CN-121981608-A - Agricultural heat index analysis method based on peak anchor point and bidirectional verification

CN121981608ACN 121981608 ACN121981608 ACN 121981608ACN-121981608-A

Abstract

The method comprises the steps of firstly generating a daily average air temperature sequence of a target area pixel by utilizing satellite remote sensing ground surface temperature data and geographic environment auxiliary data. And then calculating an air temperature trend sequence through symmetrical moving window moving average, and determining the maximum value in the air temperature trend sequence as a peak anchor point. And (3) carrying out forward and reverse bidirectional search along a time axis in the air temperature trend sequence by taking a peak anchor point as a reference, and selecting a downward-penetrating position and an upward-penetrating position of a preset limit temperature threshold value as a growth season final day point and a growth season initial day point of the determined candidates. And constructing a check window by taking the candidate point as the center, and rechecking the average air temperature sequence of the day to determine the initial day and the final day of the growing season. And finally, calculating pixel-level agricultural heat indexes to generate an agricultural heat resource distribution map. According to the method, multi-pixel parallel processing is realized through vectorization index and matrix comparison operation, and the calculation cost and the engineering complexity are remarkably reduced.

Inventors

ZHU XIN
HUANG RAN
WU KAIHUA
HU ZHIHAO

Assignees

杭州电子科技大学

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (8)

1. An agricultural heat index analysis method based on peak anchor points and bidirectional verification is characterized in that a pixel-level agricultural heat index is calculated to generate an agricultural heat resource distribution map by defining a growth season interval between a growth season early day and a growth season late day, and the agricultural heat index analysis method is characterized in that multi-time-phase satellite remote sensing earth surface temperature data and geographic environment auxiliary data are combined to generate a daily average air temperature sequence of a target area pixel by pixel; determining a candidate growing season early day point by taking a peak anchor point as a reference, reversely searching an upward-penetrating position of the temperature relative to a preset limit temperature threshold value along a time axis in a pixel-level air temperature trend sequence, and then forwardly searching a downward-penetrating position of the temperature relative to the preset limit temperature threshold value along the time axis; And respectively taking the candidate initial day point and the candidate final day point as centers, constructing an initial day check window and a final day check window, checking back the daily average air temperature sequence, determining the date which firstly satisfies the daily average air temperature not lower than the limit temperature threshold value as the initial day of the growing season in the initial day check window, and determining the date which finally satisfies the daily average air temperature not lower than the limit temperature threshold value as the final day of the growing season in the final day check window.
2. The agricultural heat index analysis method based on peak anchor point and bidirectional verification as recited in claim 1, wherein the pixel is selected from the group consisting of Longitude and latitude of (a) Altitude of sea Zenith angle of observation Number of year As auxiliary variable, and satellite remote sensing earth surface temperature observation information of multiple phases Together as an input feature vector : The daily average air temperature is measured by a weather station And (3) for the target value, establishing an eye air temperature estimation model by utilizing XGBoost algorithm, and generating an average air temperature estimation value of clear sky pixel by pixel.
3. The agricultural heat index analysis method based on peak anchor point and bidirectional verification as set forth in claim 2, wherein XGBoost algorithm is integrated Regression tree To approximate the slave feature vector Nonlinear mapping to daily average air temperature: Wherein, the Representing the space of the regression tree function, Representing a predicted daily average air temperature; algorithm training by minimizing an objective function with regularized terms The realization is as follows: wherein n represents the number of pixels; Is the first Complexity penalty term of the granularity tree; Is the first The number of leaf nodes of the tree, Is the first Tree first Predictive weights for individual leaf nodes; In order to control the penalty for the number of leaf nodes, For controlling leaf node weights Regularized intensity.
4. The agricultural heat index analysis method based on peak anchor points and two-way verification according to claim 2 is characterized by comprising the steps of selecting a clear sky image of satellite remote sensing as a reference image and other images as non-reference images, constructing XGBoost regression models by using the reference image and auxiliary variables, and carrying out prediction complementation on pixels of an image to be reconstructed, which are shielded by cloud layers, to generate seamless daily average air temperature data.
5. The agricultural heat index analysis method based on peak anchor point and bidirectional verification of claim 1, wherein a sliding window with the length of 5d is adopted for smoothing a daily average air temperature sequence to construct an air temperature trend sequence { }: Wherein, the Mean daily air temperature, d=1, 2, represents the d-th natural day of the year, , Is the natural day of the current year Beyond the limit of And when the range is in the range, supplementing the air temperature data of the missing sequence from the adjacent years.
6. The agricultural heat index analysis method based on peak anchor point and bidirectional verification according to claim 1, wherein when extreme value index operation is performed on the pixel-level air temperature trend sequence, if a plurality of daily sequences have the same maximum value, the earliest daily sequence is selected as the peak anchor point.
7. The agricultural heat index analysis method based on peak anchor point and bidirectional verification according to claim 1, wherein the duration and the effective accumulation temperature of the growing season are calculated according to the determined initial day and the final day of the growing season, and the agricultural heat resource space map of the target area is output.
8. A computer readable storage medium having stored thereon a computer program which, when executed in a computer, causes the computer to perform the method of any of claims 1 to 7.

