A rice maturity recognition method based on radar time series observation and temperature analysis

Abstract

The present application belongs to the technical field of spaceborne radar data processing and remote sensing mapping, and particularly relates to a rice maturity recognition method based on radar time series observation and temperature analysis. The present application realizes the capture of diversified periodic characteristics of time series scattering and the detection of scattering wave troughs through time series reconstruction and wave trough recognition, determines the potential phenology period time corresponding to the scattering wave troughs through potential rice phenology estimation, evaluates the temperature suitability of the potential rice phenology period through the temperature limitation of rice phenology, eliminates the scattering wave troughs that do not meet the temperature condition in combination with the rice growth mechanism and the rice maturity law, realizes the recognition of the rice scattering wave troughs and the correction of the overestimation of rice maturity, and finally realizes the recognition of the rice maturity area, is suitable for large-area rice maturity monitoring under the conditions of obvious differences in natural and social conditions, complex and diverse farming practices and frequent meteorological interference of clouds and mists, and solves the three problems of time series scattering diversity, lack of prior phenology information and overestimation of rice maturity existing in the prior art.

Description

Technical Field

[0001] This invention belongs to the field of spaceborne radar data processing and remote sensing mapping technology, specifically relating to a method for identifying rice ripening status based on radar time-series observation and temperature analysis. Background Technology

[0002] Rice ripening system refers to the number of times rice is planted in a paddy field within a year. Depending on farming practices, single-season, double-season, and triple-season rice may be planted in the field. Accurately understanding the rice ripening system is crucial for grain yield estimation and agricultural policy formulation. Remote sensing observation is an important means of monitoring rice ripening systems over large areas. By conducting long-term observations of paddy fields, dynamic monitoring of rice planting can be achieved. Optical satellites are the main data source for remote sensing monitoring of rice ripening systems. However, the rice growing season is often cloudy and rainy, and optical imaging is easily affected by unfavorable weather conditions. Therefore, high-resolution rice ripening system monitoring based on optical satellite data can usually only be successful in small sample areas, while large-area rice ripening system monitoring can only rely on low-resolution optical satellite data, making it difficult to guarantee accuracy.

[0003] Radar can provide stable and high-quality imaging unaffected by weather conditions, making radar-equipped satellites an increasingly important alternative data source for rice ripening monitoring. Long-term radar observations can acquire temporal scattering data from paddy fields, allowing analysis of the dynamic changes in the physicochemical properties of rice plants and soil under different ripening conditions. However, large-scale rice ripening monitoring based on temporal scattering data faces three unresolved issues: the diversity of rice temporal scattering due to varied farming practices; the ambiguity between rice and non-rice scattering characteristics due to a lack of prior phenological information; and the overestimation of rice ripening conditions due to abnormal farmland changes. These three issues severely restrict the operational application of radar data in large-scale rice ripening monitoring. Summary of the Invention

[0004] To address the aforementioned problems and shortcomings, and to resolve the three major issues existing in current methods for identifying rice maturity using radar data—namely, the diversity of temporal scattering, the lack of prior phenological information, and the overestimation of rice maturity—this invention provides a method for identifying rice maturity based on radar temporal observation and temperature analysis. This method captures diverse periodic characteristics of temporal scattering and detects scattering troughs through temporal reconstruction and trough identification. It determines the potential phenological period corresponding to the scattering trough by estimating potential rice phenology, assesses the temperature suitability of the potential rice phenological period through rice phenological temperature constraints, and eliminates scattering troughs that do not meet the temperature conditions by combining rice growth mechanisms and rice maturity patterns. This achieves the identification of rice scattering troughs and the correction of overestimation of rice maturity, ultimately enabling the identification of rice maturity regions.

