Research On The Method Of Crop Area Measurement Based On GF-1 Remote Sensed Data

Under the background of 3th national agricultural census,aiming at the primary task of the census,conducting an investigation to crop area and spatial distribution accurately,combining the problems existing in the automatic extraction of crop area,this paper proposes a kind of measurement results for large region crop method effectively and quickly,gives basic data supports for crop area estimation,so as to provides a scientific evidence for agricultural development plan.According to the difficulty level of remote sensing measurement,this paper proposes three levels of measurement region,plain region of Ningxia province,terraced fields of Gansu province and broken region of Guizhou province are selected as the representative region for the three levels.Combining the phonological data and GF-1 WFV-16 m data in study region,the domestic remote sensing datum which acquires during the March and April in 2016 are selected for crop area measurement.Data source is given priority by GF-1,and complementary by TH-1.In order to improve the efficiency of remote sensing data processing,with the characteristic of GF-1 and TH-1 panchromatic and muti-spectral data acquiring at the same time but poor accuracy of matching,the paper selects the method of “correct and fusion first,ortho-image rectification later” to process the remote sensing data bulkly and fastly.In the study region,ortho-image rectification is made as results respectively.By using object-oriented automatic classification for 4-band orthogonal projection data;and artificial visual interpretation for 3-band orthogonal projection data after ture color synthesis,the crop area is measured respectively.Finally,taking the field sample survey datum as true value,the results by the two methods are analyzed from visual effect,accuracy of measurement and measuring time.Through the research,the following results are given in this paper:(1)In study region,the GF-1 WFV-16 m datum which is acquired at before the crop,vigorous growth period,after harvest are compared and analyzed,combining with the phonological data,and based on single-phase of the high resolution remote sensing data,the crop area measurement is completed.As a result,the paper proposes a rapidly,largely region crop measurement method which is made full use of large wide,short return period of GF-1 WFV 16 m data and high resolution and fusion data of GF-1.(2)According to the characteristics of the GF-1 and TH-1 data,using the processing of “correct and fusion first,ortho-image rectification later”,combining the advantage of remote sensing data processing software fully,the making process of DOM are summed up,such as registration,band combination,fusion,correcting,mosaic,color adjustment,cutting and so on.A series of methods for batching,processing quickly are included.A solution of making lots of DOM production is provided for large-scale crop remote sensing measurements in the future.(3)This paper selects the methods of object-oriented classification and artificial visual interpretation,crop areas are measured using remote sensing in three typical study region,and taking the field sample survey datum as true value,analyzing the results of two kinds of measurement.The results show that two kinds of measurement are likeness on the spatial distribution;the total accuracy can be up to 90% in three typical study region,which could satisfy the application requirements.however,in the time of processing,comparing with the artificial visual,object-oriented classification method can prove double efficiency.For another,with the increasing of measure region,the speed is more obvious.There,when you need to get a wide range of agricultural remote sensing measurement results quickly,using object-oriented automatic classification method is a better choice.