A Crop Yield Estimation Method Based on GEE Integrated Remote Sensing Image and Deep Learning Method

Abstract

The invention discloses a crop yield estimation technology based on GEE comprehensive remote sensing image and deep learning method, including: S1: using the GEE platform to spatially match the remote sensing data and meteorological data within the research period with the geospatial data; S2: extraction research The annual grid planting area of ​​the crops in the research period in the region; S3: use the GEE platform to extract the monthly value influencing factors and soil attribute factors in the key growth period of the crops; S4: calculate the particle spatial position information of the crop grids in each county ; S5: use the deep learning framework keras platform to build a machine deep learning yield prediction model, and realize the localization of the model; S6: use the localized model to predict the crop yield of the year to be predicted in the research period.

Description

technical field [0001] The invention relates to the technical field of agricultural remote sensing, in particular to a crop yield estimation technology based on Google Earth Engine (GEE) platform integrated remote sensing images and deep learning methods. Background technique [0002] Agriculture is the source of basic materials for human social life and the foundation of the national economy. Food production is the core of the agricultural sector, and food security is related to the survival of the people and the stability of the country. Timely, accurate, and large-scale monitoring of crop growth status and forecasting of grain production are of great significance to protecting the interests of farmers and ensuring the food security of the region and the whole country. At present, the mainstream yield estimation models are divided into four categories: statistical yield estimation models, light energy utilization efficiency models, crop growth models and coupling models. ...

AI Technical Summary

Problems solved by technology

Crop growth mechanism models are relatively complex, requiring the input of a large number of precise parameters, and the acquisition cost of some parameters is very high, which restricts the promotion of crop models in a large area, so this type of model is mainly limited to point-scale yield estimation and prediction

The coupling model has great uncertainty in the assimilation method, data selection and assimilation accuracy, and requires a large amount of input data, the calculation time is too long, and the accumulation of errors often leads to the final estimation accuracy is not high enough

Therefore, the current yield estimation research is mostly based on administrative (national, provincial, city-county), point (such as agricultural gas stations, experimental fields) and field scales, and there is still a lack of gridded, high-precision, and large-scale yield estimation research

Images