CN114707412B

CN114707412B - SWAT model optimization method based on vegetation canopy time-varying characteristics

Info

Publication number: CN114707412B
Application number: CN202210365650.0A
Authority: CN
Inventors: 刘铁刚; 郝伟罡; 刘超; 金中武
Original assignee: Sichuan University
Current assignee: Sichuan University
Priority date: 2022-04-08
Filing date: 2022-04-08
Publication date: 2023-09-29
Anticipated expiration: 2042-04-08
Also published as: CN114707412A

Abstract

The application provides a SWAT model optimization method based on vegetation canopy time-varying characteristics, which comprises the following steps: obtaining MODIS LAI data and remote sensing data of a target area year by year; converting the MODIS LAI data into time and space resolution, and correcting according to the actually measured pattern LAI data of the target area to obtain target area corrected MODIS LAI data; the target area correction MODIS LAI data is used for replacing a LAI simulation calculation module in the SWAT model and is used as the input of the model; and correcting MODIS LAI data by using the target area to extract the waiting period data, and using the waiting period data to replace a threshold daily length definition sleep period mode in a SWAT model as an input of the model. According to the application, the vegetation type, the leaf area index and the physical weather period are taken as the vegetation canopy feature factors, the multi-parameter dynamic expression of the remote sensing canopy feature factors is realized on the basis of the SWAT model, the time-varying feature of the vegetation canopy is considered, and the capability of the model for describing the time-varying feature of the vegetation canopy is improved.

Description

SWAT model optimization method based on vegetation canopy time-varying characteristics

Technical Field

The application relates to the field of hydrology and climate monitoring, in particular to a SWAT model optimization method based on time-varying characteristics of vegetation canopy.

Background

At present, the global climate change and human activity influence are increasingly aggravated, and hydrology is gradually focused on the research of the water circulation change rule under the changing environment. Underlying elements such as climate, vegetation, soil and topography have an impact on the watershed hydrologic process. Soil and topography can be considered to be "steady-state" while climate and vegetation exhibit a high degree of time variability. Studies have shown that large scale vegetation changes caused by climate change and human activity have altered land water circulation in many areas. The hydrologic response under vegetation changes is a hotspot problem in current hydrologic research.

Quantitative analysis methods of vegetation change effects on hydrologic processes can be divided into three categories: hydrologic model method, comparative watershed experimental method and analysis method based on Budyko et al theory. The hydrologic model can conduct qualitative and quantitative research on the hydrologic response process of vegetation change, can fully consider variability of parameters and variables, and is an effective means for analyzing the hydrologic response of vegetation change. The ecological hydrologic process related to vegetation in the model comprises canopy interception, transpiration evaporation, soil moisture and the like. The vegetation parameters are important input parameters in the hydrologic model, such as vegetation type, leaf area index, canopy height, root depth and the like. The leaf area index is the most important index for representing the vegetation canopy structure and growth state in a hydrological model, and is generally used for simulating the influence of vegetation change on canopy interception, transpiration evaporation, runoff, sediment and other processes in the hydrological model. Dynamic changes of vegetation canopy structures and hydrologic characteristics are detailed in the model, and the dynamic changes have an important effect on researching vegetation change hydrologic response.

SWAT (Soiland Water Assessment Tool) is a semi-distributed hydrologic model with a strong physical mechanism and has been widely used for ecological hydrologic problem research in varying environments. In studying the hydrographic response of vegetation changes, there are mainly several key issues. First, the plant LAI is simulated in the SWAT model by a simplified crop growth model EPIC (Erosion Productivity Impact Calculator), which differs significantly from the telemetry data. Second, sleep period settings that characterize vegetation climate in the SWAT model also have an adaptability problem. The SWAT model assumes that the vegetation growth cycle is controlled by air temperature and solar length, resulting in a sleep period that is inconsistent with the actual situation. Again, land utilization change is a continuous process, but SWAT cannot reflect time-varying characteristics of different land utilization or vegetation types.

Current research is generally based on a hydrological model to analyze the effect of land use changes on the hydrologic process. However, vegetation changes include not only changes in land utilization or vegetation type, but also changes in canopy feature factors such as canopy structural features (e.g., leaf area index) and climatic features of the vegetation. These factors change in both the hydrologic process and the hydrologic parameters, and these three factors change in relation to each other. Therefore, how to describe the dynamic change of vegetation canopy structure in the hydrologic model framework is a problem that has to be solved to simulate the hydrologic response of vegetation change based on the hydrologic model.

