An R package for spatial coverage sampling and random sampling from compact geographical strata by k-means

Walvoort, D.J.J.; Brus, D.J.; Gruijter, J.J. de

doi:10.1016/j.cageo.2010.04.005

Cited by 172 publications

(113 citation statements)

References 11 publications

Supporting

Mentioning

105

Contrasting

Unclassified

Order By: Relevance

“…The third key implication for sampling is that, in general, 25 samples in a landscape unit should be sufficient to characterise mean C s (to within 20% of the true value) for nominal depth intervals of 0-0.1 m and 0-0.3 m. It is arguable that the 'grey area' of 20% is too large for a viable carbon-monitoring scheme but a smaller target, say 10% of the true mean, entails a sampling effort that is likely to be prohibitive. We recommend the use of stratified simple random sampling, with strata of equal area defined by geographic coordinates (Walvoort et al, 2010), and at least two samples collected per strata. We reiterate that the depth intervals we have cited throughout are only nominal: the results relate strictly to soil mass in the profile, which Gifford and Roderick (2003) and McBratney and Minasny (2010) recommended for reporting of C s .…”

Section: Discussionmentioning

confidence: 99%

“…Under ST the area is divided into a pre-specified number of sub-units ('strata'), then all locations within each stratum have equal probability of being allocated to a sample. Using the R statistical software (R Development Core Team, 2011), stratification was done on the basis of spatial coordinates, with a constraint that the strata have equal area (Walvoort et al, 2010). Design SY is a grid whose origin has been chosen at random.…”

Section: Model Formmentioning

confidence: 99%

See 1 more Smart Citation

Soil carbon stock in the tropical rangelands of Australia: Effects of soil type and grazing pressure, and determination of sampling requirement

Pringle¹,

Allen²,

Dalal

et al. 2011

Geoderma

View full text Add to dashboard Cite

On-going, high-profile public debate about climate change has focussed attention on how to monitor the soil organic carbon stock (C s ) of rangelands (savannas). Unfortunately, optimal sampling of the rangelands for baseline C s -the critical first step towards efficient monitoring -has received relatively little attention to date. Moreover, in the rangelands of tropical Australia relatively little is known about how C s is influenced by the practice of cattle grazing. To address these issues we used linear mixed models to: (i) unravel how grazing pressure (over a 12-year period) and soil type have affected C s and the stable carbon isotope ratio of soil organic carbon (δ 13 C) (a measure of the relative contributions of C 3 and C 4 vegetation to C s ); (ii) examine the spatial covariation of C s and δ 13 C; and, (iii) explore the amount of soil sampling required to adequately determine baseline C s . Modelling was done in the context of the material coordinate system for the soil profile, therefore the depths reported, while conventional, are only nominal. Linear mixed models revealed that soil type and grazing pressure interacted to influence C s to a depth of 0.3 m in the profile. At a depth of 0.5 m there was no effect of grazing on C s , but the soil type effect on C s was significant. Soil type influenced δ 13C to a soil depth of 0.5 m but there was no effect of grazing at any depth examined. The linear mixed model also revealed the strong negative correlation of C s with δ 13 C, particularly to a depth of 0.1 m in the soil profile. This suggested that increased C s at the study site was associated with increased input of C from C 3 trees and shrubs relative to the C 4 perennial grasses; as the latter form the bulk of the cattle diet, we contend that C sequestration may be negatively correlated with forage production. Our baseline C s sampling recommendation for cattle-grazing properties of the tropical rangelands of Australia is to: (i) divide the property into units of apparently uniform soil type and grazing management; (ii) use stratified simple random sampling to spread at least 25 soil sampling locations about each unit, with at least two samples collected per stratum. This will be adequate to accurately estimate baseline mean C s to within 20% of the true mean, to a nominal depth of 0.3 m in the profile.Crown

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Model Formmentioning

confidence: 99%

Soil carbon stock in the tropical rangelands of Australia: Effects of soil type and grazing pressure, and determination of sampling requirement