[0005] A method for identifying rice ripening status based on radar time-series observation and temperature analysis, the specific steps of which are as follows:

[0006] Step 1. Temporal reconstruction and trough identification;

[0007] Obtain the annual radar VH (Vertical transmission and Horizontal reception) polarization time-series scattering data S[t], where t is the normalized annual day, ranging from 0 to 1. Use the following formula for harmonic fitting:

[0008]

[0009] 'a' is a constant term representing the time-series scattering mean. 'i' takes values ​​of 1, 2, and 3, representing the order of the cosine term. i Represents the magnitude of the i-th cosine term. Let a represent the phase of the i-th cosine term. A and A' are obtained using least-squares fitting. i and Substituting the value of into the above equation, we obtain the reconstructed temporal scattering.

[0010] The first difference S′[t] of S[t] is calculated using the following formula:

[0011]

[0012] The value of t when S′[t] is 0 is solved using Newton's method. If S′[t-1]<0, S′[t+1]>0, and S[t]<0.02, then t is the time when the trough of the scattered wave appears, and the actual accumulated days corresponding to t are d, where d is in days.

[0013] d = 365t.

[0014] Step 2. Estimation of potential rice phenology;

[0015] Phenological stages: The times of rice seedling raising stage, transplanting stage, vegetative growth stage, reproductive growth stage, and maturity stage are respectively represented by D. S D T D V D R and D M This indicates that the annual average daily temperature data is obtained, and the duration P of the cold (<10℃) period throughout the year is calculated. C P C The unit is days. For the target area:

[0016] When P C ≠0, d>240, d potentially corresponds to D R ,at this time:

[0017] D S =d-90,D T =d-60,D V =d-30,DR =d,D M =d+30;

[0018] When P C =0 or d≤240, where d potentially corresponds to D T ,at this time:

[0019] D S =d-30,D T =d,D V =d+30,D R =d+60,D M =d+90.

[0020] Step 3. Temperature restrictions for rice phenology;

[0021] For each phenological stage (rice seedling raising stage, transplanting stage, vegetative growth stage, reproductive growth stage, and maturity stage), the temperature was obtained 15 days before and after each stage, and the average temperature E was calculated. The average temperature for each stage (rice seedling raising stage, transplanting stage, vegetative growth stage, reproductive growth stage, and maturity stage) is represented by E. S E T E V E R and E M express.

[0022] For the target scattering valley, when:

[0023] E S >10℃, E T >10℃, E V >18℃, E R >18℃, E M >10℃

[0024] If it is a valid valley, then it is considered a valid valley; otherwise, it is considered an invalid valley and is removed.

[0025] The number of effective valleys N in statistical temporal scattering is used to evaluate the maximum rice ripening period S that the target area can support, according to the following rules:

[0026]

[0027] For the target area, if N>S, it is determined that the rice maturity level has been overestimated. For each effective trough, calculate the sum of the average temperatures of each phenological stage, Sum:

[0028] Sum = E S +E T +E V +E R +E M

[0029] Remove valid valleys in order of increasing Sum value. The number of valid valleys to be removed is NS. The remaining valleys are the rice valleys.

[0030] Step 4. Identification of Rice Maturation Areas

[0031] Obtain digital elevation data for the target area, extracting altitude and slope; obtain land use products, extracting arable land distribution data. Count the number of rice troughs in the area to identify rice ripening stages.

[0032] Furthermore, in step 4, the threshold conditions for extracting altitude and slope are to retain only cultivated land with an altitude below 1000m and a slope of less than 5°.

[0033] The principles involved in steps 1 to 4:

[0034] The principle of temporal reconstruction and trough identification: Radar signal coherence, irregular rainfall, and field agricultural management practices (such as irrigation, drainage, fertilization, and weeding) all lead to random noise in temporal scattering. Differences in rice ripening and agricultural phenology at the field scale result in diverse and complex temporal scattering patterns. Rice can be repeatedly planted up to three times a year; therefore, the temporal reconstruction method based on third-order cosine harmonic decomposition can effectively overcome noise interference while preserving the dominant periodicity of temporal scattering. The most important features in temporal scattering are peaks and troughs. The appearance of peaks can be influenced by various factors such as rice growth, increased soil moisture, and increased field roughness, making them unreliable indicators of rice planting cycles. Low-value (<0.02) troughs are only related to field irrigation during rice cultivation. After third-order cosine harmonic fitting, only one trough lasting longer than 30 days exists in a single rice planting cycle; therefore, low-value troughs can serve as markers of rice planting cycles.

