Spatio-temporal Cokriging Method for Assimilating and Downscaling Multi-scale Remote Sensing Data

Authors: Bo Yang*, Department of Geography, University of Cincinnati, Cincinnati, OH 45221, USA; , Hongxing Liu, Department of Geography, University of Cincinnati, Cincinnati, OH 45221, USA; , Emily Kang, Department of Mathematical Sciences, University of Cincinnati, Cincinnati, OH 45221, USA;
Topics: Spatial Analysis & Modeling, Remote Sensing, Quantitative Methods
Keywords: spatio-temporal modeling; data fusion; Cokriging; remote sensing
Session Type: Paper
Day: 4/12/2018
Start / End Time: 10:00 AM / 11:40 AM
Room: Studio 10, Marriott, 2nd Floor
Presentation File: No File Uploaded


Given that there is typically a tradeoff between a remote sensing data’s acquisition frequency and spatial resolution, no single sensing instrument is able to provide measurements at both high spatial and high temporal resolution. This paper presents a spatio-temporal Cokriging (ST-cokriging) method for fusing remote sensing data sets with different temporal sampling frequencies and different spatial locations. Our method is based on building statistical models to explicitly describe the spatio-temporal dependence within and between different data sets, and we generalize traditional cokriging in purely spatial setting to spatio-temporal context. The spatio-temporal covariance and cross-covariance structures are modeled and estimated from data, enabling us to borrow strength not only over space and time but also between different data sets. Compared with other state-of-the-art heuristic data fusion methods, this ST-cokriging method is a better solution for dealing with the spectral difference between multi-sensor source data. Furthermore, ST-cokriging can effectively fill in data gaps in space due to clouds, instrument malfunction or other reasons, filter out data noise, and generate reliable predictions associated with uncertainty measurements. In addition, we also demonstrate that this ST-cokriging method can provide more accurate and reliable results as well as associate uncertainty. This ST-cokriging method is implemented using Python language and has been incorporated into a software package within ArcGIS environment. In our study, we apply ST-cokriging to fuse daily MODIS NDVI and NIR images and Landsat TM/ETM+ NIR and NDVI images, at 250m and 30m spatial resolution, respectively, over the Sierra Nevada, Lake Tahoe Basin, US.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login