Click for RRes home page
About us
Resources
Research
Into practice
For the public
Careers
This image is chosen randomly from a collection

Bookmark and Share

This image is chosen randomly from a collection

Rothamsted Research is an institute of the Biotechnology and Biological Sciences Research Council


Bookmark and Share

Environmetrics

We are interested in the spatial variation of environmental processes over a very wide range of spatial scales. Many factors cause variation over all these scales, and this creates challenges whether we wish

Our research focuses on mathematical and statistical ways of modelling spatial variation that are applicable to these problems.

New: A recent paper in Nature (8th September 2005), in which we were collaborators with the National Soil Resources Institute, has estimated that the soils of the United Kingdom have lost some 13 million tonnes of carbon a year from 1978 to 2003, approximately the amount by which Britain has reduced its carbon emissions from the base level that was set in 1990.
Sampling to monitor changes in organic carbon content of the soil

Sampling to monitor changes in organic carbon content of the soil

Experiments on spatial variation of the nitrogen response of cereals.

Experiments on spatial variation of the nitrogen response of cereals.

Staff

Collaborations

As well as collaborations within Rothamsted we have active collaborations with staff at the

We are active within the Pedometrics Commission of the International Union of Soil Sciences

Core Problems.

Geostatistical methods have been used by environmental scientists over the last 25 to 30 years for estimating spatially correlated variables at unsampled locations or over larger regions (blocks). Geostatistics uses a model of the spatial structure of variability, called the variogram, as a basis for optimizing such estimates. We are particularly interested in problems associated with the estimation of the variogram in difficult cases such as where the variable of interest has a strong deterministic trend or where it consists of both continuous variation and the discrete effects of point contamination (Lark, 2003).

Several of our externally funded projects have developed out of this core interest. For example, past research on optimizing spatial sampling for estimation of the variogram is currently being developed into a method for adaptive spatial sampling . We are also interested in how mechanistic models that predict environmental variables can best be deployed along with observations for spatial prediction and temporal monitoring.

The assumptions of geostatistics fail when the variability of a property is not uniform in space but changes, perhaps from one part of the landscape to another. Here we have played a leading role in the application of the wavelet transform to environmental problems. We have developed inferential methods for detecting change in variation and covariation of properties (Lark and Webster, 2001) and gained insight into the underlying sources of variation, and the implications for how to sample variables or how to predict them with mathematical models (Milne et al., 2005).

Applications

Our methods have been applied in a range of projects, funded by BBSRC, Defra and the Home-Grown Cereals Authority. We have studied factors that control nitrous oxide emission from arable soils using wavelet techniques, and we are also applying these to spatio-temporal analysis of beach profiles to illuminate problems in coastal engineering. We have studied the spatial variation of crop requirement for nitrogen at within-field scales using geostatistically based methods for analysis of data from novel experimental methods. Our interests in robust geostatistics have equipped us to contribute to work for Defra on the design of monitoring schemes for the assessment of changes in soil quality over time.

Our research includes:

Recent publications