Description

Agricultural heat index analysis method based on peak anchor point and bidirectional verification Technical Field The invention belongs to the technical field of weather and agriculture, relates to remote sensing air temperature data processing and agricultural climate resource analysis, and particularly relates to an agricultural heat index analysis method based on peak anchor points and bidirectional verification. Background In agricultural weather, climate division and crop planting suitability analysis, heat resources are one of the most basic and critical ecological factors. The effective accumulated temperature calculated according to the early day and the final day of the growing season is widely applied to the prediction of the potential growing season of crops, the selection guidance of the sowing time, the division of the farming zone and the evaluation of climate resources. In the existing agricultural climate analysis, the early days and the final days of the growing season are usually determined according to a limit temperature threshold value, an average air temperature sequence of a certain year is searched, when the air temperature stably passes through a given threshold value in spring, the initial date meeting the condition is taken as the early day, and the corresponding boundary date is taken as the final day after the air temperature stably falls below the threshold value in autumn. However, in the practical application process, the air temperature sequence near the threshold value may have short-term fluctuation caused by the cooling and heating process, so that the phenomenon of multiple crossing of 'up-going-back-going-up' or 'down-going-back-up-going-down' occurs, and therefore, the accidental warm day or cold day is easily misjudged as the initial day or the final day. And under different areas and different annual scenes, the fixed window length and the criterion parameters are difficult to adapt, and the uncertainty of boundary identification is further amplified. In the existing research, the extraction and analysis of agricultural heat resources still mainly depend on a daily air temperature observation sequence of a meteorological site. Because the weather stations are very sparsely and unevenly distributed in space, there is often a lack of adequate observation coverage, especially in farmlands, mountainous areas, or even wide unmanned areas. The heat resource analysis based on the site data has obvious limitation on the spatial scale, and the requirements of developing refined research and application on agricultural heat resource distribution in the area are difficult to meet. In addition, when the heat resource index is extracted in the pixel scale, the significant calculation cost and engineering difficulty brought by large-scale high-resolution data are required, and the technical process capable of realizing large-scale production is difficult to form. Therefore, there is an urgent need for an agricultural caloric resource generation and analysis that can accurately and rapidly achieve high spatial resolution. Disclosure of Invention Aiming at the defects of the prior art, the invention provides an agricultural heat index analysis method based on peak anchor points and bidirectional verification, which uses a remote sensing tile (tile) as a data processing unit, and synchronously executes trend sequence construction, peak anchor point positioning, candidate initial final day point retrieval and window check on time sequences of all pixels in the tile, so as to obtain accurate and reliable pixel-level heat indexes and space distribution patterns, and solve the problems of low spatial resolution, limited coverage, dependence on weather station sparse observation data and lack of scale application capability in the existing agricultural heat index analysis technology. An agricultural heat index analysis method based on peak anchor points and bidirectional verification comprises the following steps: Step one, constructing a seamless air temperature data set with high space-time resolution Acquiring multi-time-phase satellite remote sensing earth surface temperature data and geographic environment auxiliary data, constructing a nonlinear mapping model of clear sky earth surface temperature and air temperature, inverting clear sky average air temperature of a target area, reconstructing pixel by pixel of cloud coverage and missing areas, generating space-time continuous and high-spatial-resolution daily average air temperature data, and providing high-precision basic data support for subsequent agricultural heat index analysis. Step two, positioning peak anchor points And (3) performing symmetrical moving window moving average on the daily average air temperature data generated in the step one in a tile scale to obtain a pixel-level air temperature trend sequence. And when the sliding window crosses the boundary, calling the data of the corresponding sequence of the adjacent