![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Designing a national soil erosion monitoring network for England and Wales |
VerfasserIn |
Murray Lark, Barry Rawlins, Karen Anderson, Martin Evans, Luke Farrow, Miriam Glendell, Mike James, Jane Rickson, Timothy Quine, John Quinton, Richard Brazier |
Konferenz |
EGU General Assembly 2014
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 16 (2014) |
Datensatznummer |
250088907
|
Publikation (Nr.) |
EGU/EGU2014-3083.pdf |
|
|
|
Zusammenfassung |
Although soil erosion is recognised as a significant threat to sustainable land use and may be
a priority for action in any forthcoming EU Soil Framework Directive, those responsible for
setting national policy with respect to erosion are constrained by a lack of robust,
representative, data at large spatial scales. This reflects the process-orientated nature of
much soil erosion research. Recognising this limitation, The UK Department for
Environment, Food and Rural Affairs (Defra) established a project to pilot a cost-effective
framework for monitoring of soil erosion in England and Wales (E&W). The pilot
will compare different soil erosion monitoring methods at a site scale and provide
statistical information for the final design of the full national monitoring network that
will:
provide unbiased estimates of the spatial mean of soil erosion rate across E&W (tonnes
ha-1 yr-1) for each of three land-use classes -
arable and horticultural
grassland
upland and semi-natural habitats
quantify the uncertainty of these estimates with confidence intervals.
Probability (design-based) sampling provides most efficient unbiased estimates of spatial
means. In this study, a 16 hectare area (a square of 400 x 400 m) positioned at the centre of a
1-km grid cell, selected at random from mapped land use across E&W, provided the sampling
support for measurement of erosion rates, with at least 94% of the support area corresponding
to the target land use classes.
Very small or zero erosion rates likely to be encountered at many sites reduce
the sampling efficiency and make it difficult to compare different methods of soil
erosion monitoring. Therefore, to increase the proportion of samples with larger
erosion rates without biasing our estimates, we increased the inclusion probability
density in areas where the erosion rate is likely to be large by using stratified random
sampling. First, each sampling domain (land use class in E&W) was divided into
strata; e.g. two sub-domains within which, respectively, small or no erosion rates,
and moderate or larger erosion rates are expected. Each stratum was then sampled
independently and at random. The sample density need not be equal in the two strata,
but is known and is accounted for in the estimation of the mean and its standard
error.
To divide the domains into strata we used information on slope angle, previous
interpretation of erosion susceptibility of the soil associations that correspond to the soil map
of E&W at 1:250 000 (Soil Survey of England and Wales, 1983), and visual interpretation of
evidence of erosion from aerial photography. While each domain could be stratified on the
basis of the first two criteria, air photo interpretation across the whole country was not
feasible. For this reason we used a two-phase random sampling for stratification
(TPRS) design (de Gruijter et al., 2006). First, we formed an initial random sample of
1-km grid cells from the target domain. Second, each cell was then allocated to a
stratum on the basis of the three criteria. A subset of the selected cells from each
stratum were then selected for field survey at random, with a specified sampling
density for each stratum so as to increase the proportion of cells where moderate
or larger erosion rates were expected. Once measurements of erosion have been
made, an estimate of the spatial mean of the erosion rate over the target domain, its
standard error and associated uncertainty can be calculated by an expression which
accounts for the estimated proportions of the two strata within the initial random
sample.
de Gruijter, J.J., Brus, D.J., Biekens, M.F.P. & Knotters, M. 2006. Sampling for Natural
Resource Monitoring. Springer, Berlin.
Soil Survey of England and Wales. 1983 National Soil Map NATMAP Vector 1:250,000.
National Soil Research Institute, Cranfield University. |
|
|
|
|
|