Research project S0/00/001 (Research action S0)
Topic and context
The project lies within the scope of the "Local vegetation" research field.
In previous years, remote-sensing tracking of agricultural production has attracted a great deal of attention and been the subject of mounting interest within public and private agri-business circles. At a time when crop type discrimination has become a common application, crop growth modelling is mostly confined to single sensor approaches. Alongside the optical data that have long been used within this sphere, radar retrodiffusion models have demonstrated the possibility of inverting radar signals onto assorted biophysical parameters such as biomass, leaf area index (LAI), plant heigth and cover rate.
Satisfactory results have been obtained for cereals, maize and sugar beet. Significant progress is also expected in the use of polarimetric SAR (Synthetic Aperture Radar) data and very-high-resolution temporal data (Wide Swath ASAR). Furthermore, remote sensing of soil water content offers a further opportunity to improve crop yield models by gauging more accurately the temporal and spatial variability of water distribution in the root area.
Given the fact that crop development is very much dependent on both water being available in the layers near the soil surface and that information about soil humidity will allow improvements to remote sensing of crops, a joint effort by teams specialising in hydrology and agronomy is likely to significantly improve the accuracy of current methods. Ultimately, the main sticking point with the operational use of growth models is the temporal resolution of the high-spatial-resolution remote detection data. The synergy between the discrimination capability of optical captors and the high SAR observation frequency enables a new approach to be developed involving various spatial and temporal scales both for hydrological aspects as well as for crop monitoring.
Objectives
The overall research objective is to develop a regional crop growth model that will be frequently updated using remote detection data and will include a spatially distributed hydrological model. The scientific challenge is two-fold: firstly, the need to map out the assimilation of different types of data acquired on different scales using a variety of sources within a regional vegetation growth model; secondly, the need to develop methods for jointly estimating hydrological and agronomical parameters on the basis of optical sensors and Synthetic Aperture Radar (SAR). Hydrological and growth models will be developed and validated using two experimental sites. The sturdiness of the proposed algorithms will be assessed systematically through cross-validation as the two sites are characterised by very different pedological and topographical conditions.
The action’s six specific objectives may be defined in greater detail as follows:
The first objective involves extraction soil roughness and its temporal evolution in order to significantly improve the mining of biophysical variables using SAR data.
The second objective is to estimate soil humidity in plots and regions using multiple-polarisation radar data.
This soil humidity will be estimated for bare soil and for soil with plant cover.
The third objective involves automated plot definition and characterisation of the variability within plots using hyperspectral imaging.
The fourth objective is to determine crop parameters for either a plot or region through the combination of SAR and optical data.
The fifth objective involves the implementation of various techniques for data assimilation within hydrological and plant-growth models. These techniques need to make allowance for the varying quality of information derived from remote-detection data. The sixth objective is to develop a regional crop growth model explicitly incorporating feedback mechanisms between water quantity in the root area and crop growth.
Finally, specific attention will also be paid to the assimilation of low-spatial-resolution information, supplied every 3 to 5 days by WS ASAR, or the low-to-medium resolution data supplied by optical sensors such as VEGETATION, MERIS and MODIS.
