Engineering topics
Publications and source records attributed to Swanlund, G. D..
The author has identified the following significant results. Automatic photointerpretation techniques were utilized to evaluate the feasibility of data for land use management. It was shown that ERTS-1 MSS data can produce thematic maps of adequate resolution and accuracy to update land use maps. In particular, five typical land use areas were mapped with classification accuracies ranging from 77% to over 90%.
There are no author-identified significant results in this report.
There are no author-identified significant results in this report.
There are no author-identified significant results in this report.
The author has identified the following significant results. The Minnesota Iron Range area was selected as one of the land use areas to be evaluated. Six classes were selected: (1) hardwood; (2) conifer; (3) water (including in mines); (4) mines, tailings and wet areas; (5) open area; and (6) urban. Initial classification results show a correct classification of 70.1 to 95.4% for the six classes. This is extremely good. It can be further improved since there were some incorrect classifications in the ground truth.
There are no author-identified significant results in this report.
There are no author-identified significant results in this report.
There are no author-identified significant results in this report.
The author has identified the following significant results. Primary conclusions from the lake acreage study are: (1) The ERTS-1 band 7 density range of 0-5 reliably indicates open water down to 2 acre size. (2) The density range 6-9 identifies swamps. (3) The depth of the water could not be determined. (4) Cloud shadows can be misread as lakes unless the clouds are detected. (5) ERTS-1 data would provide the information for classifying lakes and for monitoring fluctuations in lake area.
The author has identified the following significant results. Automatic stratification of forested land from ERTS-1 data provides a valuable tool for resource management. The results are useful for wood product yield estimates, recreation and wildlife management, forest inventory, and forest condition monitoring. Automatic procedures based on both multispectral and spatial features are evaluated. With five classes, training and testing on the same samples, classification accuracy of 74 percent was achieved using the MSS multispectral features. When adding texture computed from 8 x 8 arrays, classification accuracy of 90 percent was obtained.
Automatic land use classification in Minnesota
The author has identified the following significant results. Progress is being made in the areas of sun angle and atmospheric effects on the data and its interpretation. The effort is aimed at correction using only navigation and calendar data needed for satellite operation and derivations from the MSS data. Water estimation of value to land planners and conservationists has been demonstrated with verification of performance by comparison with a concurrent study involving map planimetry, aerial photos, and field-checking. The satellite will provide seasonal and annual coverage with timely information in a way not now feasible by conventional methods. For example, the reference data were obtained using the most recent files, which date back to 1949, and from most recent photography taken in 1968. The calendar time involved was three to four months, which is a reflection on priority pressure on trained manpower available. ERTS-1 data can help relieve this problem by providing the needed information while freeing trained manpower for more appropriate parts of the effort.
There are no author-identified significant results in this report.
There are no author-identified significant results in this report.
There are no author-identified significant results in this report.
There are no author-identified significant result in this report.