Estimating topsoil organic matter content in two experimental fields in Denmark using multi-spectral aerial imagery and topographic data

Abstract

Geografisk Tidsskrift, Danish Journal of Geography 105(2):1–14, 2005 The potential use of spectral reflectance data and topographical measurements for the estimation of topsoil organic matter (SOM) content is investigated. Intensive measurements of soil texture and topography in two Danish fields combined with digital aerial photographs, representing blue, green, red and nearinfrared wavelength regions, form the basis of the investigation. Analyses of semi- variograms indicate similarity between the spatial variation of the spectral, topographic and textural measurements, and confirms that the sampling density is sufficient to account for the spatial variation of SOM within each field. Multivariate statistics were applied to investigate the potential of spectral reflectance data, including selected spectral indices, and topographical data for prediction of SOM. The results show that the spectral reflectance data obtained within the visible and near infrared range are correlated with the SOM content in both fields, and that the spectral bands are almost co-linear. The relative importance of spectral reflectance and topographical data varied in the two fields, although the combined use of these data improved the estimation of SOM significantly in both fields. The data indicated that SOM content above 3.5% was rare in both fields although a high SOM content was found in one field, which was related to kettle holes characterised by poor drainage. The study focused on the estimation of SOM from zero to 3.5% within the 2 fields. The Predicted accuracy varied between 0.24 and 0.30% SOM, depending on the field and input parameters used in the model. The results also revealed that the data used in the investigation were insufficient to properly describe the variations of SOM within the fields.