A Study On Retrieving Information Of Mycosphaerella Larici-leptolepis Ito Etal Based On Remote Sensing

Plant diseases and insect pests in forest is a main enemy of forestry, it is called fire disaster without smoking, so its dangers is obvious. Originally, the plant diseases and pests occurs in small scale and sparsely not in big and continuous areas. It should not be ignored to monitor in early days. The small-scaled and sparse distribution, after long-time devastating circulation, will lead to big area at last, and then it can be monitored, but the danger may have been out of control. So it is very important to monitor the diseases and pests in time or early.This paper takes mu dan gang forestry in Jilin province for example. At first, it analyzed the spectral features of disease larch and the relationship between it and chlorophyll, and discussed the coupling degree between the spectrum in remote sensing image and that in field measurement; Then, it used image datum, and combined GPS and forest chart datum to build a database on retrieving the information of Mycosphaerella larici-leptolepis Ito etal., and use decision tree and object - oriented methods respectively to retrieve the disease information , based on the idea of layered and zonal classification. Finally, evaluate the retrieving accuracy, and analyze the monitoring results.The main content of this paper and the results, as followed:1 the spectral reflective features change after getting diseases, base on which remote sensing technique diagnose and monitor the forest diseases. The difference between the spectral features of diseased and healthy larch can be described as following five respects: "blue valley" and "red valley", "green peak", "red edge", near infrared high reflective region and water absorption zone.2 the position of "red edge" inpixelion of the reflective spectrum of larch under different disease degree have good linear relationship with chlorophyll a ,T , and the correlation coefficient is highest with chlorophyll a.3 regression analyses on the spectral datum of field measurement and the reflectivity of corresponding pixel in SPOT image, the result indicate that their relativity is common, and show that the reflective information in wide band image is not abundant and detailed enough. It represent the average value in a pixel, we can't finish retrieving disease information through single band, so it is necessary to combine several bands.4 according to remote sensing datum, ground spectral datum, forest phasic chart, GPS datum and scenic picture, comprehensive database of retrieving early falling diseases information was built, which mainly conclude remote sensing image database, remote sensing knowledge database, secondary information database and interpretation symbol database.5 base on the idea of layered and zonal classification, taking decision tree and object-oriented measurement respectively to realize retrieving early falling disease information, and making a comparison. The result shows that the accuracy base on decision tree is lower, while the latter don't only enhance the classification accuracy but also can avoid "Pepper and Salt Phenomenon"effectively.6 through retrieving the information of Mycosphaerella larici-leptolepis Ito etal., we found that the sick area is 2.165 km2, in which the light danger area is 1.6 km~2; this disease distribute on southeast direction much and dumpily , and then northeast, the rest places are sparse.