摘要
An exploratory spatial data analysis method (ESDA) was designed Apr.28,2002 for kriging monthly rainfall. Samples were monthly rainfall observed at 61 weather stations in eastern China over the period 1961-1998. Comparison of five semivariogram models (Spherical, Exponential, Linear, Gaussian and Rational Quadratic) indicated that kriging fulfills the objective of finding better ways to estimate interpolation weights and can provide error information for monthly rainfall interpolation. ESDA yielded the three most common forms of experimental semivariogram for monthly rainfall in the area. All five models were appropriate for monthly rainfall interpolation but under different circumstances. Spherical, Exponential and Linear models perform as smoothing interpolator of the data, whereas Gaussian and Rational Quadratic models serve as an exact interpolator. Spherical, Exponential and Linear models tend to underestimate the values. On the contrary, Gaussian and Rational Quadratic models tend to overestimate the values. Since the suitable model for a specific month usually is not unique and each model does not show any bias toward one or more specific months, an ESDA is recommended for a better interpolation result.
An exploratory spatial data analysis method(ESDA) was designed Apr.28,2002 for kriging monthly rainfall.Samples were monthly rainfall observed at 61 weather stations in eastern China over the period 1961-1998.Comparison of five semivariogram models(Spherical,Exponential,Linear,Gaussian and Rational Quadratic)indicated that kriging fulfills the objective of finding better ways to estimate interpolation weights and can provide error information for monthly rainfall interpolation.ESDA yielded the three most common forms of experimental semivariogram for monthly rainfall in the erea.All five models were appropriate for monthly rainflaa interpolation but under different circumstances.Spherical,Exponential and Linear models perform as smoothing interpolator of the data,whereas Gaussian and Rational Quadratic models serve as an exact interpolator.Spherical,Exponential and Linear models tend to underestimate the values,On the contrayr,Gaussian and Rational Quadratic models tend to overestimate the values.On the contrary,Gaussian and Rational Quadratic models tend to overestimate the values,Since the suitable model for a specific month usually is not unique and each model does not show any bias toward one or more specific months,an ESDA is recommended for a better interpolation result.
