Toggle navigation
Home
ICPA
Conference
Abstract Management
Abstract Topic Groups
Author Instructions
Registration
Registration Information
16th ICPA - Conference Registration
Registrants Map
Hotel and Travel Information
Tour
Workshops
Exhibit Hall Map
Sponsors
Conference Program
General Outline
Oral Program
Poster Program
Student Poster Awards
Keynote
Plenary Session
Awards
Photos
Conference Survey
Proceedings
Leadership
ISPA Leadership
Officers
Past Presidents
Officer Responsibilities
Country Representatives
Communities
Community Guidance
On-Farm Experimentation
Nitrogen
Latin America
Economics
African Association for Precision Agriculture
Membership
ISPA Member Benefits
Membership Form
Events
ISPA Events
ACPA
ACPA Proceedings
AfCPA
AfCPA Proceedings
CLAP
CLAP Proceedings
ECPA
ECPA Proceedings
ICPA
ISPA Webinars
OFE
AAPA
Latin American
Robotics and Automation Symposium
Event Overview
Registration
Program
Venue
Speakers
About ISPA
Newsletters
History
Jobs
Precision Ag Definition
Agriculture Course Database Submission
Publications
ICPA Proceedings
ECPA Proceedings
Contact Us
Members
Suggestion Form
Conference
Abstract Management
Abstract Topic Groups
Author Instructions
Registration
Registration Information
16th ICPA - Conference Registration
Registrants Map
Hotel and Travel Information
Tour
Workshops
Exhibit Hall Map
Sponsors
Conference Program
General Outline
Oral Program
Poster Program
Student Poster Awards
Keynote
Plenary Session
Awards
Photos
Conference Survey
Proceedings
Proceedings
Search
Authors
Topics
Years
Types
Find matching any:
Reset
» Add more types
Add filter to result:
Probability Distributions And Alternative Transformations Of Soil Test NO3-N And PO4-P, Implications For Precision Agriculture
A. Moulin
Agriculture and Agri-Food Canada
Recommendations for fertilizer N in crop production and precision agriculture depend on statistical analyses of data which represent soil NO3-N and PO4-P fertility typical of management zones and fields. Non-normal distributions of soil test N are commonly log transformed prior to statistical analysis for interpolation with methods such as kriging, regression, or principle component analysis. These data are transformed to ensure that analysis meet the assumptions of normality for the distribution of data and equality of variances. Analyses of soil test NO3-N and PO4-P in the 0-15 cm depth increment for samples (48 to 100 sites per field) in 8 fields in Manitoba identified a range of distributions including the lognormal, exponential, Johnson lognormal and Gamma. Mixed distributions consisting of 2 normal functions combined were also observed for some landforms. No single distribution characterized soil NO3-N for all landforms or fields, and the distribution functions considerably by field and by landform. The consequences of this wide range of distributions are significant, as the calculation of moments such as the mean or standard deviation based on the normal distribution for untransformed data will result in biased estimates of parameters for soil properties. The normal quantile transformation is an alternative to those based on other functions. The transformation assigns a normal quantile value to each of ranked data, resulting in a normal distribution. However the normal quantile transformation should be applied with caution to parametric analysis such as analysis of variance or principle components analysis, as the results of analyses cannot be back-transformed to means or standard deviations. Continuous functions such as the Johnson Sl, Generalized Log or normal quantile may be useful as alternatives to the lognormal in transforming data, in order to take advantage of the wide range of options for parametric analyses in analysis of variance, regression and principle component analysis. Furthermore these transformations will improve recommendations for N fertilizer based on soil test as the central moment is representative of the soil property for the spatial scale in question. However each transformation should be evaluated on a case-by-case basis with respect to the statistical analysis, soil property, spatial scale, and landform in question.
A. Moulin
Spatial Variability in Crop, Soil and Natural Resources
Poster
2014
Download paper