A new resampling method of MODIS NDVI images based on integration with Landsat 7 ETM+ data

Time series of MODIS imagery provide almost real time, continuous, and free data, but their relatively coarse spatial resolution limit their exploitation for certain applications. A comparative study of NDVI (Normalized Difference Vegetation Index) images derived from TERRA MODIS and Landsat 7 ETM+ data was carried out during past researches aimed at understanding NDVI MODIS images potentialities for the specific application of hydrologic basin characterization. In this context the possibility of following the evolution of NDVI during the year increase the effectiveness of this type of data. According to this instance a new resampling method for MODIS NDVI imagery (supposed belonging to a time series) is proposed in this paper. This is aimed at improving the MODIS spatial resolution making use of geometric reference information obtained from a classification map of the main land covers derived from Landsat ETM+ data. The basic idea is that the geometric clustering of the territory does not change even though its radiometry does; it is therefore possible to drive the resampling process of the MODIS NDVI time series to a better geometric resolution using a single ETM+ classification map. If the monitored territory suffers from fast changes in its geometry (field shapes, wooded area shapes, etc.) it will necessary to update the reference classification map more often. If successful, this approach provides a powerful monitoring system for change-detection applications: a new type of data characterized by high temporal frequency and better geometric resolution. In this work great attention was paid to the problem of the geometric co-registration of the two types of data. A specific and innovative co-registration image-to-image procedure was developed based on a planar rational polynomial transformation. Some comparisons and tests will be presented to demonstrate the added value of this approach to the problem of the exploitation of high frequency satellite acquisitions (MODIS). NDVI MODIS images resampled both through the proposed technique and through a traditional Nearest Neighbour approach will be compared; correspondent Landsat NDVI images will be used as reference data to test obtained results.

Ҳали таржима қилинмаган