On May 26, the China International Big Data Industry Expo (Big Data Expo), released leading scientific and technological achievements. The "Inversion Technology of Key Parameters in Agricultural Meteorological Remote Sensing Based on Artificial Intelligence" developed by Professor Mao Kebiao, won the "2023 Leading Scientific and Technological Achievement Award" at the Expo. Mao Kebiao is a researcher of the Institute of Agricultural Resources and Regional Planning (IARRP) of the Chinese Academy of Agricultural Sciences (CAAS).

In recent years, artificial intelligence (AI) technology has sparked a research boom in academic and engineering applications. It has also shown outstanding application potential in the inversion of geophysical parameters and agricultural meteorological remote sensing parameters. At present, most of the applications of AI technology in geosciences and agronomy are still "black boxes", and the applications have no physical meaning or lack of interpretability and versatility.

The research team has conducted many original research work on artificial intelligence geophysical parameter inversion through nearly 20 years of research (https://b23.tv/MbrOAvS). (1) A paradigm theory and criteria for geophysical parameter inversion based on artificial intelligence coupled physics and statistical methods have been proposed; (2) A paradigm for simultaneously inverting soil moisture and surface temperature based on artificial intelligence has been proposed (video lecture: https://b23.tv/Ln7PQhO); (3) The normal form of simultaneous retrieval of land surface temperature and Emissivity based on artificial intelligence is proposed (video lecture: https://v.douyin.com/DH91RfF/); (4) A paradigm for retrieving near surface air temperature based on artificial intelligence has been proposed. (video lecture: https: //v.douyin.com/DHH5Md9/); (5) A paradigm for inversion of atmospheric water vapor content based on artificial intelligence is proposed. This theory seeks to construct physical methods theoretically through physical and logical reasoning, and to build generalized statistical methods. It utilizes physical and representative statistical methods generated from multi-source data to deconstruct deep learning training and testing databases, thereby achieving the integration of deep learning with both physical and statistical methods.

In addition, the team proposed conditions for determining the formation paradigm: (1) there must be a causal relationship between the input and output variables of deep learning; (2) complete closed equation groups can be theoretically constructed between the input and output parameters.

The review committee of the expert group of the China International Big Data Industry Expo highly affirmed the research results. The introduction of the paradigm theory for artificial intelligence-based geophysical parameter inversion is a milestone in the history of geophysical parameter inversion. It breaks through the limitations of traditional methods and systematically improves the accuracy of parameter inversion, providing theoretical and technical support for creating "ChatGPT" based on artificial intelligence geophysical parameter inversion.

Paradigm Theory and Judgment Conditions for Inversion of Geophysical Parameters

2023 leading scientific and technological achievements release video lecture: https://b23.tv/MbrOAvS 

The Leading Scientific and Technological Achievement Award is a national award officially registered by the National Science and Technology Award Office. It is the only social and industrial science and technology award with the theme of big data. It has high authority and has been recognized by the industry.