Pansharpening Landsat 8 Data For Improved Agricultural Field Monitoring

Wednesday, 17 December 2014
Hankui Zhang and David P Roy, South Dakota State University, Brookings, SD, United States
Satellite data provide a synoptic view and have been used for agricultural applications including cropland distribution mapping, crop condition monitoring, crop production assessment, and yield prediction. The ability of satellite data to monitor agriculture reliably is dependent on many factors but is fundamentally constrained by the satellite spatial resolution relative to the field spatial dimensions. The recently launched Landsat 8 satellite has improved calibration, radiometric resolution, geometry and global data acquisition frequency over previous Landsat sensors. Pansharpening is an established technique to integrate higher spatial resolution panchromatic information with lower spatial resolution multi-spectral information. A new pansharpening algorithm is presented that is specific to Landsat 8 and that models the sensor spectral response functions to provide a universal algorithm that is computationally efficient and applicable to large volume data. Experiments conducted using Landsat 8 data acquired over agricultural regions with markedly different field dimensions in South Dakota, China, and India, are presented to demonstrate and quantify the utility of the 15m pansharpened Landsat 8 data over conventional 30m data.