Temperature discharge water reference temperature self-adaptive extraction method based on remote sensing image

Abstract

The invention relates to the technical field of ocean remote sensing monitoring, in particular to a remote sensing image-based warm discharge water reference temperature self-adaptive extraction method, which specifically comprises the steps of S01, data acquisition and preprocessing, S02, cumulative histogram generation, S03, curve fitting and smoothing, S04, second derivative calculation and S05, reference temperature determination. According to the method, the reference temperature is determined by analyzing the statistical distribution characteristics of the sea surface temperature data field, namely the form of the accumulative histogram, the physical demarcation point of the background water body and the warm drainage area on the temperature distribution is automatically positioned by using the second derivative, the process is completely driven by data, the interference of subjective factors is eliminated, and the accuracy of temperature distribution is improved. The method can automatically adapt to water areas with different geographical forms and different warm drainage diffusion modes, and objectivity and universality of reference temperature extraction are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ocean remote sensing monitoring, and particularly relates to a benchmark temperature adaptive extraction method for warm water discharge based on remote sensing images. BACKGROUND

[0002] The temperature rise range and intensity distribution of warm water discharge are important contents of warm water discharge monitoring. At present, the main observation methods include on-site multi-point measurement, numerical simulation and thermal infrared remote sensing. The thermal infrared remote sensing satellite monitoring technology with the ability of wide-range ground temperature observation in day and night and long time series has been widely applied to the monitoring of the temperature rise range and intensity of warm water discharge.

[0003] The extraction of benchmark temperature, also known as background temperature (i.e. the natural sea surface temperature not affected by warm water discharge), is the basic work for obtaining the temperature rise range and intensity of warm water discharge. The most commonly used methods for remote sensing extraction of the benchmark temperature of warm water discharge at present include the bay average temperature method, the discrete multi-point average method, the adjacent area substitution method and the temperature gradient method. The bay average temperature method is simple to calculate, but is only applicable to semi-enclosed sea areas with small sea surface temperature difference; the discrete multi-point average method needs to arrange points with the discharge area as the boundary reference line. However, in actual application, the determination of the boundary of the discharge area and the selection of the distribution position of the points have great uncertainty; the adjacent area substitution method needs a large amount of historical SST data obtained based on remote sensing, and the temperature value of the selected area will also be different due to the factors such as warm water discharge, wind, rainfall and season. The threshold value of the temperature gradient method is mostly determined by experience, and lacks a unified and objective selection standard. In addition, due to the difference in geographical environment, different benchmark temperature extraction methods need to be used when carrying out warm water discharge monitoring in different regions.

[0004] Therefore, the existing methods have the problems of limited regional applicability, dependence on manual intervention and strong subjectivity, which leads to the problem that the benchmark temperature cannot be extracted automatically in batches. There is an urgent need for a new method that can overcome the above defects and automatically and adaptively extract the benchmark temperature of warm water discharge. SUMMARY

[0005] The present application aims to provide a method for automatically extracting the benchmark temperature of warm water discharge which can be applied to semi-enclosed and open sea areas at the same time without manual setting of regions and threshold values.

[0006] To achieve the above-mentioned purpose, the present application adopts the following technical solutions: A benchmark temperature adaptive extraction method for warm water discharge based on remote sensing images, comprising: S1: data acquisition and preprocessing: acquiring sea surface temperature remote sensing image data within a preset range around the warm water discharge outlet, respectively performing resampling, land and sea farming area mask, DN value conversion and preprocessing of removing cloud and shadow covered pixels, ensuring that only valid pixels are retained in the image to form a data set; S2: cumulative histogram generation: traverse all valid sea surface temperature pixels in the dataset, count the number of pixels of different temperature values, and construct a frequency distribution histogram of sea surface temperature; based on the frequency distribution histogram, start from the lowest temperature value, gradually accumulate the proportion of the number of pixels in each temperature interval to the total number of valid pixels, and generate a cumulative histogram curve representing the cumulative distribution characteristics of the pixel temperature value; S3: curve fitting and smoothing: taking the peak point of the histogram in S2 as the reference, taking SST max as the threshold, taking SST max -1 and SST max +2 as the fitting interval range, and using a preset nonlinear function model to fit the cumulative histogram curve, a smooth fitting curve is obtained which can accurately represent the overall trend and is continuous and derivable; S4: second derivative calculation: derivative operation is performed on the smooth fitting curve to calculate the second derivative in the entire temperature definition domain, generating a second derivative curve which reflects the change of the curvature of the original cumulative histogram curve; S5: reference temperature determination: searching and positioning the global or local minimum value point on the second derivative curve, and determining the abscissa temperature value corresponding to the minimum value point as the reference temperature of the current remote sensing image; the minimum value point corresponds to the inflection point in the physical sense where the cumulative histogram curve changes from rapid growth to slow growth, which distinguishes the background water temperature region and the warm water discharge affected region.