[0035] The principle of estimating potential rice phenology: A complete rice planting cycle includes the seedling stage, transplanting stage, vegetative growth stage, reproductive growth stage, and maturity stage, each lasting approximately 30 days. Rice typically spends its seedling stage (30 days) in a specific seedbed, and the remaining 120 days are completed in the paddy field (fields where crops are grown on a large scale). The scattering trough is related to field irrigation during rice cultivation. The most significant scattering trough within a rice planting cycle is usually caused by field irrigation during the transplanting stage. However, if the temperature in the rice-growing area is not above 10°C year-round, and rice is transplanted after July, farmers will implement heat preservation irrigation during the reproductive growth stage (mid-September), resulting in the most significant scattering trough during this period. Therefore, when the temperature in a region is not above 10°C year-round, and the scattering trough occurs after September, this trough should be considered a potential reproductive growth stage for rice, and the other four rice phenological stages are estimated based on 30-day intervals. In other cases, the scattering trough should be considered as the potential rice transplanting period, and the other four rice phenological periods are also calculated based on a 30-day interval.

[0036] The principle of temperature limitation in rice phenology: 10℃ is the lower limit of temperature for rice plant survival; all phenological stages within a rice planting cycle should be supplied with temperatures above 10℃. 18℃ is the lower limit of temperature for effective accumulation of photosynthetic products in rice; both the vegetative and reproductive growth stages within a rice planting cycle should be supplied with temperatures above 18℃. Therefore, after calculating the duration of each phenological stage corresponding to the scattering trough, it is necessary to assess whether each phenological stage meets the temperature requirements for rice growth. Only when the temperature of all phenological stages reaches the lower limit requirement is the scattering trough considered an effective trough. Furthermore, when the number of effective troughs exceeds the maximum rice maturity that the region can support, it is judged that there is an overestimation of the rice maturity. The maximum rice maturity that a region can support is assessed by calculating the duration of the cold period (<10℃), which is the period of the year when rice cultivation is impossible. Considering that a rice cropping cycle typically takes 120 days in the field, when the cold period is 0 days, the remaining time throughout the year is sufficient for three rice cropping cycles, and the maximum rice maturity system the region can support is 3. When the cold period is between 0 and 120 days, the remaining time throughout the year is sufficient for two rice cropping cycles, and the maximum rice maturity system the region can support is 2. When the cold period is between 120 and 240 days, the remaining time throughout the year is sufficient for one rice cropping cycle, and the maximum rice maturity system the region can support is 1. When the cold period exceeds 240 days, the remaining time throughout the year is insufficient to complete one rice cropping cycle, and the maximum rice maturity system the region can support is 0. For areas where the rice maturity system is overestimated, the difference between the number of effective troughs and the maximum rice maturity system the region can support is calculated; this difference represents the number of effective troughs that need to be removed. Rice has higher temperature requirements than other terrestrial processes. The temperature accumulation of the potential phenological period corresponding to its scattering valley should be higher than that of other terrestrial processes. Therefore, the effective valley with lower temperature accumulation is the valley that needs to be removed.

[0037] The principle of rice ripening area identification: After identifying the troughs corresponding to rice planting cycles in the temporal scattering, it is necessary to further restrict the identification area of ​​rice ripening. The need for hot and humid climate for rice means that paddy fields are mainly distributed below 1000 meters above sea level, while the water storage requirements of paddy fields mean that they are mainly distributed on slopes below 5°. In addition, the cultivated land area in the land use product indicates the potential distribution of paddy fields. Therefore, combining altitude, slope and land use product can further reduce the impact of scattering fluctuations in non-rice areas on rice ripening identification and improve mapping accuracy.

