Spatial-temporal Interpolation of Precipitation for Downscaling Grid-based Climate Data

Authors: Mengchao Xu*, George Mason University
Topics: Temporal GIS, Water Resources and Hydrology, United States
Keywords: downscaling, interpolation, precipitation, MERRA 2
Session Type: Paper
Day: 4/4/2019
Start / End Time: 1:10 PM / 2:50 PM
Room: Buchanan, Marriott, Mezzanine Level
Presentation File: No File Uploaded

Downscaling refers to a prediction of a finer spatial resolution from a lower resolution dataset. In remote sensing, downscaling involves a process of decrease pixel size of input imagery. Main forms of downscaling include dynamical downscaling and statistical downscaling, where dynamical downscaling requires global climate models to support local conditions and statistical downscaling requires local variable involvement. Spatial-temporal interpolation is an indirect downscaling method, which uses “neighboring” points to predict values between two known pixels. The selection of points would take both spatial and temporal effect into consideration. Traditional ways of image interpolation use spatial neighboring as the foundation; however, temporal variability analysis should also be integrated to provide a more comprehensive prediction. From a statistical point of view on temporal variability, it is assumed that due to “randomness”, it will average to a central tendency or to a temporal trend if enough observations are taken. However, the varying temporal effects attract the most interest to the scientists and are the most challenging part to predict or estimate. The objective of this research is to exam major methods of spatial and temporal interpolation on the MERRA 2 dataset for the downscaling purpose and propose or improve methods for local-scale prediction.

Abstract Information

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

To access contact information login