With the development of quantitative remote sensing

the scaling problems attract more and more attention. The discrepancy between observation scale

model scale and land surface process scale may lead to different conclusions. Now

how to effectively scale remotely sensed information at different scales already becomes one of the most important research focuses of remote sensing. The aim of our research is to compare and analyze two general scaling methods

the Taylor Series Expansion Model （TSM） and the Computational Geometry Model （CGM）

and apply them to the scaling of leaf area index （LAI）. Firstly

the necessity and importance of scaling are analyzed. Secondly

based on the research of description for the same object using different scale data

the mechanism of scaling effects is presented. Then

the two general models

TSM and CGM

are briefly introduced and their advantages and disadvantages are discussed in detail. Finally

through the retrieval of leaf area index

the two models are comprehensively compared and analyzed in three distinct landscapes. The result shows that the relative scaling error increases with the heterogeneity of land surface. The relative scaling error is 2% in the relatively homogeneous woodland; however

it arises up to 7% in crop-water mixed areas. Apparently

the TSM can better characterize the scale effect and obtain more accurate surface parameters when both small scale （high resolution） data and large scale （low resolution） data are available. The relative scaling error can be reduced to less than 1% for all these test landscapes when TSM is used in scaling. In contrast

CGM can not produce rational result and the relative error is still large. It may be due to using inappropriate weights or data ranges in the model. More study about CGM is needed. On the whole

it is necessary to select the suitable scaling model according to the practical applications. The scaling makes the remote sensing products at different scales comparable and the surface parameter retrieval results more accurate. Scaling technique will provide a powerful technical support for applications in resources survey

environment and disaster monitoring

and other relevant fields.