Multi-Sensor & Temporal Data Fusion for Cloud-Free Vegetation Index Composites

Abstract

Remotely sensed data from satellite sensors such as Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR) provide almost daily global coverage. Satellite sensor data are used to create scientific data products that include surface reflectance, land surface temperature, sea-surface temperature and many others, as well as ancillary metadata like satellite viewing angle and data quality information. Vegetation indices, like Normalized Difference Vegetation Index (NDVI) (Jensen, 2000), derived from reflectance products of satellite sensors, are generally used as indicators of relative abundance and activity of green vegetation, often including leaf-area index, percentage green cover, chlorophyll content, green biomass, and absorbed photosynthetically active radiation. Frequently reflectance data products needed to create vegetation indices include undesired cloud, water vapour, aerosols, or other poor quality pixels. Continuous monitoring of occurrences such as droughts, frosts, floods, major fires, forest stress, or natural disasters are just a few of the circumstances when daily cloud-free vegetation index composites data are of high utility. The traditional approach to creating a single synthetic cloud-free image that includes ideal values selected from a temporal set of possibly cloudy satellite images collected over a continuous time period of interest is called multi-temporal compositing (MTC). MTC compositing is generally used to create vegetation indices images from data products with high temporal and low spatial resolution such as those produced by the National Oceanic and Atmospheric Administration’s (NOAA) AVHRR sensor or NASA’s MODIS (Justice, 1998). Various methods of MTC have been utilized to produce scientific data products including Maximum Value Compositing (MVC), Constrained View Maximum Value Compositing (CV-MVC) (Cihlar et. al., 1994, Heute et. al., 1999), and CVMVC which incorporates sensor data quality information. The motivation for investigating multi-sensor and temporal fusion for creating hightemporal frequency composites is to overcome the limitations of single-sensor MTC methods and deliver continuous monitoring capabilities that exceed the temporal frequency of currently available 8-day, 10-day, 14-day, and 16-day composite vegetation index data products. Currently available composite products do not provide sufficient frequency and temporal detail to capture and quantify important events, do not deliver data for continuous environmental monitoring, and provide temporally sparse inputs precluding effective agricultural productivity modelling. O pe n A cc es s D at ab as e w w w .in te ch w eb .o rg