[0038] In summary, this invention achieves the capture of diverse periodic features of temporal scattering and the detection of scattering troughs through temporal reconstruction and trough identification. It determines the potential phenological period corresponding to the scattering troughs by estimating potential rice phenology, assesses the temperature suitability of potential rice phenological periods by limiting rice phenological temperature, and eliminates scattering troughs that do not meet temperature conditions by combining rice growth mechanisms and rice ripening patterns. This enables the identification of rice scattering troughs and the correction of overestimation of rice ripening, ultimately achieving rice ripening area identification. It is suitable for large-area rice ripening monitoring where natural and social conditions differ significantly, farming practices are complex and diverse, and cloud and fog weather interference is frequent. This invention solves the three major problems of temporal scattering diversity, lack of prior phenological information, and overestimation of rice ripening in current methods of using radar data for rice ripening identification. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the process of the present invention;

[0040] Figure 2 The temporal reconstruction and trough identification diagram is shown in the example.

[0041] Figure 3 This is a potential rice phenological estimation diagram for an example.

[0042] Figure 4 The rice phenological temperature limitation diagram is shown in the example.

[0043] Figure 5 This is an example of a rice ripening area identification map. Detailed Implementation

[0044] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0045] A method for identifying rice ripening status based on radar time-series observation and temperature analysis (e.g.) Figure 1 As shown in the figure, the development environment in this embodiment is GEE (Google Earth Engine), and the programming language is JavaScript.

[0046] Step 1: Retrieve Sentinel-1 ascending VH polarization data for Leizhou City, Guangdong Province in 2020. Use the imageCollection function to construct the time-series scattering, and use the image.linearRegression function to perform time-series harmonic fitting to obtain the undetermined coefficients a and A. i and Reconstructing the temporal scattering. First-order difference operations on the temporal scattering are performed using the GEE mathematical operation library, and scattering valley identification is performed using the GEE logic operation library to extract the valley date d.

[0047] Step 2: Retrieve the daily average temperature data of ERA5-Land in Leizhou City, Guangdong Province in 2020, and count the duration P when the temperature is below 10℃. C According to P C The values ​​of d and d are used to determine the potential phenological period of rice corresponding to the trough of the scattered wave. The remaining phenological periods are then calculated at 30-day intervals to obtain D. S D T D V D R and D M .

[0048] Step 3: Use the Filter.date function and the reduce('mean') function to calculate the average temperature E for each phenological stage. S E T E V E R and E M The GEE logic operation library is used to determine whether the temperature at each phenological stage meets the growth requirements of rice. Valleys that do not meet the temperature conditions are removed, and the remaining valleys are considered valid valleys. The GEE logic operation library is used to calculate the maximum rice maturity S that each region can support. When the number of valid valleys N is greater than S, the Sum value of each valid valley is calculated, and N / S valid valleys with lower Sum values ​​are removed, with the remaining valleys being considered rice valleys.

[0049] Step 4: Retrieve elevation, slope, and farmland distribution data from NASADEM and ESA WorldCover respectively. Use the GEE mathematical operation library to extract farmland areas with elevations below 1000m and slopes below 5°, and count the number of rice valleys at different locations within these areas to identify rice ripening areas.

[0050] This embodiment processes time-series Sentinel-1 radar data, ERA5-Land temperature data, NASADEM, and ESAWorldCover data. Figure 2 The temporal reconstruction and trough identification diagram is shown in the example. Figure 3 This is a potential rice phenological estimation diagram for an example. Figure 4 The rice phenological temperature limitation diagram is shown in the example. Figure 5 This is an example of a rice ripening area identification map.

