Minimize Land Monitoring

Crop monitoring by multitemporal fusion of MERIS FR and Landsat TM data, test site Albacete, Spain

15 October 2013

Web Content Image

Remote sensing applications benefit from the use of data fusion approaches based on spectral unmixing, which have proven their value for delivering spectrally consistent fused images while reducing the mixed pixel problem. In this study the scientists propose a multi-temporal setting of the algorithm that is applied to full MERIS and Landsat TM time series. The final goal is to complete or fill gaps in the Landsat time series by using MERIS data with more frequent coverage. Hence obtaining consistent time series at high spatial resolution and with the enhanced spectral properties of MERIS.

The potential of multitemporal land cover mapping and monitoring using MERIS data has been assessed by numerous studies at global and national scales. However, better resolution is frequently required to properly assess land cover changes in heterogeneous landscapes or for crop monitoring. The enhanced time series obtained using Landsat-like (20-30 m) and MERIS-like (250-500m) images allow developing operational applications that require monitoring rapidly-varying phenomena at high spatial resolution, such as precision agriculture, irrigation advisory services, and near real-time change detection.

The performance of the method was illustrated in two time series acquired over Albacete, Spain in 2004 and 2009. The temporal NDVI profiles showed that rapidly varying phenomena were captured in a more consistent way if the downscaled (fused) products are included in the time series. The results show the validity of the processing scheme, both in terms of fusion quality assessment and consistency of the derived NDVI multitemporal products.

Read more