Inge Aalders

Agricultural census data and land use modelling

Modelling land use change is often constrained by imperfect and incomplete data sources. This paper explores three modelling methodologies and their ability to predict agricultural land use on the basis of information from the Scottish Agricultural Census. This dataset, which contains information on ownership, land use and employment statistics for the majority of Scotland, is restricted by law concerning the level of detail which can be provided, and as such is both the best available source of information for agricultural practice in Scotland and is partial and incomplete. It is demonstrated that the methodologies applied to the problem (neural network, Bayesian network and decision tree), with a limited number of relevant drivers included in the modelling process, are capable of use for the prediction of changes in land use, suitable for policy analysis. The reasons for selecting these particular modelling approaches included a need to deal with a large amount of noisy, inaccurate data, and the fact that each is capable of successfully investigating and quantifying unknown relationships between dataset variables. The greatest success, measured as a combination of accuracy, data-handling flexibility and ease of model comprehension by the user, was achieved by the decision tree method.

The Risk of Peat Erosion from Climate Change: Land Management Combinations—An Assessment with Bayesian Belief Networks

ABSTRACT Modeling risk factors to soils is constrained by the lack of key data and understanding that explicitly and quantitatively link specific threats to risk. Peat erosion results from the complex interaction of climatic, topographic, and anthropogenic influences acting over a long period of time. With numerous contemporary factors operating to perpetuate the erosion processes, it is often difficult to identify with certainty what actually are the initial and subsequent drivers of erosion. In this situation, expert opinion forms a vital source of information. Here we demonstrate how Bayesian Belief Networks (BBN) can be used to combine quantitative data from the National Soils Inventory of Scotland (NSIS) with qualitative expert knowledge to estimate risk of peat erosion in Scotland. This model was used to identify the main factors associated with peat erosion. It was shown that climatic variables (increased temperature, decreased precipitation) are the most important risk factors for perpetuating peatland erosion. However, the BBN approach also indicated that maintaining good vegetation cover is a significant mitigating factor. It would follow that land management practices that impact negatively on vegetation cover would also exacerbate peatland erosion given a hot dry climate.

The effect of image compression on synthetic PROBA-V images

We have carried out an in-depth investigation into the effects of image compression on synthetic Probe for On-Board Autonomy – Vegetation (PROBA-V) scenes and Landsat-derived image tiles. The two image compression implementations used were the TER implementation and a bespoke implementation of the Consultative Committee for Space Data Systems (CCSDS) Blue Book standard, which are functionally identical but operate on different image architectures. This work included (1) the development of an approach for producing synthetic scenes that were appropriate in terms of structure and content, and (2) evaluation of the image compression approach on the two kinds of image in terms of their usefulness for land-cover mapping. The synthetic image (SI) generation approach has been rigorously tested and produces images that are statistically similar to real scenes, both compressed and uncompressed. The results of our work show that the effects of image compression vary significantly between bands and with different compression ratios, and that the impact of image compression on image quality does vary with spatial scale. We also found indications of increased error rate at boundaries within the imagery. While the SI generation process and the processing chain of this imagery are not completely consistent with PROBA-V imagery, agreement was found among many of the results produced by the two approaches.