Pringle¹,

Allen²,

Dalal

et al. 2011

Geoderma

View full text Add to dashboard Cite

show abstract

“…It would be straightforward to do the same computations on sample designs obtained for non-uniform regions by spatial coverage methods such as those presented in the "spcova" package for the R platform (Walvoort et al, 2010).…”

Section: Discussionmentioning

confidence: 99%

Planning spatial sampling of the soil from an uncertain reconnaissance variogram

Lark

Hamilton

Kaninga

et al. 2017

SOIL

View full text Add to dashboard Cite

Abstract. An estimated variogram of a soil property can be used to support a rational choice of sampling intensity for geostatistical mapping. However, it is known that estimated variograms are subject to uncertainty. In this paper we address two practical questions. First, how can we make a robust decision on sampling intensity, given the uncertainty in the variogram? Second, what are the costs incurred in terms of oversampling because of uncertainty in the variogram model used to plan sampling? To achieve this we show how samples of the posterior distribution of variogram parameters, from a computational Bayesian analysis, can be used to characterize the effects of variogram parameter uncertainty on sampling decisions. We show how one can select a sample intensity so that a target value of the kriging variance is not exceeded with some specified probability. This will lead to oversampling, relative to the sampling intensity that would be specified if there were no uncertainty in the variogram parameters. One can estimate the magnitude of this oversampling by treating the tolerable grid spacing for the final sample as a random variable, given the target kriging variance and the posterior sample values. We illustrate these concepts with some data on total uranium content in a relatively sparse sample of soil from agricultural land near mine tailings in the Copperbelt Province of Zambia.

show abstract

“…The equal area spatial coverage sampling (EASCS) algorithm was applied in this study for the selection of optimal points. For the purpose of improving the efficiency of sampling, regular grids are first established on the reference image: a grid cell size of about 1/2500 to 1/5000 of the size of the study area is considered suitable [Walvoort et al, 2010]. The mean value of the squared shortest distance (MSSD) is used as an objective function for sampling the GCPs using geographically compact areas as strata, and this function can be minimized by the well-known k-means clustering algorithm.…”

Section: Optimal Ideal Points Positions Samplingmentioning

confidence: 99%

Optimal selection of GCPs from Global Land Survey 2005 for precision geometric correction of Landsat-8 imagery

Peng

et al. 2015

European Journal of Remote Sensing

View full text Add to dashboard Cite

To conduct precision geometric correction of Landsat-8 data, all ground control points from the Global Land Survey (GLS) 2005 are typically selected, thereby making the process time consuming and labor intensive. This paper developed an optimal selection algorithm for choosing representative points. The optimal technique consists of three steps, including 1) evaluating the spatial distribution patterns of points from GLS2005, 2) extracting ideal points positions from each scene based on the spatial distribution patterns obtained in the first step, and 3) selecting real representatives GCPs from the original large number of GCPs based on the positions of ideal points. One hundred individual Landsat-8 images were chosen for precision geometric correction to assess the robustness and efficiency of the method. Experimental result demonstrated that the approach could only consume 1/10 processing time or less when compared to that using the full set of original GCPs while still achieving comparable geometric accuracy. The developed technique will make an important contribution to improving the efficiency of precision geometric product generation systems for Landsat-8 images.

show abstract

An R package for spatial coverage sampling and random sampling from compact geographical strata by k-means

Cited by 172 publications

References 11 publications

Soil carbon stock in the tropical rangelands of Australia: Effects of soil type and grazing pressure, and determination of sampling requirement

Soil carbon stock in the tropical rangelands of Australia: Effects of soil type and grazing pressure, and determination of sampling requirement

Planning spatial sampling of the soil from an uncertain reconnaissance variogram

Optimal selection of GCPs from Global Land Survey 2005 for precision geometric correction of Landsat-8 imagery

Contact Info

Product

Resources

About