Disclosure of Invention

In order to solve the technical problems, the application provides a SWAT model optimization method based on vegetation canopy time-varying characteristics, which takes vegetation types, leaf area indexes and weathers as vegetation canopy characteristic factors, realizes multi-parameter dynamic expression of remote sensing canopy characteristic factors under the frame of the SWAT model, and constructs a hydrological simulation model considering canopy structure parameter time-varying characteristics.

In order to achieve the above object, the present application provides the following technical solutions:

a SWAT model optimization method based on vegetation canopy time-varying features comprises the following steps:

s1, obtaining MODIS LAI data and remote sensing data of a target area year by year;

s2, converting the MODIS LAI data into time and space resolution, and correcting according to the actually measured sample area LAI data of the target area to obtain target area corrected MODIS LAI data;

s3, replacing the LAI simulation calculation module in the SWAT model with the MODIS LAI data corrected by the target area to be used as the input of the model;

and S4, correcting MODIS LAI data by using the target area to extract the waiting period data, and using the waiting period data to replace a threshold daily length definition dormancy period mode in the SWAT model to serve as input of the model.

In some preferred embodiments, the method for performing temporal and spatial resolution conversion on the MODIS LAI data in step S2 includes:

s201, resampling the preprocessed MODIS LAI data according to the spatial resolution of the remote sensing data;

s202, classifying vegetation of the remote sensing data; establishing the reflectivity quantity relation between the resampled MODIS LAI data and the remote sensing data according to the vegetation classification result;

s203, converting the spatial resolution of the MODIS LAI data into the same spatial resolution as the remote sensing data according to the reflectivity quantity relation;

and S204, processing the time resolution of the MODIS LAI data into 1 day by adopting a linear interpolation method.

In some preferred embodiments, the method for correcting the area observation LAI data according to the target area in step S2 includes:

s211, setting a plurality of block sample areas in the target area according to the spatial resolution of the converted MODIS LAI data, and carrying out actual measurement observation on the sample areas to obtain actual measurement sample area LAI data;

s212, training a MODIS-ESU LAI conversion model by adopting MODIS LAI data and actually measured sample area LAI data at the same time and at the same position;

s213, inputting the target area MODIS LAI data into a trained MODIS-ESU LAI conversion model to obtain target area correction MODIS LAI data.

In some preferred embodiments, step S3 further comprises:

s301, setting an LAI simulation calculation module in a SWAT model to be in an inactive state;

s302, calculating and assigning a target area correction MODIS LAI data average value of each hydrologic response unit, wherein the target area correction MODIS LAI data average value is used as input of a vegetation growth subprogram in the SWAT model.

In some preferred embodiments, step S4 further comprises:

s401, correcting MODIS LAI data extract candidate data comprising a starting point SOG and a terminal point EOG of vegetation growing period by using a multi-period weathered inversion method by using the target area;

s402, setting a threshold daily length definition sleep period mode in a SWAT model to be in an inactive state;

and S403, respectively calculating and assigning the average value of the start point SOG and the end point EOG of the vegetation period of each hydrological response unit as the input of a simulated biomass growth program in the SWAT model.

Advantageous effects

According to the application, the vegetation type, the leaf area index and the physical weather period are taken as the vegetation canopy feature factors, the multi-parameter dynamic expression of the remote sensing canopy feature factors is realized on the basis of the SWAT model, the time-varying feature of the vegetation canopy is considered, and the capability of the model for describing the time-varying feature of the vegetation canopy is improved.

Drawings

FIG. 1 is a schematic flow chart of a preferred embodiment of the present application;

FIG. 2 is a flow chart of performing time and spatial resolution conversion on MODIS LAI data according to a preferred embodiment of the present application;

FIG. 3 is a flow chart of the calibration of the LAI data according to the regional observation of the target region according to the preferred embodiment of the present application;

FIG. 4 is a flow chart of a preferred embodiment of the present application for implementing the target area correction MODIS LAI data instead of the LAI simulation calculation module in the SWAT model as the model input;

FIG. 5 is a flow chart of a preferred embodiment of the present application for implementing a sleep period mode defined by substituting the objective day length data for the threshold day length in the SWAT model as the model input;

FIG. 6 is a schematic diagram of an initial vegetation growth module for editing a SWAT model in accordance with a preferred embodiment of the present application;

FIG. 7 is a schematic diagram showing the configuration of a climatic information input module for editing SWAT models according to a preferred embodiment of the present application;

FIG. 8 is a graph showing the spatial distribution of leaf area indexes at different times in a preferred embodiment of the present application;

FIG. 9 is a schematic view showing spatial distribution of mean and rate of change of vegetation in the eversion period over time for multiple years according to a preferred embodiment of the present application;

Detailed Description

The present application will be further described with reference to the accompanying drawings, in order to make the objects, technical solutions and advantages of the present application more apparent. In the description of the present application, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "top," "bottom," "inner," "outer," and the like indicate or are based on the orientation or positional relationship shown in the drawings, merely to facilitate description of the present application and to simplify the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present application.

