Multi-scale crop phenological period remote sensing dimensionality reduction prediction method

Abstract

The invention relates to a multi-scale crop phenological period remote sensing dimensionality reduction prediction method. The method comprises the following steps: 1, obtaining multi-year remote sensing data and phenological data of a certain crop in a plurality of planting areas; 2, preprocessing the remote sensing data; 3, performing dimension reduction on the preprocessed remote sensing data to generate input data with variable time sequence length; 4, establishing a crop identification and prediction model by using a long short-term memory network, the established crop identification and prediction model being a multi-output model, and integrating the identification function of the current phenology and the prediction function of the next phenology stage into one model; 5, training a crop identification and prediction model; 6, testing the crop identification and prediction model; and 7, applying the crop phenological period identification and prediction model to crop phenological period identification and prediction in any growth stage and any size area. The method can be applied to phenology identification and prediction of different scales, and can meet the requirements of field scale and monitoring and prediction of administrative scales of all levels such as villages, towns, counties, cities and provinces.

Description

1. Technical field: [0001] The invention relates to the technical field of crop production, in particular to a multi-scale crop phenological period remote sensing dimension reduction prediction method. 2. Background technology: [0002] Agriculture is the basis for human survival and development. The acquisition and prediction of large-scale agricultural information is of great significance for guiding agricultural production and ensuring food security. Crop phenology is a cyclical natural phenomenon in annual units caused by crops affected by the environment and human activities. [0003] Crop phenology is important information for crop growth. The identification and prediction of crop phenology period is an important method for crop species identification, classification and crop yield estimation. An important basis for making planning decisions, etc. [0004] The traditional method of phenological identification adopts the method of artificial field observation, which i...

AI Technical Summary

Problems solved by technology

For example, Chinese patent application 202011432779.6 "Method and System for Time Point Prediction of Plant Key Phenological Periods Based on Deep Learning" establishes a prediction model based on meteorological data, without considering human factors, varieties, planting time, etc.

However, the formation mechanism of the specific phenological periods of crops is relatively complicated, and it is difficult to fully consider all factors.

[0008]At present, in the case of using remote sensing data for representation learning in order to obtain a certain recognition or prediction result, such as phenological period prediction, crop yield prediction and other applications, model training and The use must be based on the same scale. For example, in some studies, models trained at the field scale can only be used for field-scale predictions, and models trained at the county-level scale can only be used for county-level predictions.

Images