MODIS land data storage, gridding, and compositing methodology: Level 2 grid

TLDR

The approach offers scientific and data‑processing advantages over conventional remote‑sensing storage methods and can be applied to any dataset with subpixel‑accurate geolocation. The authors describe a generic MODIS land gridding and compositing algorithm that leverages a novel data‑storage structure to exploit multiple surface observations more fully than conventional methods, illustrating it with simulated data and discussing practical storage considerations. The methodology preserves sensor‑grid intersection information, enabling more accurate algorithm development, meeting diverse MODIS land product generation needs, and allowing users to create application‑specific datasets while facilitating efficient processing and reprocessing.

Abstract

The methodology used to store a number of the Moderate Resolution Imaging Spectroradiometer (MODIS) land products is described. The approach has several scientific and data processing advantages over conventional approaches used to store remotely sensed data sets and may be applied to any remote-sensing data set in which the observations are geolocated to subpixel accuracy. The methodology will enable new algorithms to be more accurately developed because important information about the intersection between the sensor observations and the output grid cells are preserved. The methodology will satisfy the very different needs of the MODIS land product generation algorithms, allow sophisticated users to develop their own application-specific MODIS land data sets, and enable efficient processing and reprocessing of MODIS land products. A generic MODIS land gridding and compositing algorithm that takes advantage of the data storage structure and enables the exploitation of multiple observations of the surface more fully than conventional approaches is described. The algorithms are illustrated with simulated MODIS data, and the practical considerations of increased data storage are discussed.