Generation and assessment of MODIS time series using quality information

Abstract

Monitoring and modeling extensive Earth surface processes for regional to global applications such as carbon budgeting or biomass estimation requires time series derived from remotely sensed imagery. Time series are also needed for discrimination of long-term land cover change from short-term variations, mapping of vegetation dynamics and improved land cover mapping and update. The results of these applications, however, clearly depend on the quality of the time series. Cloud coverage, high aerosol content, adverse view and illumination angles, or sensor defects affect and corrupt the data and may lead to false conclusions. Value-added MODIS data contain detailed pixel level quality information. This source of meta-data highly suits for data analysis or generation of time series. A software package, called Time Series Generator (TiSeG), has been developed to analyze data quality and estimate the quality of time series to be generated. TiSeG meets the challenge to weight the data quality against the quantity of available data for meaningful time series construction.