Method for realizing large-area near real-time monitoring on phenological period of rice based on compactly polarimetric radar

Abstract

The invention discloses a method for realizing large-area near real-time monitoring on the phenological period of rice based on a compactly polarimetric radar. The method comprises the steps of (I) planning preparation and field measurement, namely selecting an appropriate radar data source, planning the time of field synchronous measurement by use of satellite crossing time, carrying out field synchronous measurement to obtain a GPS vector file and the data of rice sample point, and dividing the rice sample points into two sets, namely a set A1 of training samples and a set A2 of verification samples, (II) rice classification, namely obtaining appropriate compactly polarimetric radar images, preprocessing and extracting and analyzing key radar parameters, establishing a classification algorithm to obtain a result and performing classification accuracy evaluation by use of the verification samples, and (III) phenological inversion, namely analyzing each sample in the set A1, extracting and analyzing the key radar parameters, establishing an inversion algorithm, verifying the accuracy by use of the samples of the set A2, and adjusting and determining the effective radar parameters and the effective algorithm. The method for realizing large-area near real-time monitoring on the phenological period of rice based on the compactly polarimetric radar has the advantages of large monitoring range, high refinement and good practical generalization performance.

Description

technical field [0001] The invention relates to the field of crop phenology monitoring based on remote sensing data, in particular to a large-area near-real-time monitoring method for rice phenology based on compact polarization radar. Background technique [0002] Phenology refers to a natural phenomenon that occurs with a yearly cycle under the influence of environmental factors and human activities. The phenological period of a crop is the date corresponding to the critical growth period of the crop. The phenological information of crops not only reflects the local environmental conditions at that time, but also reflects the accumulation of environmental conditions in the past period of time. As a part of the terrestrial ecosystem, the phenology of crops is affected by changes in climate, hydrology and other factors, which will directly affect the final yield of crops. At the same time, the phenological information of crops is important agricultural information, an impo...

AI Technical Summary

Problems solved by technology

[0010] (1) At present, the identification of rice by remote sensing only stays in the classification of rice and non-rice. There is no effective method for the classification of different varieties of rice, such as hybrid rice and japonica rice, which cannot satisfy the monitoring of different varieties of rice. and appraisal needs;

[0011] (2) An important defect of radar remote sensing inversion of rice phenology is that the obtained rice phenology is not precise enough, that is, for the 10 growth stages of rice, the existing remote sensing inversion technology can only distinguish 4-5 of them Phenophases, some of which are so small that they cannot be distinguished

For example, when Spanish scholars inverted the rice phenology in 2011, they could not distinguish between the heading stage and the flowering stage, milky stage and full-ripening stage of rice, which are very similar phenological stages.

The growth of rice in these phenological periods is ambiguous, obviously unable to meet the needs of increasingly sophisticated agricultural applications;

[0012] (3) In the existing application of radar remote sensing inversion of rice phenology, some of the images used are relatively small in width (such as fully polarized Fine Quad mode images), and can only monitor the growth of rice within a county or city Situation; Some widths are relatively large (such as dual-polarization ScanSAR mode images), and the accuracy is not enough

[0013] In summary, the existing methods have limitations in the cognitive scale and deficiencies in the identification methods in the actual application of farmland monitoring.

Images