[0007] Optionally, the cumulative histogram curve in the fitting interval is nonlinearly fitted by using a generalized logistic function, and the second derivative of the fitting curve is calculated.

[0008] Compared with the prior art, the present application has the following advantages: (1) strong universality: the present application determines the reference temperature by analyzing the internal structural characteristics of the temperature distribution, which exists in both open and semi-closed sea areas, so the method can be applied to different types of sea areas.

[0009] (2) objective and automatic: the whole process does not need manual reference area or experience threshold setting, avoiding subjective factors and easy to realize automatic processing of business system.

[0010] (3) good adaptability: the method does not depend on absolute temperature value, but finds the mathematical feature point of temperature distribution curve, so it can adapt to different tidal states, different seasons and different geographical locations of reference temperature extraction.

[0011] (4) Accuracy and reliability: Through verification with multiple images under different tidal conditions and geographical locations, the reference temperature extracted by this method is only slightly different from the result of the method recommended in the "Technical Specification for Satellite Remote Sensing Monitoring of Temperature Drainage of Coastal Nuclear Power Plant (Trial)" (absolute difference ≤ 0.15℃), which proves its accuracy and reliability. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the adaptive extraction method for temperature reference temperature of thermal drainage based on remote sensing imagery in this application.

[0013] Figure 2 This is an SST remote sensing image of a semi-enclosed marine nuclear power plant in an embodiment of this application.

[0014] Figure 3 This is an SST remote sensing image of an open-sea nuclear power plant in an embodiment of this application.

[0015] Figure 4 for Figure 2 A schematic diagram of the SST cumulative histogram and its generalized logic function fitting curve for the corresponding region.

[0016] Figure 5 for Figure 3 A schematic diagram of the SST cumulative histogram and its generalized logic function fitting curve for the corresponding region.

[0017] Figure 6 for Figure 4 A schematic diagram of the second derivative curve of the corresponding fitted curve.

[0018] Figure 7 for Figure 5 A schematic diagram of the second derivative curve of the corresponding fitted curve. Detailed Implementation

[0019] The technical solution of this application will be described in detail below with reference to specific embodiments. It should be noted that these embodiments are only used to illustrate this application and are not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions or improvements made by those skilled in the art to the technical solution of this application without departing from the spirit and substance of this application should be included within the scope of protection of this application.

[0020] like Figure 1 As shown in the figure, this application discloses an adaptive extraction method for thermal drainage reference temperature based on remote sensing images, which includes the following steps.

[0021] S01: Data Acquisition and Preprocessing: Acquire data within a preset range (no less than 100km) around the thermal discharge outlet. 2The target water area SST (Sea Surface Temperature) remote sensing image data within the range is covered, and the remote sensing image data is preprocessed, at least including resampling, land (including tidal flats) and sea surface aquaculture area mask, DN value conversion, removing cloud and shadow covered pixels, etc., to generate a data set containing effective sea surface temperature pixel points, to ensure that only effective pixels are retained in the image.

