Sampling in space and
time for survey and monitoring of natural resources
Dates: 2, 3, 4 April 2012
WAGENINGEN UNIVERSITY
Course leader: Dick Brus (Team Soil Geography, Wageningen-UR)
SCOPE
Designing an effective and efficient sampling scheme is an important first step in a research project: it largely determines the quality of the final result. Also, problems with statistical inference can often be avoided when we start thinking before the data are collected. This post-graduate course aims at presenting an overview of sampling methods for spatial inventory (survey) and monitoring of natural resources such as soil, groundwater, surface waters, forests, habitats, landuse/landcover et cetera. Both random sampling strategies and model-based (geostatistical) sampling strategies will be explained, and rules for choosing between alternative strategies will be given. The third day we will go into the problem of sampling in space and time, so that changes over time of, for instance, quality-indicators for soil or surface water, or the spatial extent of habitats can be estimated from the sample data. The course is a mix of theory and practice, with case studies that are analysed using the free R software. Course material: parts of J.J. De Gruijter, D.J. Brus, M. Bierkens - M. Knotters (2006), Sampling for Natural Resource Monitoring (Springer).
PROGRAMME
Day 1:
Sampling in space: probability
sampling for estimating means and totals
|
9.00-9.15 |
Welcome |
|
9.15-9.45 |
Overview of course and some basic concepts |
|
9.45-10.30 |
Simple and Stratified simple random sampling: stratification variables,
proportional and optimal allocation of sampling points, stratification
effect |
|
10.30-11.00 |
Break |
|
11.00-12.15 |
Geographical stratification |
|
12.15-13.15 |
Lunch |
|
13.15-15.00 |
Systematic random sampling: how to estimate the sampling variance?
Two-stage random sampling: how to select primary sampling units? How to
estimate the sampling variance? Relative efficiency |
|
15.00-15.30 |
Break |
|
15.30-17.15 |
Discussion of other sampling
design types brought up by
the participants. Probability sampling in practice: pitfalls |
|
9.00-10.30 |
Increasing precision by exploiting covariates: the regression estimator
Two-phase sampling for
stratification and regression |
|
10.30-11.00 |
Break |
|
11.00-12.15 |
Sampling on regular grids: optimization of grid spacing |
|
12.15-13.15 |
Lunch |
|
13.15-15.00 |
Spatial coverage sampling |
|
15.00-15.30 |
Break |
|
15.30-17.15 |
Model-based sampling:
optimization of sampling pattern for ordinary kriging and universal
kriging |
|
9.00-10.30 |
Design-based versus model-based sampling strategies: fundamental
differences; how to choose? |
|
10.30-11.00 |
Break |
|
11.00-12.15 |
Design-based, model-based and hybrid approaches
Basic types of space-time design: static-synchronous, independent
synchronous, serially alternating, rotational panel, supplemented panel |
|
12.15-13.15 |
Lunch |
|
13.15-15.00 |
Design-based sampling for estimating space-time means, totals |
|
15.00-15.30 |
Break |
|
15.30-1715 |
Design-based sampling for estimating the temporal trend of spatial means
A hybrid, design-based and model-based sampling approach for estimating
the temporal tend of spatial means |
|
COURSE FEE |
|
| PhD's PE-RC/SENSE with TSP |
€ 200.- |
| Other PhD's and WU staff | € 350.- |
| Other participants | € 700.- |
Course fee includes a reader, coffee/tea and lunches
* For PE-RC PhD candidates, the reduced fee only applies if the thesis is defended at Wageningen University.
Please download the registration form and send it to pe-office@wur.nl
Full registration only occurs once you have provided us with the requested information. Accordingly you will receive an official registration confirmation by one of the course organizers.
Further information please contact:
Dr. C. Van de Vijver
Tel: + 31 (0)317-485116
Email: Claudius.vandeVijver@wur.nl