阅读说明：本技术 基于gee的大区域湖泊透明度快速制图方法 (GEE-based large-area lake transparency rapid drawing method ) 是由 刘�东 段洪涛 于 2019-11-27 设计创作，主要内容包括：本发明涉及基于GEE的大区域湖泊透明度快速制图方法,采集不同湖区湖泊的透明度实测数据；从GEE计算云平台获取透明度实测数据对应的遥感数据,包括遥感反射率及备注信息数据,并从中筛选出有效匹配数据；基于透明度实测数据和有效匹配数据,建立透明度遥感模型,并应用于GEE全球遥感产品实现不同时空尺度的透明度快速制图。本发明结合遥感大数据和GEE计算机云平台,通过分析同步的现场透明度观测和陆地遥感反射率,构建了适用于陆地遥感反射率的大区域透明度反演模型。可以实现大区域、不同时间尺度湖泊透明度的快速制图。为大区域湖泊水质评价和综合对比提供数据支撑。(The invention relates to a GEE-based method for quickly drawing transparency of large-area lakes, which is used for collecting actually measured transparency data of lakes in different lake areas; obtaining remote sensing data corresponding to transparency measured data from the GEE computing cloud platform, wherein the remote sensing data comprises remote sensing reflectivity and remark information data, and screening effective matching data from the remote sensing reflectivity and remark information data; and establishing a transparency remote sensing model based on the transparency measured data and the effective matching data, and applying the transparency remote sensing model to GEE global remote sensing products to realize rapid transparency mapping of different space-time scales. The method combines remote sensing big data and a GEE computer cloud platform, and constructs a large-area transparency inversion model suitable for the land remote sensing reflectivity by analyzing synchronous field transparency observation and land remote sensing reflectivity. The method can realize the rapid drawing of the lake transparency in large areas and different time scales. And data support is provided for water quality evaluation and comprehensive comparison of large-area lakes.)

1. A GEE-based method for quickly drawing transparency of a large-area lake is characterized by comprising the following steps:

(1) collecting actually measured transparency data of lakes in different lake regions;

(2) obtaining remote sensing data corresponding to transparency measured data from the GEE computing cloud platform, wherein the remote sensing data comprises remote sensing reflectivity and remark information data, and screening effective matching data;

(3) establishing a transparency remote sensing model based on the transparency measured data and the effective matching data, wherein the transparency remote sensing model comprises the following steps:

a) randomly selecting actually-measured transparency and matched MODIS remote sensing reflectivity as a model construction data set, analyzing the linear correlation between the remote sensing reflectivity of different wave bands and transparency actual measurement data, and selecting green light and red light wave bands with good linear correlation;

b) testing the correlation between different combination forms of green light and red light wave bands and actually measured transparency, screening a model, and carrying out parameterization by using a model construction data set to construct a transparency remote sensing inversion model;

(4) the model is applied to global reflectivity stored in GEE to realize rapid drawing of different space-time scale transparencies in a large area.

2. The method according to claim 1, wherein in step (1), the transparency measurement data is obtained based on a black and white Secchi disc measurement of 30cm diameter.

3. The method of claim 1, wherein in step (2), the remote sensing data is directly selected from MODIS land reflectivity data stored by GEE.

4. The method according to claim 3, wherein in the step (2), the remote sensing data is acquired by: and programming and acquiring remote sensing reflectivity and remark information data corresponding to each transparency measured data on the GEE computing cloud platform by taking +/-3 h as a time window and 3 pixels as a space window.

5. The method according to claim 3, wherein in the step (2), the remark information data comprises quality evaluation data, and valid matching data is quickly screened out based on the quality evaluation data.

6. The method according to claim 5, wherein the quality evaluation information is represented in binary, and comprises cloud coverage, mountain shadow, land and aerosol condition information of the extracted pixels, and the remote sensing reflectivity data which is matched with the transparency measured data and has small influence on non-cloud coverage, non-mountain shadow, non-land and aerosol is screened out based on the quality evaluation information, and effective matching data is rapidly screened out.