[0022] Specifically, the Landsat 9 Collection 2 Level-2 scientific data product published by the United States Geological Survey (USGS) can be selected, which contains SR (Surface Reflectance, multi-spectral surface reflectance) and ST (Surface Temperature, thermal infrared band surface temperature) data. The reason for selecting this data source is that it provides ST data that has been processed by an official algorithm and has undergone preliminary atmospheric correction, with guaranteed data quality, and is provided free of charge worldwide, with good accessibility. The spatial resolution of the ST data is 100 meters, and the revisit period is 16 days. When obtaining data, the image with no cloud or less cloud coverage is preferentially selected by querying the metadata, for example, the cloud coverage of the research area is less than 10%, to minimize the impact of clouds on the integrity of the SST data. The image data format is usually GeoTIFF, and the pixel value is stored in 16-bit unsigned integer (DN value).

[0023] In order to accurately extract the SST of the water area from the image, water-land separation needs to be performed, and the MNDWI (Modified Normalized Difference Water Index) is used to achieve this in the embodiment. The calculation formula of MNDWI is: MNDWI=(Green-SWIR2) / (Green+SWIR2) wherein Green corresponds to the surface reflectance of band 3 (0.53-0.59 μm, i.e. green band) of Landsat9 image, and SWIR2 corresponds to the surface reflectance of band 7 (2.11-2.29 μm, i.e. short wave infrared 2 band). After calculation, a threshold is set, for example, MNDWI>0, and the pixel points greater than the threshold are determined as water body, and the rest are land or other ground objects. This step can generate a binary water mask. Considering that tidal changes will cause dynamic changes in the water-land boundary, a fixed coastline vector file drawn at the lowest tide level is further used as the final research area boundary in the embodiment. The water mask generated by the above MNDWI is intersected with the fixed coastline vector file to obtain a research area mask for extracting SST, thereby effectively excluding the interference of other heat source areas such as intertidal bare beaches.

[0024] Understandably, the DN value of a Landsat 9 Level-2 ST product needs to be converted to a temperature value in degrees Celsius using a specific linear transformation formula. This formula is: ST = DN * s + h + k, where ST is the Kelvin temperature, DN is the DN value of the Landsat 9 Level-2 ST product, s is the scaling factor (0.00341802 for ST products), h is the offset (149), and k is the temperature at 0 K (-273.15°C). These parameters are provided by the USGS in the product metadata file. After this linear transformation, each water pixel in the image has a floating-point value representing its surface temperature in degrees Celsius.

[0025] S2: Cumulative Histogram Generation: Based on the effective pixels of the SST image obtained in S1, a 15km × 15km area is taken outward from the drainage outlet as the center, such as... Figure 2 and 3 As shown, using 0.1℃ as the bin window, all valid sea surface temperature pixels in the dataset are traversed, and the number of pixels at different temperature values ​​is counted to construct a frequency distribution histogram of sea surface temperature. Starting from the lowest temperature value, the proportion of pixels in each temperature range to the total number of valid pixels is accumulated step by step to generate a cumulative histogram curve representing the cumulative distribution characteristics of pixel temperature values.

[0026] Based on the frequency distribution histogram described above, the cumulative distribution is calculated. Starting from the statistical interval with the lowest temperature, the ratio of the number of pixels in each statistical interval to the total number of effective pixels is accumulated. Let the total number of effective pixels be Ntotal, and the number of pixels in the i-th temperature statistical interval be ni. Then, the cumulative ratio P(Tk) corresponding to the upper boundary temperature Tk of the k-th statistical interval is: P(Tk) = (Σ{i=1 to k}ni) / Ntotal. Performing this calculation on all temperature statistical intervals yields a series of discrete data points (Tk, P(Tk)). Connecting these data points forms the cumulative histogram curve of SST. This cumulative histogram curve is a monotonically increasing S-shaped curve, with the horizontal axis representing the temperature value and the vertical axis representing the cumulative ratio of pixels with temperatures lower than or equal to that temperature value.

