FEDERALNOE GOSUDARSTVENNOE BJUDZHETNOE UCHREZHDENIE NAUKI INSTITUT PROBLEM EHKOLOGII I EHVOLJUTSIIIM. A.N. SEVERTSOVA ROSSIJSKOJ AKADEMII NAUK (IPEHEH RAN)
发明人:
PUZACHENKO JURIJ GEORGIEVICH,Пузаченко Юрий Георгиевич,PUZACHENKO MIKHAIL JUREVICH,Пузаченко Михаил Юрьевич,SANDLERSKIJ ROBERT BORISOVICH,Сандлерский Роберт Борисович,KRENKE ALEKSANDR NIKOLAEVICH,Крен
申请号：
RU2012113106
公开号：
RU2565280C2
申请日:
2012.04.04
申请国别(地区):
RU
年份:
2015
代理人:
摘要：
FIELD: agriculture.SUBSTANCE: invention relates to forestry, in particular to remote methods of forest resource inventory over wide areas. The method is characterised in that a remote sensing of the inventory territory is carried out with a resolution of 2-40 m on three to ten spectral channels in the range of 0.45-90 microns. A digital terrain model is created. The sensing of the inventory territory is carried out in different seasons. The digital terrain model is divided into hierarchical levels. The data obtained is formed into a database with a line structure and georeferencing, in each line of which a set of spectral characteristics is placed for each channel for each period and a set of hierarchical levels of terrain with their characteristics, which is classified according to the iterative procedure of K-average at K=2. The spectral characteristics in the obtained classes are compared with predetermined stable spectral images of the land cover types. The main correspondence of the obtained classes is isolated, on specific types of land cover. The classes are isolated, having forestry value. In each class having forestry value, the number of control points is determined on the relation Li=wlog2mi, where mi - is the frequency of class i in the whole sample, w - is the coefficient of accuracy depending on the prevalence of the type of land cover, and Li - is the number of control points in the class i. The control points are placed inside the isolated classes. On the control points the forest inventory field description is carried out. The forest inventory characteristics resulting from descriptions are added to the database. For each forest inventory characteristic its states are interpolated to all forest inventory territory. The accuracy of the interpolation is determined. The homogeneous areas are isolated relative to interpolated states of forest inventory characteristics. Based on the given threshold of similarity the homogeneous areas less than the taken