As shown in fig. 1, a method for optimizing a SWAT model based on time-varying characteristics of vegetation canopy comprises the steps of:

in the present application, the target area refers to the basin to be studied, which is affected by human activities and climate change.

The MODIS (Moderate-resolution Imaging Spectroradiometer, medium resolution imaging spectrometer) is a large space remote sensing instrument developed by the united states space navigation office, and according to the capability of direct broadcasting, the original MODIS data stream can be obtained freely through a real-time tracking antenna. In addition, some web sites on the internet provide processed data for public payment. LAI (Leaf Area Index) refers to the multiple of the total Area of plant leaves per unit Area of land. It is related to the density, structure (single layer or multiple layer), biological characteristics (branch angle, leaf angle, shade tolerance, etc.) and environmental conditions (illumination, moisture, soil nutrition status) of vegetation, and is a common comprehensive index for representing vegetation utilization light energy status and canopy structure.

The MODIS LAI data specifically comprises MODIS reflectivity data and LAI data.

It should be understood that, for the acquired image data, preprocessing steps such as filtering, noise reduction, clipping and the like are required. In some preferred embodiments, the MODIS LAI data may be processed using a S-G (Savitzky-Golay) smoothing filter technique to remove noise interference due to atmospheric conditions.

In some preferred embodiments, the telemetry data may be: landsat data (Landsat 5. TM., landsat7 ETM+, landsat8 OLi) with a temporal resolution of 1 day and a spatial resolution of 30m. It should be understood that, what kind of remote sensing data is specifically adopted by those skilled in the art does not image implementation of the technical scheme and implementation of the technical effects of the present application, and therefore, the present application is not intended to be further limited to specific remote sensing data types.

S2, converting the MODIS LAI data into time and space resolution, and correcting according to actually measured pattern (ESU) LAI data of a target area to obtain target area corrected MODIS LAI data;

it should be appreciated that the spatial resolution of the MODIS LAI data tends to be low, while the resolution of the remote sensing image is high, so that to enable the adaptation of both, a conversion of the MODIS LAI data into temporal and spatial resolution is required. In some preferred embodiments, specific steps are presented, as shown in FIG. 2, including:

s202, classifying vegetation of the remote sensing data; establishing the reflectivity quantity relation between the resampled MODIS LAI data and the remote sensing data according to the vegetation classification result; wherein, the classification can adopt an unsupervised classification method.

S203, converting the spatial resolution of the MODIS LAI data into the same spatial resolution as the remote sensing data according to the reflectivity quantity relation; preferably, a time-adaptive reflectance fusion model (STARFM) may be used in converting the spatial resolution of the MODIS LAI data.

Further, it will be appreciated by those skilled in the art that in existing SWAT models, the HRU (hydrographic response unit) is commonly defined by vegetation type, soil type and grade. The specific method for classifying vegetation of the remote sensing data can be as follows:

a) Reclassifying year-by-year vegetation type data in the remote sensing data, merging vegetation types with small area proportion and small change thereof into other similar types, and reducing the types of the vegetation types;

b) Spatially superposing year-by-year vegetation type raster data in a simulation period according to a year sequence, wherein a vegetation type map generated after superposition has attribute information such as vegetation patch numbers, vegetation patch types, years and the like;

c) And superposing the new vegetation type map with the soil type map and the gradient map, and calling vegetation type information of different years year by year in subsequent use to generate HRUs of all the years. It should be appreciated that the plaque locations, numbers and areas are the same in HRUs of different years at this time, but vegetation type properties may change.

In other preferred embodiments, as shown in fig. 3, the method for correcting the area observation LAI data according to the target area includes:

s211, setting a plurality of block sample areas in the target area according to the spatial resolution of the converted MODIS LAI data, and carrying out actual measurement observation on the sample areas to obtain actual measurement sample area LAI data; the arrangement is that ground observation data are considered to have reality, MODIS LAI data have space-time continuity, and the reliability of space-time distribution of the MODIS LAI data is enhanced by combining the ground observation data with the MODIS LAI data.

