Time-Series Data Mining in a Geospatial Decision Support

Time-Series Data Mining in a Geospatial Decision Support
System¤
Dan Li, Sherri Harms, Steve Goddard, William Waltman, Jitender Deogun
Department of Computer Science and Engineering
University of Nebraska-Lincoln, Lincoln NE 68588-0115

Abstract
This paper presents an overview of the motivation for, and the use of time-series data mining in, a Geospatial Decision Support System (GDSS). Our approach is based on a combination of time-series data mining algorithms and spatial interpolation techniques. The initial focus of the system is to facilitate drought risk management. We develop two association rule mining algorithms and two interpolation methods, which help drought experts predict local weather conditions or potential yield impact based on the global weather patterns.

Keywords: Geospatial Decision Support System, Data Mining, Interpolation.

1. Introduction
Drought is a natural process of Great Plains landscapes and results in significant economic, social, and environmental impacts. Historically, more emphasis has been placed on the response component of drought management, with little or no attention to mitigation, preparedness, and prediction or monitoring (Wilhite 2001). Thus, through the National Science Foundation (NSF) Digital Government program, the USDA RMA is working with the University of Nebraska–Lincoln Computer
Science and Engineering (CSE) Department, National Drought Mitigation Center (NDMC), and High Plains Regional Climate Center (HPRCC) to develop a Geospatial Decision Support System (GDSS) to improve the quality and accessibility of temperature and precipitation data for drought assessment and drought risk management. Figure 1 shows two drought assessment maps. The drought map of Nebraska can be generated in real-time produced by our GDSS system for any specified time interval from the project’s home page: http://nadss.unl.edu/.
A common question in risk analysis is “How are events related in time?” In a risk management application where a time-series is a factor, it is important to study the relationships of the parameters that occur together. Data mining algorithms have the potential to identify these relationships.
Predicting events and identifying sequential rules that are inherent in the data help domain experts learn from past data and make informed decisions for the future. For example, decision-makers are interested in discovering associations between the periodical occurrence of El Ni˜no and the periodical occurrence of natural hazards. Data mining techniques can help us build abstract models to represent the reality and to support risk management and mitigation of natural hazards. In the rest
of this paper, we demonstrate the integration of spatio-temporal knowledge discovery techniques in the GDSS using a combination of data mining methods applied to geospatial time-series data.

¤This research was supported in part by NSF Digital Government Grant No. EIA-0091530 and NSF EPSCOR,
Grant No. EPS-0091900.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s