7. The method according to claim 3, wherein in the step (3), a transparency-sensitive wavelength band-555 nm green light wavelength band and 645nm red light wavelength band are selected to construct a transparency remote sensing inversion model.

8. The method according to claim 3, wherein in the step (3), the transparency remote sensing inversion model constructed by the optimal band combination and segmentation method is as follows:

[SDD]＝a×x^b,x＞T

[SDD]＝c×exp(d×x),x≤T

x＝(R555+R645)/2 (1)

wherein [ SDD ]]Is the transparency of the lake picture element; r555 and R645 are land remote sensing reflectivity (sr) of MODIS data in 555nm and 645nm wave bands respectively^-1) (ii) a T is the median separation threshold; a, b, c and d are fitting parameters.

9. The method of claim 1, wherein the step (3) further comprises:

and carrying out precision verification on the constructed model by using the residual transparency measured data and the matched remote sensing reflectivity.

10. The method according to claim 1, wherein in the step (4), the constructed model is applied to GEE-stored global-length-of-day MODIS remote-sensing reflectivity so as to realize national area>20km²And quickly drawing the transparency of the lake at different space-time scales.

Technical Field

The invention relates to the field of satellite remote sensing technology and application thereof, in particular to a GEE-based method for quickly mapping transparency of a large-area lake.

Background

The transparency (SDD) is an important index for monitoring lakes, and can comprehensively reflect the water quality conditions of lakes. The reduction of the transparency of the lake is not beneficial to the penetration of light through a water column, is not beneficial to the growth of phytoplankton and aquatic vegetation in the lake, and is easy to cause the deterioration of the ecological environment of the lake. In addition, under the influence of severe human activities, the eutrophication condition of lakes becomes increasingly serious, and phytoplankton is increased in an outbreak manner; the phytoplankton can reduce the transparency of the lake water body, and the strong light absorption substance (CDOM) generated by the decomposition of the phytoplankton can also reduce the transparency of the lake, so that the transparency is also an important index for evaluating the eutrophication condition of the lake. Through the development of scientific technology in recent decades, the lake transparency has been developed from the traditional on-site point observation to the time series monitoring of whole lakes and even regions. But is limited by large regional space difference and computer processing capability, and national, intercontinental or global lake transparency remote sensing synchronous satellite remote sensing monitoring algorithms are not reported for a while.

Disclosure of Invention

The invention aims to provide a GEE-based method for quickly drawing transparency of a large-area lake, so as to quickly realize the monitoring of the transparency of the large-area lake in different time scales.

In order to achieve the technical purpose, the remote sensing big data are analyzed from the large-area field transparency measured data and the GEE computer cloud platform, and reflectivity values of different wave bands of the remote sensing data corresponding to the field transparency sampling data time are extracted. And then, analyzing the synchronous remote sensing reflectivity value, eliminating data greatly influenced by cloud, mountain shadow, land, aerosol and the like, analyzing the transparency and the remote sensing reflectivity of synchronous observation, and selecting a remote sensing wave band or a wave band combination suitable for reflecting transparency change. And finally, constructing a remote sensing inversion model of the lake water body based on the change mechanism of the transparency of the lake water body, and verifying the constructed model.

The specific technical scheme is as follows:

a GEE-based method for quickly drawing transparency of a large-area lake comprises the following steps:

(1) collecting actually measured transparency data of lakes in different lake regions;

(2) obtaining remote sensing data corresponding to transparency measured data from the GEE computing cloud platform, wherein the remote sensing data comprises remote sensing reflectivity and remark information data, and screening effective matching data;

(3) establishing a transparency remote sensing model based on the transparency measured data and the effective matching data, wherein the transparency remote sensing model comprises the following steps:

a) randomly selecting actually-measured transparency and the matched MODIS remote sensing reflectivity as a model construction data set, analyzing the linear correlation between the remote sensing reflectivity of each waveband and transparency actual measurement data, and selecting green light and red light wavebands with good linear correlation;

b) testing the correlation between different combination forms of green light and red light wave bands and actually measured transparency, screening a model, constructing data set parameterization by using the model, and constructing a transparency remote sensing inversion model;

(4) the model is applied to global reflectivity stored in GEE to realize rapid drawing of different space-time scale transparencies in a large area.