[0051] As can be seen from the above embodiments, the present invention has achieved rice ripening identification in Leizhou City, Guangdong Province in 2020, with an accuracy of 79.25%. The method provided by the present invention effectively captures the temporal scattering periodic characteristics and potential rice scattering attenuation, determines the key phenological period of potential rice planting cycles, assesses the temperature suitability of potential rice phenological periods, and, combined with rice growth mechanisms and rice ripening patterns, identifies rice scattering troughs, corrects overestimation of rice ripening, and realizes rice ripening monitoring under the background of significant differences in natural and social conditions, complex and diverse farming practices, and frequent cloud and fog weather interference. It can promote the operational application of radar data in large-area rice ripening monitoring and provide information support for food security and social stability.

Claims

1. A rice maturity recognition method based on radar timing observation and temperature analysis, characterized in that, The following steps are performed in Google earth engine: Step 1. Time series reconstruction and trough identification; Obtain annual radar VH polarization time series scattering data S[t], t is the normalized annual day, the value range is from 0 to 1, use the following formula for harmonic fitting: where a is a constant term representing the mean of the time series scatter; i takes values 1, 2, 3 in turn, representing the order of the cosine term; A i represents the amplitude of the i-th order cosine term, represents the phase of the i-th order cosine term; the values of a, A i and are obtained by using least squares fitting, and substituted back into the above formula, i.e. to obtain the reconstructed time series scatter; where the imageCollection function is used to construct the time series scatter, and the image.linearRegression function is used to realize the time series harmonic fitting; Calculate the first-order difference S'[t] of S[t], the formula is as follows: Use Newton method to solve t value when S'[t] is 0, if S'[t-1]<0, S'[t+1]>0, S[t]<0.02, then t is the time of scattering trough, the actual annual day corresponding to t is d, the unit of d is day; d = 365t Step 2. Potential rice phenology estimation; Phenological stage: the time of rice seedling stage, transplanting stage, vegetative growth stage, reproductive growth stage and maturity stage is represented by D S , D T , D V , D R and D M respectively; obtain annual average daily temperature data, and count the length of cold period P C , P C in days in the whole year, and the cold period refers to the temperature <10℃; for the target region: When P C ≠ 0, d > 240, d potentially corresponds to D R At this time: D S = d - 90, D T = d - 60, D V = d - 30, D R = d, D M = d + 30; When P C = 0 or d < 240, d potentially corresponding to D T In this case: D S = d - 30, D T = d, D V = d + 30, D R = d + 60, D M = d + 90; Step 3. Rice phenology temperature limit; For each phenological period, the temperature of 15 days before and after the period was obtained, and the average temperature E was calculated; the average temperatures of the rice seedling stage, transplanting stage, vegetative growth stage, reproductive growth stage, and maturation stage were represented by E S , E T , E V , E R , and E M , respectively. For the target scattering trough, if: E S >10 °C, E T >10 °C, E V >18 °C, E R >18 °C, E M >10 °C It is an effective trough, otherwise it is considered as an invalid trough and removed; Statistical effective trough number N in time series scattering, evaluate the maximum rice cropping system S that the target area can carry, the rules are as follows: For the target area, when N>S, it is determined that there is overestimation of rice cropping system; for each effective trough, calculate the sum Sum of the average temperature of each phenological period: Sum = E S + E T + E V + E R + E M According to the order of Sum value from low to high, remove the effective troughs in turn, the number of effective troughs to be removed is N-S, and the remaining troughs are the rice troughs; Step 4. Rice cropping system area identification; Obtain the digital elevation data of the target area, extract the elevation and slope; obtain the land use product, extract the cultivated land distribution data; count the number of rice troughs in the region, and realize the identification of rice cropping system.

2. The method for rice maturity recognition based on radar time series observation and temperature analysis according to claim 1, characterized in that: In step 4, the threshold condition for extracting elevation and slope is to retain only the cultivated land with elevation below 1000m and slope less than 5°.

Images