REMOTE-SENSING AND THE MEASUREMENT OF GEOGRAPHICAL ENTITIES IN A FORESTED ENVIRONMENT .1. THE SCALE AND SPATIAL AGGREGATION PROBLEM

The hypothesis tested in this study was that remote sensing constitute s a particular case of an arbitrary uniform spatial sampling grid used to obtain measurements about geographical entities that induces the s cale and aggregation effect responsible for haphazard analysis results . The main objective was to evaluate the impact of measurement scale a nd spatial aggregation on the information content and classification a ccuracies of airborne MEIS-II data acquired over a midlatitude tempera te forested environment. The original MEIS-II data were resampled to f our spatial resolutions, namely 5 m, 10 m, 20 m, and 30 m. Forest clas ses were established according to three progressive levels of spatial aggregation. Descriptive statistics (Wald-Wolfowitz runs test, mean an d variance) were calculated on transects of pixels representing each f orest class delineated on the images at every spatial resolution. A ma ximum-likelihood classification was also performed for each combinatio n of spatial resolution and aggregation level. The results reveal that , except for the mean, changing the measurement scale and the aggregat ion level of the classes greatly affects the values of the descriptive statistics. The Z value of the Wald-Wolfowitz runs test decreases wit h decreasing spatial resolution. The effect is more pronounced when th e classes are progressively aggregated. For most classes, the variance decreases with the decrease of spatial resolution. In such cases, the impact of changing the measurement scale is greater than the change o f aggregation level. Per-class accuracies are also considerably modifi ed depending on the measurement scale and the aggregation level. Withi n a particular aggregation level, some classes are better classified a t fine spatial resolutions, while others require coarser spatial resol utions. Three major conclusions can be stated from these results: 1) T he information content of remote sensing images is dependent on the me asurement scale determined by the spatial resolution of the sensor; 2) neglecting the scale and aggregation level when classifying remote se nsing images can produce haphazard results having little correspondenc e with the objects of the scene; and 3) there is no unique spatial res olution appropriate for the detection and discrimination of all geogra phical entities composing a complex natural scene such as a forested e nvironment. These conclusions provide a theoretical foundation from wh ich original solutions to the problem of appropriate scales of measure ment for geographical entities can be experimented. Logically, there e xists an optimal spatial resolution for each entity of interest, corre sponding to its intrinsic spatial and spectral characteristics.