As a further improvement of the invention, in the step (1), transparency measurement data is obtained based on a black and white Secchi disc measurement with a diameter of 30 cm.

As a further improvement of the invention, in the step (2), the remote sensing data directly selects and uses MODIS land reflectivity data stored by GEE.

Further, the remote sensing data acquisition mode is as follows: and programming and acquiring remote sensing reflectivity and remark information data corresponding to each transparency measured data on the GEE computing cloud platform by taking +/-3 h as a time window and 3 pixels as a space window.

Further, in the step (2), the remark information data includes quality evaluation data, and the valid matching data is screened based on the quality evaluation data. The quality evaluation information is represented in a binary system and comprises cloud coverage information, mountain shadow information and land and aerosol state information of the extracted pixel, non-cloud coverage data, non-mountain shadow data, non-land and aerosol data which are matched with the transparency measured data are screened out based on the quality evaluation information, and effective matching data are obtained.

As a further improvement of the invention, in the step (3), a transparency sensitive wave band-555 nm green light wave band and 645nm red light wave band are selected to construct a transparency remote sensing inversion model.

Further, in the step (3), the transparency remote sensing inversion model constructed by the optimal band combination and segmentation method is as follows:

[SDD]＝a×x^b,x＞T

[SDD]＝c×exp(d×x),x≤T

x＝(R555+R645)/2 (1)

wherein [ SDD ]]Is the transparency of the lake picture element; r555 and R645 are land remote sensing reflectivity (sr) of MODIS data in 555nm and 645nm wave bands respectively^-1) (ii) a T is the median separation threshold; a, b, c and d are fitting parameters.

Further, the step (3) further comprises:

and carrying out precision verification on the constructed model by using the residual transparency measured data and the matched remote sensing reflectivity.

As a further improvement of the present invention, the method further comprises: the constructed model is applied to the global scale of days MODIS remote sensing reflectivity stored in GEE, so that the national area is realized>20km²And quickly drawing the transparency of the lake at different space-time scales.

The method is characterized in that a large-area transparency inversion model suitable for the terrestrial remote sensing reflectivity is constructed by combining remote sensing large data and a GEE computer cloud platform and analyzing synchronous field transparency observation and the terrestrial remote sensing reflectivity. The method can realize the rapid drawing of the lake transparency in large areas and different time scales. And data support is provided for water quality evaluation and comprehensive comparison of large-area lakes.

Drawings

FIG. 1 is a remote sensing inversion model of lake transparency.

FIG. 2 is a remote sensing inversion accuracy evaluation of lake transparency.

Detailed Description

The invention is further explained by taking the rapid drawing process of the transparency of large and medium lakes in different areas of China as an example and combining the description of the attached drawings and the embodiment:

(1) representing the on-site data acquisition of lakes in different lake regions throughout the country. Chinese breadth, large difference of different lake material compositions, and the phenomenon of foreign matter common spectrum in lake water body remote sensing, so that the previous water color remote sensing research takes some lakes in individual or region as research objects. In order to construct a remote sensing algorithm suitable for nationwide lakes, 2233 sampling point bit data covering 298 lakes in different lake areas of the country are researched and used, the sampling number of each lake is 3-20 and is different, and the sampling time is 2008 + 2018. Following the NASA published specifications for optical measurements of bodies of water, the field transparency measurement used a 30cm diameter black and white Secchi disc.

