Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2009, Issue 6, Vol. 2, pp. 459-465
Estimation of accuracy of satellite definition of cropland in framework
of method with using of different resolution data
Национальный центр космических исследований и технологий
Национальное космическое агентство, Республика Казахстан, Ул. Шевченко 15, 050010, Казахстан
The level of errors in the technology of a satellite estimation of the spring cropland acreage in Northern
Kazakhstan was rated. The spring wheat and barley are a regional monoculture and cultivated on large fields
(400 hectares). The technology of the satellite data analysis is based on use of the different spatial resolution
images: IRS LISS III (23 m) and EOS MODIS (250 m); with allocation of two classes: annual spring cultures
and other classes. The separate fields form an agricultural mask. The single covering by the high resolution
images is used for an estimation of the fields sizes. For agriculture masks in the size from above 15 thousands of
hectares the errors level is rated less than 3 % with probability of occurrence is equal 0,95. The errors of
recognition of spring annual cultures at the analysis of 4 years MODIS monitoring data are rated as about 1 %.
Thus, the satellite technology of an estimation of the spring cropland acreage in Northern Kazakhstan, used in
interests of the Ministry of Agriculture of Kazakhstan, has a level of summary errors less than 4 %.
Keywords: Satellite images, space resolution, agriculture masks, errors, annual cultures recognition, pixel count, field area
Full text