A Comparison of Bangladesh Climate Surfaces from the Geostatistical Point of View

Avit Kumar Bhowmik

doi:10.5402/2012/353408

Abstract

This study analyses the degree and margin of differences among the surfaces of annual total precipitation in wet days (PRCPTOT) and the yearly maximum value of the daily maximum temperature (TXx) of Bangladesh, produced by thin plate spline (TPS), inverse distance weighting (IDW), ordinary kriging (OK), and universal kriging (UK) methods of spatial interpolation. From the surface differences, the maximum and minimum differences are observed between the surfaces produced by TPS and IDW, and OK and UK, respectively. The residual plots from cross-validation depict that IDW and OK methods mostly under predict and TPS and UK methods mostly overpredict the observed climate indices’ values. Both the tendency of methods’ over and underprediction and the surface-differences decrease with the increase in the number of available spatial point observations. Finally, two performance measures—the index of agreement (d) and the coefficient of variation of prediction (ρf)—imply that there is a little difference in the prediction ability of the four different methods. The performance of the spatial interpolation improves with the increase in the number of available spatial points, and eventually the predicted climate surfaces get similar. However, UK shows better interpolation performance in most of the years.

Highlights

The intrinsic responsibility of the professionals, who deal with climate, is to provide insights regarding climate variability at any place at any time
The maximum and minimum differences are observed between the surfaces produced by thin plate spline (TPS) and inverse distance weighting (IDW), and ordinary kriging (OK) and universal kriging (UK), respectively
This paper analyzes the degree and margin of differences among the surfaces of the PRCPTOT and the TXx of Bangladesh produced by the TPS, IDW, OK, and UK methods of spatial interpolation

Summary

Introduction

The intrinsic responsibility of the professionals, who deal with climate, is to provide insights regarding climate variability at any place at any time. Spatial interpolation methods as geostatistical prediction methods, with their inherent properties and applications, have successfully been implemented to combine different georeferenced climate variables and parameters in such a way that it is possible to give consistently derived estimates at any place for a certain time [3,4,5,6,7,8]. In light of the discussion above, this study compares climate surfaces that have been produced by two deterministic and two stochastic spatial interpolation methods, by evaluating their surface differences, residuals, and predictability It creates the high-resolution continuous surfaces of two climate indices—PRCPTOT and TXx. The PRCPTOT characterizes the annual total precipitation in wet days, and the TXx corresponds to the yearly maximum value of the daily maximum temperature. Is applicable for this study in spite of the irregular spatial point configuration

Methodological Framework

Results and Discussions

Conclusion