Abstract: Error resources of shelterbelt porosity estimated by measuring shelterbelt photo with digitized photographic silhouettes were analyzed, of which, the random error is studied. Results show that the bigger random error variance of crown and trunk porosities is the upper limit of whole variance, that the random errors of crown, trunk, and whole porosities in each typical shelterbelts studied distribute in accordance with normal distribution, that the random error variance of whole porosity is irrelevant to tree species and patterns of the shelterbelts, and that the random error variances between crown and trunk porosities of the shelterbelts, which are composed of Populus×beijingensis or P.×xiaozhuanica or other hybrid popular arranged by rectangle or triangle, have no significant difference, however, for local popular, the random error variance of trunk porosity is signifieantly bigger than that of crown. The random error limits of generalactual porosity estimated from average value of measuring photo with digitized photographic silhouettes were defined, and the idea of applying the limits for calculating the number of observation and delimiting range to be measured were also discussed. Finally, available way for establishing the perfect method such as "estimation of shelterbelt porosity by measuring photo with the help of digitized photographic silhcuettes" is advanced. The available way is that both random error and error from both projective aberration and contractive aberration in the porosity respectively, is restricted by increasing the number of samples from different parts of the same belt, and is emended by calculating with their model.

Key words: Porosity, Random error, Farmland shelterbelt, Digitized photographic silhouettes