(2) And extracting the land remote sensing reflectivity synchronously observed on site. The GEE computing cloud platform integrates various remote sensing reflectivity data. For Chinese regional transparency remote sensing inversion, MODIS remote sensing reflectivity is recommended, the data set can cover the world once every day from 2 months to 24 days in 2000, and the method has unique advantages for researching long-time sequence change of lake transparency. According to the field observation date and time of national lake transparency data, a GEE background database and a strong geographical computing cloud are utilized to obtain MODIS remote sensing reflectivity, quality evaluation and other remark information corresponding to each field observation transparency in a programming mode. And analyzing remote sensing reflectance values of MODIS satellite synchronous observation, and eliminating data values influenced by cloud, mountain shadow, land, aerosol and the like. The method comprises the following specific steps:

a) and (3) obtaining MODIS reflectivity and quality evaluation information corresponding to each measured data by autonomous programming on the GEE platform by using +/-3 h as a time window and 3 pixels as a space window according to a satellite-ground matching principle recommended by NASA.

b) The quality evaluation information of the MODIS is represented in a binary system, the quality evaluation information comprises information such as cloud coverage, mountain shadow and land and aerosol conditions of the extracted pixels, the MODIS remote sensing reflectivity which is matched with actually measured data and has small influences on non-cloud coverage, non-mountain shadow, non-land and aerosol can be screened out according to the information, and finally the obtained effective matching number is 489.

(3) And (5) constructing and evaluating a transparency remote sensing model. The transparency of the lake water body is the result of the interaction of various substances in the water; the increase of the suspended particles in the water body can increase the back scattering of the water body to light and reduce the transparency of the water body, thereby increasing the reflectivity of the satellite remote sensing; the increase of the CDOM substances of the water body can increase the light absorption of the water body, reduce the transparency of the water body and reduce the reflectivity of the satellite remote sensing. The strong scattering of suspended particles is mainly shown in the long wave band, and the strong absorption of CDOM is mainly shown in the short wave band. According to some regional lake researches, the transparency is usually well correlated with the remote sensing reflectivity of short-wave green light or long-wave red light wave bands. Based on the research foundations, a transparency remote sensing algorithm suitable for Chinese lakes is constructed, and the method comprises the following specific steps:

a) randomly selecting 75% of actually-measured transparency and matched MODIS remote sensing reflectivity (N & lt366 & gt) as a model construction data set, analyzing linear correlation between reflectivity of each wave band of MODIS and the actually-measured transparency through autonomous programming, and selecting green light and red light wave bands with good linear correlation;

b) testing the correlation (linearity, exponent, power function and the like) between different combination forms (ratio, difference, sum and the like) of green light and red light wave bands and actually measured transparency through autonomous programming, determining a model with the best correlation, constructing data set parameterization by using the model, and constructing a remote sensing inversion model suitable for inverting the transparency of the Chinese lake, wherein the remote sensing inversion model is shown in a graph (1) and a formula (2):

wherein [ SDD ]]Is lake pixel transparency (cm); r555 and R645 are MODIS remote sensing reflectivity (sr) of MODIS data in 555nm and 645nm wave bands^-1)。

c) And (3) carrying out precision verification on the constructed model formula (2) by using the remaining 25% of field observation transparency and the matched MODIS remote sensing reflectivity (N is 123): the relative error of the model construction data set is 32.42 percent, the relative error of the model inspection data set is 32.72 percent, and the accuracy requirement (<45 percent) of the international universal remote sensing inversion is met, as shown in a figure (2).

(4) And (5) rapidly drawing the transparency of different time scales. The constructed model is programmed autonomously by means of a GEE geographic computing cloud platform and online data access functionsThe model is applied to the global daily scale MODIS remote sensing reflectivity stored in GEE, thereby synchronously calculating the area>20km²The transparency of the lakes of 412 Chinese lakes at different time scales (yearly average, monthly average and climatic month), and the final transparency result is derived for normalizing the quick drawing.