Scale is a key concept for understanding the complexity of earth system.And it is regarded as one of the main challenges of earth observation.It is crucial to select the optimal spatial resolution of remote sensing image according to its application field and its characteristics.Based on analyzing the scale characteristic of remote sensing images

this paper analyses the scale selection and discusses geo-statistics based method of quantificationally selecting the optimal spatial resolution of remote sensing image.Especially

this paper analyses traditional local variance method and its defects.As for local variance method

it is suggested to measure the relationship between the size of the objects in the scene and spatial resolution

and then calculate the mean value of the standard deviation by passing a n pixel by n pixel moving window for each pixel on successively spatially degraded images

and then takes the mean of all local variances of the successively spatially degraded images as an indication of the spatial elements within the scene of the image

according to which the optimal spatial resolution whose mean local variance is maximum can be estimated.So the traditional local variance method computes the mean of all local variance on the different ground area

which results in that the local variance does not fall substantially with the successively degradation of spatial resolution of the image

consequently the computational results are non-comparable

and the traditional method can not achieve satisfactory result.Breaking through the limitation

this paper proposes the modified local variance method based on variable window sizes and variable resolution to quantitatively select the optimal spatial resolution of remote sensing images

which are high spatial resolution images with large window size and low spatial resolution images with small window size

so that the relevant ground area is kept consistent.The actual process inevitably involves the ideal decimal window size

which is proposed to be computed based on the spatial statistics theory.Consequently

the optimal spatial resolution of remote sensing image can be computed by comparing the modified mean local variance.This paper takes three pieces of IKONOS images which stand for building district

farmland and forest individually as primary experimental image and the modified local variances are computed for the three kinds of landscape individually.The experimental results show that this geo-statistics based method of quantificationally selecting the optimal spatial resolution of remote sensing image has theoretical and instructional meaning: the spatial resolution of 3—5m

3—5m and 1—5m is respectively suitable for landscape of building district

farmland and forest;only the local variance based on variable window size and variable resolution can indicate the actual change of local variance with the degradation of spatial resolution of the image;local variance method adopts proper window size to reflect the change of landscape property

so it can reflect the micro-characters and is suitable for study on the fine scale landscape or the artificial landscape.