In the present application, applicant considers that the following disadvantages exist in terms of directly applying the existing SWAT model to the time-varying characterization of vegetation canopy structures: (1) the spatial distribution of vegetation types in the simulation process does not change with time, (2) the leaf area index simulation precision has larger uncertainty, and (3) the model dormancy period description is not matched with the actual physical characteristics of vegetation canopy. Therefore, consider optimizing the SWAT model. In the present application, an important aspect is to edit a vegetation growth module (growth. F) of the SWAT model, and its alternative schematic is shown in fig. 6; another aspect is a climatic information input module editing the SWAT model, an alternative illustration of which is shown in fig. 7.

since the SWAT model simulates the plant growth process by a simplified crop growth model EPIC (Erosion Productivity Impact Calculator), it assumes that only the average air temperature on the day (T _av ) Exceeds the basal temperature (T) _base ) Plants will grow at this time. Using cumulative latent heat unit fraction (fr _PHU ) The leaf area index change process was simulated and plant biomass change was calculated using the leaf area index according to the Montetth method. Such plant growth modules do not describe well the actual leaf area index variation process. Especially in areas where human activities such as returning and manual harvest are affected strongly and areas where natural disasters such as earthquakes occur, leaf area indexes simulated by the method cannot reflect actual vegetation canopy change characteristics.

Therefore, the applicant considers using the target area corrected by the ground observation data to correct the MODIS LAI data as the input of the model instead of the original LAI simulation algorithm, so as to avoid the above-mentioned problems. In some preferred embodiments, a specific operation method of the step S3 is provided, as shown in fig. 4, including:

Further, in the existing SWAT model, plant climates are reflected by setting a dormant period, assuming that vegetation does not grow during the dormant period. After the rest period, the leaf area of both the arbor and perennial plant was set to a minimum (set in the plant growth bank). The model defines the beginning and end of the sleep period using a threshold daily length. When the daily length is lower than the threshold daily length, all the annual plants except the warm season enter a dormant state, and part of leaves are converted into residues; when the daily length is above the threshold, the plant dormancy ends and the model begins to simulate biomass growth. The setting mode of the dormancy period has larger difference with the actual plant climate characteristics, and particularly has poor adaptability to the growth cycle of tropical and subtropical plants. Moreover, for mountainous areas with complex topography and climate conditions, meteorological sites are rare, and it is difficult to simulate the influence of reactive climate change on vegetation climate through a model.

Therefore, the applicant considered to avoid the above-described problem by defining the sleep period mode instead of the threshold daily length of the SWAT model using, as model inputs, object candidate data (mainly SOG and EOG) extracted from the target area correction MODIS LAI data corrected by the ground observation data. In some preferred embodiments, a specific operation method of the step S4 is provided, including:

In some preferred embodiments, a specific operation method of the step S4 is provided, as shown in fig. 5, including:

s401, correcting MODIS LAI data extract candidate data comprising a starting point SOG and a terminal point EOG of vegetation growing period by using the target area by adopting a multi-period weatherometer inversion method; the method for extracting the weathered data is numerous, and can be determined by a person skilled in the art according to practical situations, and in this embodiment, a universal multi-period weathered inversion method (UMPM) is introduced, and the method is mainly divided into a data preprocessing part 3, a weathered feature parameter extraction part 3, and a parameter inspection part 3.

1) The data preprocessing is to eliminate noise in the original data, reconstruct the LAI time sequence number, and enable the LAI time sequence number to meet the requirements of a method for extracting the climatic characteristic parameters.

2) For extracting the climatic characteristic parameters, the adopted strategy is to firstly determine the climatic growth cycle number, and then determine the key climatic nodes in each growth cycle respectively. And determining an inversion strategy of the key object weather nodes by adopting a combination of a Logistic function fitting method and a piecewise linear fitting method. Selecting a Logistic fitting method as a main algorithm; and when the inversion of the main algorithm fails or is not applicable, selecting a standby algorithm for inversion.

3) The parameter inspection is to inspect false values possibly occurring in the process of extracting the climatic characteristic parameters so as to ensure the reliability and stability of the product result.

The extraction method of the weathered data can be as follows: extracting a vegetation growth period starting point (SOG) and a vegetation growth period end point (EOG), and a peak starting point (SOP) and a vegetation growth period end point (EOP) by adopting a general multi-period weather inversion method.