[0027] Spatially, thermal discharge from nuclear power plants can be divided into two main parts: the temperature rise zone and the non-temperature rise zone. The temperature rise zone is characterized by high temperatures, a small spatial distribution area, a large temperature range, and significant spatial variations. The non-temperature rise zone, in contrast, is characterized by relatively low temperatures, a large spatial distribution area, a relatively small temperature range, and relatively gentle spatial variations. These characteristics are specifically reflected in the SST cumulative histogram as follows: the temperature rise zone is mainly distributed in areas with narrow cumulative ratio intervals and large temperature distribution ranges (e.g.,...). Figure 4 and 5The area within the box above the dividing point has a smaller slope for the cumulative ratio with temperature, and the slope change is more significant; while the non-temperature rise region is mainly distributed in areas with a larger cumulative ratio range and a smaller temperature distribution span (such as...). Figure 4 and 5 The cumulative ratio (the portion within the box below the dividing point) exhibits a relatively stable trend with a large change (slope) in temperature. Therefore, the method of this invention identifies a boundary temperature threshold when transitioning from a non-temperature-rising region to a temperature-rising region. Regions exceeding this threshold can be defined as temperature-rising regions, and this threshold is the reference temperature.

[0028] S3: Curve Fitting and Smoothing: Using the histogram peaks in S2 to correspond to SST max Based on the SST value threshold [SST] max -1, SST max +2] represents the fitting interval range. A preset nonlinear function model is used to fit the cumulative histogram curve to obtain a smooth fitting curve that can accurately characterize the continuous and differentiable overall trend of change.

[0029] It is understandable that the S-shaped characteristic of the SST cumulative histogram closely matches the curve shape of the generalized logic function. This embodiment uses the generalized logic function, whose mathematical expression is: F(x) = f min +(f max -f min ) / (1+exp(-k*(xg))), where x is the independent variable, representing the SST temperature value, F(x) is the dependent variable, representing the cumulative ratio corresponding to temperature x, and f min and f max These are the lower and upper limits of the curve, which physically correspond to the minimum and maximum values ​​of the cumulative ratio. In this application, f min Theoretically, it is 0, f max Theoretically, it is 1 (or 100%), g is the x-value corresponding to the inflection point of the curve (the point of maximum growth rate), i.e., the SST median point, and k is the growth rate of the curve, which determines the steepness of the S-curve. Before fitting, the temperature range for fitting needs to be reasonably selected. Theoretically, the temperature value at the peak should be slightly lower than the boundary temperature. At the same time, the temperature rise intensity classification and mapping specifications should be standardized, covering areas above the boundary temperature by more than 1°C. Therefore, this invention uses the SST temperature value of the histogram peak center point as the benchmark, and extends it by 1°C and 2°C towards the low temperature and high temperature values, respectively, as the statistical analysis temperature range of the invention method, thereby ensuring that the subsequent temperature rise statistics are greater than 1°C, and ensuring that the boundary temperature extracted in the next step of curve fitting analysis is within this range. The SST corresponding to the peak point of the histogram in S2 is statistically obtained. max Fuqing Nuclear Power Plant SST max The temperature was 11.0℃ at the Fangchenggang Nuclear Power Plant SST. maxis 20.8℃, and the fitting temperature interval is [SST max -1, SST max +2], thus [12.17℃, 15.17℃] (Fuqing nuclear power plant) and [21.8℃, 24.8℃] (Fangchenggang nuclear power plant) are obtained.

[0030] S04: Second derivative calculation: the second derivative of the smoothed fitting curve is calculated, and a second derivative curve is generated, which reflects the change of the curvature of the original cumulative histogram curve. The function can be expressed as: F"(x)=(f max -f min )*k^2*e^(-k(x-g) )*((e^(-k(x-g) )-1) / (1+e^(-k(x-g) ) )^3.

[0031] It can be understood that, as shown in FIGS. Figure 6 and 7 , near the histogram peak, the fitting curve is downward convex, so the second derivative is negative, and the minimum value of the second derivative exists in this area, representing the most convex curve at this point. On both sides of the histogram, the fitting curve tends to be flat, and the convexity and concavity decrease, and the second derivative tends to zero. The cumulative histogram fitting curve shows the characteristics of small slope and sudden change (inflection point) of slope change at the boundary temperature position. That is, when the temperature moves to the high temperature area, the temperature slope and its change of the fitting curve both show a trend of first increasing and then decreasing, and when the temperature reaches the boundary temperature, the slope change reaches the minimum. When the temperature crosses the boundary temperature and enters the temperature rise area, the temperature slope continues to gradually decrease, but as it gets closer to the warm water center, the temperature changes more sharply, showing a gradually increasing trend of temperature slope change.

