| Protective forest is an important part of land cover types, and it has very important significance for the earth’s ecological balance and environmental regulation. With the continuous development of our country’s economic construction and it has made remarkable achievements, the ecological damage to the environment caused by urban construction and industrial production is also becoming prominent.The lack of resources from the forest and the adverse effects of corruption is very prominent in all those ecological problems.Huining County is located in the Longxi Loess Plateau,the central of Gansu province,which has characteristics of broken terrain,plenty ravines, dry climate,scarce rainfall and low vegetation coverage, so the ecological environment is very fragile.The major ecological crisis in Huining County are soil erosion and desertification etc. Over the past decade, people’s environmental awareness on the importance and urgency has increased significantly; the local government is also increased investment in forestry construction by vigorously supporting of national policy and economy. Area of protective forest has showed a significant, rapid growth. Give full play to the advantages of3S technology for monitoring and research of the afforestation can provide an important basis for the environmental protection and ecological construction decisions. However, new protective forest is always in young state which cause low vegetation cover rate, how to effectively monitoring remains to be explored.This paper which based on long timing Landsat satellite images,extrcacted the protective forest land cover information lied in the area of Longxi Loess Plateau north of Huining county, using object-based image analysis method and GIS spatial analysis techniques.A rule set which has strong applicability for extracting protective forest is established through features statistical analysis. On this basis, the paper discussed the area and distribution of protective forest land cover, anlyzed the variation trend, the main conclusions are as follows:1Medium resolution images of Landstat series had lots of advantages like free accsess, wide range cover etc. As Landsat8satellite had already been successfully lauched and operated, the continuity of image data has been secured.So Landsat images is still a better choice in the mornitoring of land cover aimed at large area and long time sequence.2There are plenty of ravines in Loess Plateau, the terrain there is very undulating and complex.Most of Protective forest located in the steep mountain or edge of tableland, with combination of image data and terrain data, the efficiency and accuracy of information extraction can be improved effectively.3The image segmentation is a key step in object-based analysis. This study is based on discrepancy measures to evaluate the image segmentation quality.Image segmentation results obtained from729kinds of parameter combinations were analysised compared to reference data set in two sides:geometric discrepancy and arithmetric discrepancy.On the calculation of PSE which shows geometric relationship and NSR which shows arithmetric relationship and a comprehensive evaluation of factors ED2, we obtained the optimal combination of segmentation parameters:scale:30, shape:0.2, compactness:0.8.4Classification based on rule set take the complexity, diversity and uncertainty of the world into consideration. Combining with expert knowledge in the field of geosciences and using variety of features obatained from object-based image analysis method like texture, shape, class-related relationship etc,it can achieve information extraction effectively.This paper aimed protective forest as main objective land cover type, features of spectral texture and shape were statistical analysised to identify the feature combination which have the largest difference and most conducive to.Based on this, rule set for information extraction was established.The results showed that extraction accuracy of protection forest achieved higher than86%which meant a effective classification.5By processing and statistical analysising the rule set classification results based on spatial analysis, the results showed that protection forest in the study area between1992and2011showed a rising trend from1992to2002, the growth area amount to2546hectares, an increase of16.7%. Between the year2002to2011, the growth area reached30,454hectares, which almost nearly quadruple. At the same time, the total amount of class cropland like irrigated land and dryland remained unchanged. The amount increased between1992-2002was small, which scattered in the study area.However, between2002-2011part of its growth throughout the study area which mainly distributed contiguous in the south of Touzhaizi Sifangwu Township, in the central of study area Baicaoyuan Township and Tumenjian Township. Also, compared with the increase amount of protection forest, the decrease amount was very little which basically could be negligible. |
PDF Full Text Request