Skip to Main Content
 

Global Search Box

 
 
 
 

Files

ETD Abstract Container

Abstract Header

A New Approach to Spatio-Temporal Kriging and Its Applications

Abstract Details

2011, Master of Science, Ohio State University, Computer Science and Engineering.

Stochastic spatio-temporal variability is often observed in naturally occurring phenomena. It had always been a challenge to predict their behavior in space and time. Statistical techniques exist that may be united to model and predict the spatio-temporal behavior of these phenomena.

In this research we present a new approach to spatio-temporal data analysis. A new Spatio-Temporal Kriging model was built to predict the spatio-temporal behavior of atmospheric temperature data, gathered from heterogeneous sensors for over 10 years at 63 locations in the US. Kriging interpolates the best linear unbiased estimate of a value at an unobserved point in space, based on the weighted linear combination of surrounding observations, minimizing the prediction error. Spatial and temporal associations in the data were initially modeled separately, using Universal Kriging (UK) and Autoregressive (AR) techniques respectively and then combined to spatio-temporally predict temperatures, k days into the future in a given spatial domain.

ARIMA (Autoregressive Integrated Moving Average) model was used to compare the performance of our Spatio-Temporal Kriging model. Our model performed twice as better with 2.47°C of average standard error (SE) in prediction estimates as compared to 4.49°C from ARIMA. Confidence interval (95% CI) for prediction estimates from ARIMA model was ±8.80°C as compared to ±4.84°C from our Spatio-Temporal Kriging model. Uncertainty in predictions observed from both the models may be largely associated to the presence of strong temporal correlation in the observations at locations near the Great lakes, also observed from slowly decaying autocorrelation function (ACF) at these locations.

A new Space-Time linear model was also built using regression that yielded poor results, because it only captured the effect of latitude on temperature, i.e. temperature drops as we move up north.

We also introduced a novel concept of Kriging based virtual sensor (KVSense) that may be used for temporarily replacing the faulty wireless sensor(s) and also to emulate the working of a real sensor at inaccessible areas.

We concluded by discussing, possible novel energy harvesting (energy conservation and wireless sensor power rejuvenation) strategies for wireless sensor networks (WSN) configured in a spatial domain based on mined spatio-temporal knowledge on availability of ambient (sunlight, wind, etc.)

Srinivasan Parthasarathy, PhD (Advisor)
Gagan Agrawal, PhD (Committee Member)
112 p.

Recommended Citations

Citations

  • Agarwal, A. (2011). A New Approach to Spatio-Temporal Kriging and Its Applications [Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1306871646

    APA Style (7th edition)

  • Agarwal, Abhijat. A New Approach to Spatio-Temporal Kriging and Its Applications. 2011. Ohio State University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1306871646.

    MLA Style (8th edition)

  • Agarwal, Abhijat. "A New Approach to Spatio-Temporal Kriging and Its Applications." Master's thesis, Ohio State University, 2011. http://rave.ohiolink.edu/etdc/view?acc_num=osu1306871646

    Chicago Manual of Style (17th edition)