Remote sensing is an important source of land cover data required by many GIS users. Land cover data are typically derived from remotely–sensed data through the application of a conventional statistical classification. Such classification techniques are not, however, always appropriate, particularly as they may make untenable assumptions about the data and their output is hard, comprising only the code of the most likely class of membership. Whilst some deviation from the assumptions may be tolerated and a fuzzy output may be derived, making more information on class membership properties available, alternative classification procedures are sometimes required. Artificial neural networks are an attractive alternative to the statistical classifiers and here one is used to derive a fuzzy classification output from a remotely–sensed data set that may be post–processed with ancillary data available in a GIS to increase the accuracy with which land cover may be mapped. With the aid ancillary informatio...

Land cover classification by an artificial neural network with ancillary information

For the d etailed definition of future earth observation missions for a griculture and forestry, it is required to take into account the reflection and emission characteristics of the targets of interest. For active microwave instruments, the radar signatures of agricultural crops and forests are required. In view of the lack of sufficiently comprehensive historical data sets, ESA decided, partly in cooperation with the J oint Research Centre of the European Comnunity (JRC), to carry out a series of airborne radar campaigns to measure the radar echo from agricultural crops, bare soils and forests as a function o f radar frequency, polarisation, and incidence a ngle during the growing season on several tests sites in Europe. During radar coverage, cwprehensive ground data collection was carried o ut. In this paper, an overview will be given of the complementary deployment of airborne scattermeters and imaging radars d uring AGRISCATT, as well as the ground data collection. Samples of the first results will be shown.

Radar Signature Measurements During The Agriscatt Campaigns

Separable processes represent a convenient class of models for data collected on a regular rectangular lattice. Three model-based tests, for testing separability and testing axial symmetry and separability together, are presented. These are shown to be much more powerful than existing model-free tests using the sample periodogram, provided the model assumptions hold. A simulation study also suggests that these tests are not very sensitive to small departures from the assumed process.

/pdf/model-based-tests-for-simplification-of-lattice-processes-59okyo2hv5.pdf

Model-based tests for simplification of lattice processes

AgriSAR 86 data from the Feltwell Site, U.K., was affected by a variety of radiometric distortions. These distortions prevent accurate calibration of the images, but analysis of the image statistics, after some radiometric corrections have been performed, permits a number of qualitative inferences about variations in crop backscatter in time and space. Cereals and sugar beet appear to exhibit different incidence angle responses. The separability of wheat and sugar beet varies with time and incidence angle. The separation of spring barley and wheat varies with time. Cereals show greater spatial variability than sugar beet. Individual fields exhibit great apparent variability in their temporal responses; for sugar beet this can in the main be explained by the effects of speckle, but there is some evidence of real variation for winter wheat.

Inferences on spatial and temporal variability of the backscatter from growing crops using AgriSAR data

The AgriSAR '86 data set for the Feltwell, U.K. test site is assessed for its ability to provide information on the spatial and temporal variation of backscatter from growing crops. The image data is shown to contain several sources of radiometric distortion which affect any attempt at image calibration. Of these, the simplest to correct is that caused by range interpolation in the amplitude data. All interpolated pixels should be discarded from the amplitude data supplied by CNES. Correction from slant range to ground range is comparatively straightforward. Effects due to variations in antenna pattern, in the range and azimuth direction, require assumptions about the angular variation of the backscatter of crops and statistical homogeneity of ground cover before they can be removed. The properties of system noise appear complicated and are not yet properly characterized, but interfere with corrections for the antenna pattern. Offsets in the two channels of the complex data also interfere adverse...

The radiometric quality of AgriSAR data

http://ieeexplore.ieee.org/iel2/4289/12379/00570547.pdf

C. Yanasse

Papers

The radiometric quality of AgriSAR data

Inferences on spatial and temporal variability of the backscatter from growing crops using AgriSAR data

Change Detection In Agrisar Images