[0032] S05: Reference temperature determination: search and locate the global or local minimum value point on the second derivative curve, and determine the abscissa temperature value corresponding to the minimum value point as the reference temperature of the current remote sensing image, obtaining the reference temperatures 13.18℃ and 23℃. The minimum value point of the second derivative is the point with the fastest change of curvature, and the cumulative histogram fitting curve shows a small slope and a sudden change point (inflection point) of slope change at the boundary temperature position. The temperature value corresponding to this point can be objectively and stably defined as the transition point from the reference temperature to the temperature rise area.

[0033] Those skilled in the art can understand that, whether in winter (overall low temperature) or summer (overall high temperature), whether in open sea (baseline temperature distribution is concentrated, peak shape is sharp) or semi-closed sea (baseline temperature distribution is wide, peak shape is flat), the SST histogram will present a bell-shaped structure with high in the middle and gradually decreasing on both sides. Taking the temperature corresponding to the minimum point of the second derivative of the fitting curve of the cumulative histogram as the baseline temperature, it has sufficient technical rationality and internal logic, and objectively identifies the statistical demarcation point between the background water body and the water body affected by the warm water discharge. The method can always find the structural feature point (i.e. the point with the most drastic change in curvature) of the distribution curve in this structure, rather than an absolute temperature value or a region that needs to be manually set, and therefore has adaptability and robustness to different geographical environments and seasons.

[0034] The above are preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in the specification (including the abstract and drawings) can be replaced by other equivalent or similar features unless specifically described. That is, each feature is only an example of a series of equivalent or similar features unless specifically described.

Claims

1. A method for self-adapting extraction of reference temperature of warm water discharge based on remote sensing image, characterized in that, Comprise: S1: data acquisition and preprocessing: obtain the sea surface temperature remote sensing image data in the preset range around the warm water outlet, respectively, resample, land and sea farming area mask, DN value conversion, and remove cloud and shadow covered pixel preprocessing, ensure that only valid pixels are retained in the image, form a data set; S2: cumulative histogram generation: traverse all valid sea surface temperature pixel points in the data set, count the number of pixel points of different temperature values, and construct a frequency distribution histogram of sea surface temperature; Based on the frequency distribution histogram, from the lowest temperature value, the proportion of the number of pixel points in each temperature interval to the total number of valid pixel points is accumulated step by step, and a cumulative histogram curve representing the cumulative distribution characteristics of pixel temperature value is generated; S3: Curve fitting and smoothing: SST corresponding to the peak point of the histogram in S2 max Take the SST value threshold [SST max -1, SST max +2] as the fitting interval range, and use a preset nonlinear function model to fit the cumulative histogram curve to obtain a smooth fitting curve that can accurately represent the overall trend of the curve, which is continuous and derivable. S4: second derivative calculation: derivative operation is performed on the smoothed fitting curve, and the second derivative of the whole temperature definition domain is calculated, a second derivative curve is generated, and the second derivative curve reflects the change of the curvature of the original cumulative histogram curve; S5: reference temperature determination: searching and positioning the global or local minimum value point on the second derivative curve, determining the abscissa temperature value corresponding to the minimum value point as the reference temperature of the warm water discharge corresponding to the current remote sensing image; The minimum value point corresponds to the inflection point of the cumulative histogram curve in physical meaning, which is the turning point from rapid growth to slow growth, which distinguishes the background water temperature area and the warm water discharge influence area.

2. The method of claim 1, wherein, The cumulative histogram curve is fitted by using a preset nonlinear function model, specifically: using a generalized logistic function formula to perform nonlinear fitting on the cumulative histogram curve in the fitting interval, and calculating the second derivative of the fitting curve.

Images