 |
| Titel |
Parameterising root system growth models using 2D neutron radiography images |
| VerfasserIn |
Andrea Schnepf, Bernd Felderer, Peter Vontobel, Daniel Leitner |
| Konferenz |
EGU General Assembly 2013
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 15 (2013) |
| Datensatznummer |
250071774
|
|
|
|
|
|
| Zusammenfassung |
| Root architecture is a key factor for plant acquisition of water and nutrients from soil. In particular in view of a second green revolution where the below ground parts of agricultural crops are important, it is essential to characterise and quantify root architecture and its effect on plant resource acquisition.
Mathematical models can help to understand the processes occurring in the soil-plant system, they can be used to quantify the effect of root and rhizosphere traits on resource acquisition and the response to environmental conditions. In order to do so, root architectural models are coupled with a model of water and solute transport in soil.
However, dynamic root architectural models are difficult to parameterise. Novel imaging techniques such as x-ray computed tomography, neutron radiography and magnetic resonance imaging enable the in situ visualisation of plant root systems. Therefore, these images facilitate the parameterisation of dynamic root architecture models. These imaging techniques are capable of producing 3D or 2D images. Moreover, 2D images are also available in the form of hand drawings or from images of standard cameras. While full 3D imaging tools are still limited in resolutions, 2D techniques are a more accurate and less expensive option for observing roots in their environment. However, analysis of 2D images has additional difficulties compared to the 3D case, because of overlapping roots.
We present a novel algorithm for the parameterisation of root system growth models based on 2D images of root system. The algorithm analyses dynamic image data. These are a series of 2D images of the root system at different points in time. Image data has already been adjusted for missing links and artefacts and segmentation was performed by applying a matched filter response.
From this time series of binary 2D images, we parameterise the dynamic root architecture model in the following way:
First, a morphological skeleton is derived from the binary images by a closing and a thinning step. Then, a weighted graph is produced from this skeleton, where root tips and branch points are the nodes of the graph. For each connecting edge, the pixel coordinates are stored in a list. Finally, a root system growth model is used to determine individual roots within the graph. In this way, the sequential appearance of each sub-branch is maintained.
We demonstrate the use of this algorithm to determine parameters for the root system growth model of Leitner et al. (2010). We use 2D radiography images of Lupine plants. Parameters that are gained from the images include the length of the apical and basal zones, the internodal distances, the number of branches per root, the branching angels, the root radii, and the root growth rate. Computed parameter values are means of four replicates, i.e. the means over four root systems grown under the same conditions. The root systems were classified according to their branching order, and average parameter values were determined for each root order. Based on these parameters, the dynamics of root system growth can be recaptured and analysed. |
|
|
|
|
|
|