The Moderate-Resolution Imaging Spectroradiometer (MODIS) is a key instrument on the Earth Observing System (EOS) spacecraft. The instrument views the entire surface of the Earth every 1-2 days with 36 spectral bands, and has spatial resolutions at nadir of 250, 500 and 1000 m. The data in this study were provided by the Space Science and Engineering Center, University of Wisconsin-Madison (SSEC/UW-Madison), after being transformed into McIDAS format. This is equivalent to the MODIS level 1 product distributed by the Distributed Active Archive Center (DAAC) located at NASA’s Goddard Space Flight Center (GSFC). The level 1 product is processed from level 0 data that has been time-referenced and annotated with ancillary information, including radiometric and geometric calibration coefficients, and geolocation information.

The purpose of this presentation is to evaluate the instrument character using spatial structure and correlation functions. The structure value can be expressed at each of the fixed two points r1 and r2 as defined by Gandin (1963). However a modification is required for the application of the technique to satellite observations, since measurements at fixed locations are not available. Discrete structure values can be computed for all possible pairs of satellite measurements as the modification. This objective analysis requires the assumptions of isotropy and homogeneity to make the structure value dependent only on the scalar separation distance. Therefore, we can analyze the structure values over the observation field with no cloud-contaminated pixels. In other words, the structure function analysis requires clear areas over satellite images. Gandin showed that the noise-included structure value is increased by twice the mean-squared error. Therefore, the error can be estimated at zero distance separation by means of the extrapolation of structure function. This study object is noise level of the MODIS emissive infrared spectral bands (3-15 ́m) using structure function analysis. I will present comparisons of the estimated noise levels from selected observations. Similarly, correlation function analysis will be included in the presentation as a further discussion.