Gis-Based Soil Sampling Methods for Precision Farming of Rice
Sampling is the first step in the process of precision farming that relies on spatial data. Soil sampling projects are costly and time consuming and selecting a representative sample that can estimate the statistical and spatial properties of soil is a challenge that can cause impasse in the precisi...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
2006
|
| Online Access: | http://psasir.upm.edu.my/id/eprint/612/2/600447_fk_2006_81_abstrak_je__dh_pdf_.pdf http://psasir.upm.edu.my/id/eprint/612/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Sampling is the first step in the process of precision farming that relies on spatial data. Soil sampling projects are costly and time consuming and selecting a representative sample that can estimate the statistical and spatial properties of soil is a challenge that can cause impasse in the precision farming projects and may dissuade the farmers to adopt precision farming. While the random sampling can ensure the unbiasedness of the results, it may not cover the whole study area. Systematic and stratified sampling designs have the potential to reduce the number that is needed for sampling the soil. A trial has been done with sampling on the interpolated map of 2003 data from soil survey of the Sawah Sempadan rice irrigation scheme at North of Selangor Malaysia in 2003 and the result were analyzed both statistically and spatially. For predicting the mean, systematic and stratified scheme produce good results, but stratified sampling could predict the mean with less standard error and narrower confidence interval of mean.
In terms of reproducing the spatial variation and mapping, stratified sampling showed weaknesses with minimum number of samples, while having results comparable to random scheme with three-fold less number of samples. Systematic sampling showed intermediate precision for Nitrogen and Potassium, while higher precision with three-fold samples less than random scheme for Phosphorus. In general, systematic design with 70 samples proved to have good results for the macronutrients mapping.
|
|---|