®

Accuracy Assessment

Since 1972, the Landsat satellite sensor systems (MSS & TM/ETM+) have made remotely sensed data readily available and have offered an efficient means of collecting information about the environment.  Recent advances in remote sensing technologies and the increasing availability of high spatial and spectral resolution earth observation satellite data provide great potential for acquiring detailed spatial information to identify and monitor environmental problems within specific areas at desirable spatio-temporal scales.

In environments disturbed by anthropogenic processes, transitions in building materials, density, size and shape, vegetation, and intensive socio-economic activities often transform the landscape towards heterogeneity.  Many factors such as reference data, spatial and spectral resolution, radiometry, rectification, variations in vegetation phenology and physiology, and urban development cycles, which can influence the accuracy of an image classification derived from remotely sensed data, must be investigated.

As more advanced digital satellite remote sensing techniques become available, digital image classification becomes more complex.  Despite these advances, computer-assisted image classification is still unable to produce land cover maps and statistics with high accuracy therefore, it has become common practice to assess the reliability of the results.  Because image classification or change-detection analysis maps are often used to assist the development of land planning and management practices, assessment of their accuracy is important.  Errors contained in the classified images can and will adversely affect the accuracy and validity of the resulting products, such as maps and reports.  To adequately assess the accuracy of a remotely sensed classification, we collect accurate ground and reference data.

For more information please contact us at: info@applied-geospatial.com