{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Remote Sensing - a tool for environmental observation

Hence if multi temporal remote sensing techniques are

Info icon This preview shows pages 93–96. Sign up to view the full content.

View Full Document Right Arrow Icon
grow from late April to late July or early August. Hence, if multi-temporal remote sensing techniques are applied, the dates of image acquisition should be matched with the growing stages of the studied crops. A sequence of image acquired in January, June and August are mostly ‘best’ for multi-temporal crop classification. A disadvantage of multi-temporal remote sensing is that it is expensive: multiple images must be acquired and radiometrically and geometrically corrected. Especially, the geometric correction must be very accurate otherwise the temporal change detected might be caused by ‘non-fitting’ agricultural lots. An example of a multi-temporal land use classification in the Po-valley in Italy is described by Azzali (1985). Figure 6.7 shows the temporal profiles of the crops. Figure 6.7 Temporal profile / phenological cycle of winter wheat (Jensen, 2000).
Image of page 93

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
93 Figure 6.7 Cont’d Phenological Cycles of sugar beets, cotton and alfalfa and images collected over a period of 12 months, original in colour (Jensen, 2000).
Image of page 94
94 Change detection is one of the most powerul applications of Earth observation. Several regions in the world are currently undergoing rapid, wide-ranging changes in land cover or land use. Well-known examples are of course the ongoing deforestation process in the Amazon, central Africa and Indonesia. Other examples are the changing land ice sufrace and length of glaciers as a result of our changing climate. earth observation provides a fast and relatively cheap way to inventory and monitor these changes. A prerequisite for the successful application of remote sensing for change detection is that the changes in land cover result in changes in radiance values and that these radiance changes are large with respect to other causes of radiance changes such as changing atmospheric conditions, differences in soil moisture and differences in sun angle (Mas, 1999; Singh, 1989). Change detection methods can basically be subdivided into the following broad classes: 1. Image differencing: registered images acquired at different times are subtracted to produce a residual image which represents the change between the two dates.Pixels of no radiance change are distributed around the mean while pixels of radiance change are distributed in the tails of the statistical distribution. 2. Vegetation Index Differencing: The NDVI is calculated (the normalized ratio of red and near infrared reflectance) for the images acquired at different dates. Next the NDVI values of the two images are subtracted and result in high positive or high negative values for change areas. 3. Direct multi-date classification: This method is based on the single analysis of a combined dataset of the two dates images aiming at identifying areas of changes. Two images are first merged into one multi-band image set. Classes where changes are occurring are expected to present statistics significantly different from where change did not take place and can so be identified. Unsupervised classsification methods e.g. the isodata algorithm to reveal the spectral distance patterns in the data.
Image of page 95

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Image of page 96
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}