Examples

This example takes the Minjiang upstream basin as an example. After 1999, the country enforces the policy of returning to the forest and returning to the grass, and the vegetation is gradually restored, but the venturi earthquake occurring in 2008 causes serious damage to the local ecosystem and vegetation. The forest area after earthquake is reduced by 2.75%, the damaged area of conifer forest is the most, and then the forest area after disaster is gradually recovered after broad-leaved forest and needle-broad hybrid forest. At present, a leaf area index module in the existing SWAT model cannot accurately simulate LAI changes under the influence of multiple factors such as earthquake damage, human activities and the like, and the SWAT model optimization method provided by the application takes vegetation types, leaf area indexes and physical periods as vegetation canopy characteristic factors, realizes multi-parameter dynamic expression of remote sensing canopy characteristic factors on the basis of the SWAT model, and can effectively improve the depicting capability of the model on vegetation leaf area index space-time changes. As shown in fig. 8, based on the remote sensing leaf area index (MOD 15 A2) analysis, the leaf area index upstream of Minjiang was reduced by 5.94% in 2001 and 2018; the area proportion of the low value area (LAI is less than or equal to 2) is increased from 22.09% to 29.32%, and vegetation degradation is obvious in the southeast Wenchen area of the river basin. Under the influence of various factors such as climate change, human activity, earthquake and the like, the condition of the Minjiang upstream vegetation changes obviously.

On the other hand, the current SWAT model determines the beginning and the end of a growing period or a dormant period according to the temperature and the daily length, and the distribution of mountain meteorological sites is rare, so that the space-time change of vegetation climate information is difficult to accurately reflect. By adopting the SWAT model optimization method provided by the application, as shown in fig. 9, remote sensing climatic information extracted based on MODIS LAI data shows a significant advance trend in the green returning period of the grasslands and deciduous broadleaf forests at the upstream of Minjiang in 2001-2016, and a non-significant advance trend in the mixed forests. Moreover, the rate of change of the turning green period is higher in the high altitude section (> 3500 m), and the trend is more remarkable. Therefore, the application adopts the extracted weather information based on the remote sensing data as the model input, and can effectively improve the depicting capability of the model on the vegetation weather index spatial distribution pattern and evolution.

Further, the method may be implemented in any type of computing platform operatively connected to a suitable computing platform, including, but not limited to, a personal computer, mini-computer, mainframe, workstation, network or distributed computing environment, separate or integrated computer platform, or in communication with a charged particle tool or other imaging device, and so forth. Aspects of the application may be implemented in machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, optical read and/or write storage medium, RAM, ROM, etc., such that it is readable by a programmable computer, which when read by a computer, is operable to configure and operate the computer to perform the processes described herein. Further, the machine readable code, or portions thereof, may be transmitted over a wired or wireless network. When such media includes instructions or programs that, in conjunction with a microprocessor or other data processor, implement the steps described above, the application described herein includes these and other different types of non-transitory computer-readable storage media. The application also includes the computer itself when programmed according to the methods and techniques of the present application.

The foregoing has shown and described the basic principles, principal features and advantages of the application. It will be understood by those skilled in the art that the present application is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present application, and various changes and modifications may be made without departing from the spirit and scope of the application, which is defined in the appended claims. The scope of the application is defined by the appended claims and equivalents thereof.

Claims

1. The SWAT model optimization method based on the time-varying characteristics of the vegetation canopy is characterized by comprising the following steps:

s4, correcting MODIS LAI data extract waiting period data by using the target area, and using the waiting period data to replace a threshold daily length definition dormancy period mode in a SWAT model as the input of the model;

the method for performing temporal and spatial resolution conversion on the MODIS LAI data in step S2 includes:

2. The method of optimizing a SWAT model based on time-varying characteristics of a vegetation canopy as claimed in claim 1, wherein the method of correcting the regional observation LAI data from the target region in step S2 comprises:

s212, training a MODIS-ESU ‍ LAI conversion model by adopting MODIS LAI data and actually measured sample area LAI data at the same time and at the same position;

s213, inputting the target area MODIS LAI data into a trained MODIS-ESU ‍ LAI conversion model to obtain target area correction MODIS LAI data.

3. The method of optimizing a SWAT model based on time-varying characteristics of a vegetation canopy of claim 1, wherein step S3 further comprises:

4. The method of optimizing a SWAT model based on time-varying characteristics of a vegetation canopy of claim 1, wherein step S4 further comprises:

s401 and ‍, correcting MODIS LAI data extract candidate period data comprising a starting point SOG and a terminal point EOG of vegetation growing period by using a multi-period climatic inversion method;