GROUND TRUTH PLANNING FOR SYNTHETIC APERTURE RADAR (SAR): ADDRESSING VARIOUS CHALLENGES USING STATISTICAL APPROACH
While planning target parameter retrieval experiment in farmers’ fields using Synthetic Aperture Radar (SAR) remote sensing, it is very important to answer three questions. The first question is, ‘How large should a farmer’s field be from where the target parameter sample should be collected?’. The answer to this question is very critical particularly when the experiment is to be carried out in South Asian countries where in general, agricultural land holding by an individual farmer is small. The second question of importance is that “what should be the sample size for model development?” or in other words “how many such large enough farmers’ fields needs to be taken so that enough target parameter samples will be available to be able to develop a meaningful relationship between remotely sensed parameter and the observed target parameter?”. The third important question is what should be the minimum sample size for meaningful model validation? In this paper answers to all these questions are obtained by adopting statistical approach when the experiment is planned for target parameter retrieval using synthetic Aperture radar (SAR). The required size of the sampling unit is determined by taking into account the characteristic-fading phenomenon of SAR signal and probabilistic statistical approach based upon the margin of error permissible by the experimenter in estimating average. The answer to the second question about how many farmers fields or in other words, sampling units be chosen during ground truth data collection such that a nearly true relationship can be developed between observed target parameter on ground and SAR backscatter, is obtained based upon the margin of error permissible by the experimenter in a regression estimate. Finally answer to third question has been reached by developing a procedure using ‘precision power approach’ to arrive at the minimum size of validation sample required by the experimenter for a meaningful validation exercise. It should be noted that the approach presented in this paper to determine sample size for model development and model validation are general in nature and are equally applicable to other remote sensing data sets as well.
Synthetic Aperture Radar (SAR), fading phenomenon, probabilistic statistical approach, precision power approach, sampling unit size, sample size, model development, model validation, ground truth, target parameter retrieval, Radarsat-1/2 SAR, Envisat-1 AS
- There are currently no refbacks.