Pixel-level vegetation water consumption calculation method based on multi-dimensional feature space

Abstract

The invention provides a pixel-level vegetation water consumption calculation method based on a multi-dimensional feature space in the technical field of remote sensing monitoring, and the method comprises the steps: calculating the surface net radiation Rn through satellite remote sensing data and GLDAS grid meteorological data, and then calculating the soil heat flux G; constructing a four-dimensional feature space by using four variables including canopy coverage Fc, surface temperature Ts, near-surface temperature Ta and wind speed Ws; then the maximum surface temperature Tsmax and the minimum surface temperature Tsmin are determined; a high-dimensional dry boundary and a high-dimensional wet boundary are obtained through fitting solving by means of the least square method; and calculating the daily water consumption of the vegetation in the target area through the evaporation ratio. According to the method, pixel-level accurate observation is provided for vegetation water consumption in a macroscopic area, support can be provided for hydrological cycle research in an ecological system, and a basis can be provided for a water consumption rule and irrigation requirements of crops.

Description

Technical Field

[0001] The invention relates to the technical field of remote sensing monitoring of vegetation physiology, and in particular to a pixel-level vegetation water consumption calculation method based on multi-dimensional feature space. Background Art

[0002] Vegetation water consumption, that is, evapotranspiration of vegetation, is a key indicator to characterize the moisture status of vegetation, and is also an important part of surface water balance and energy balance. By accurately grasping the vegetation water consumption, support can be provided for regional water resources management. In current research and applications, vegetation water consumption is mostly directly observed through ground observation equipment, such as lysimeters, eddy systems, large-aperture scintillometers and other equipment, or indirectly estimated through regional water balance. However, these field measurement methods are usually sensitive to the method and location of field installation. When it comes to monitoring large areas, these measurement methods are inefficient and it is difficult to capture the spatial heterogeneity of regional vegetation water consumption.

[0003] With the development of remote sensing technology and various sensors in recent years, methods for inverting vegetation water consumption based on remote sensing observation information have gradually been proposed. These methods can be roughly divided into statistical regression methods and energy balance methods. The former directly fits the relationship between vegetation water consumption and remote sensing observation information through regression methods to establish a functional relationship between the two and then simulate water consumption. This method is simple and easy to implement, but it is not very interpretable and has poor portability. The latter uses remote sensing information and combines it with the principle of energy balance to invert the various energy components of the surface, and determines the energy component used to evaporate water, that is, the latent heat flux, and finally calculates the actual evapotranspiration. This type of method has a complete theoretical basis and high simulation accuracy, but the calculation process is complicated. Therefore, later generations proposed a method based on the Bowen ratio to simplify the solution of latent heat flux, such as the characteristic space method of NDVI-Ts.

[0004] At present, these methods all have certain limitations, which are mainly reflected in the following aspects:

[0005] (1) The method based on ground observation and spatial interpolation is less efficient for observing vegetation water consumption in larger areas.

[0006] (2) Remote sensing methods based on statistical regression have low interpretability and poor portability.

[0007] (3) The characteristic space method of NDVI-Ts is difficult to accurately characterize the upper limit of surface wetness and dryness, and constructing the characteristic space through a simple vegetation index has theoretical defects. Summary of the invention

[0008] In view of the deficiencies in the prior art, the present invention provides a pixel-level vegetation water consumption calculation method based on a multidimensional feature space, which can efficiently and accurately monitor the vegetation pixel-level water consumption and its spatiotemporal distribution characteristics at a regional scale.

[0009] To achieve the above object, the present invention adopts the following technical solutions:

[0010] A method for calculating vegetation water consumption at pixel level based on multidimensional feature space comprises the following steps:

[0011] S1. Download and organize MODIS satellite remote sensing data of the target area for the past 30 days or more, including: surface albedo α, canopy coverage Fc, normalized difference vegetation index NDVI, surface temperature Ts, and surface radiation rate ε;

[0012] S2. Download and organize the GLDAS grid meteorological data of the target area for 30 days or more that matches the MODIS data date, including: near-surface temperature Ta, wind speed Ws;

[0013] S3, using the resampling tool in ArcGIS software to downsample the above two categories to the same spatial resolution;

[0014] S4. Calculate the surface net radiation Rn using the surface albedo α, surface temperature Ts, and near-surface temperature Ta;

[0015] S5. Calculate soil heat flux G using surface albedo α, surface net radiation Rn, surface temperature Ts, and normalized vegetation index NDVI;

[0016] S6, construct a four-dimensional feature space using four variables: canopy coverage Fc, surface temperature Ts, near-surface temperature Ta, and wind speed Ws;

[0017] S7, canopy coverage Fc with a step size of 0.01, near-surface temperature Ta with a step size of 0.1K, wind speed Ws with a step size of 0.1m / s to construct the minimum sliding unit, and determine the maximum surface temperature Ts corresponding to each minimum sliding unit max and minimum surface temperature Ts min ;

[0018] S8, use the least squares method to fit Ts max , Ts min The nonlinear functional relationship with canopy coverage Fc, near-surface air temperature Ta, and wind speed Ws, namely, the high-dimensional dry limit and high-dimensional wet limit;

[0019] S9, using the surface temperature Ts of different pixels in the target area, combined with the high-dimensional dry boundary and high-dimensional wet boundary solved above, calculate the Bowen ratio ER of the target pixel;

[0020] S10. Calculate the daily water consumption of vegetation in the target area by using the evaporation ratio ER and the assumption that the daily evaporation ratio is constant.

[0021] To optimize the above technical solutions, the specific measures taken also include:

[0022] Furthermore, in step S4, the formula for calculating the surface net radiation Rn using the surface albedo α, the surface temperature Ts, and the near-surface air temperature Ta is as follows:

[0023]

[0024] Where Gsc is the solar constant, usually 1376W / m 2 ; θ is the solar zenith angle; Z is the target area elevation; J is the Julian day; σ is the Boltzmann constant, usually 5.67×10 -8 W / m 2 K 4 .

[0025] Furthermore, in step S5, the soil heat flux G is calculated using the surface albedo α, the surface net radiation Rn, the surface temperature Ts, and the normalized vegetation index NDVI:

[0026]

[0027] Where t depends on the observation time. The value is 0.9 before 12:00, 1.0 between 12:00 and 14:00, and 1.1 after 14:00.

[0028] Further, in step S8, the least square method is used to fit Ts max , Ts min The nonlinear functional relationship with canopy coverage Fc, near-surface temperature Ta, and wind speed Ws is as follows:

[0029] Ts max =a1Fc 2 +b1Ta 2 +c1W 2 +d1Fc×Ta+e1Ta×Ws+f1Ws×Fc+g1

[0030] Ts min =a2Fc 2 +b2Ta 2 +c2W 2 +d2Fc×Ta+e2Ta×Ws+f2Ws×Fc+g2

[0031] In the formula, the parameters a1, b1, c1, d1, e1, f1, g1 and a2, b2, c2, d2, e2, f2, g2 are solved by the least square method respectively and are dimensionless. Among them, a1, b1, c1, a2, b2, c2 are the coefficients of the quadratic term, d1, e1, f1, d2, e2, f2 are the coefficients of the linear term, and g1 and g2 are constant terms.

[0032] Furthermore, in step S9, the surface temperature Ts of different pixels in the target area is used, combined with the high-dimensional dry boundary and high-dimensional wet boundary solved above, and the formula for calculating the evaporation ratio ER of the target pixel is as follows:

[0033]

[0034] Furthermore, in step S10, the evaporation ratio ER is used, combined with the assumption that the daily evaporation ratio is constant, to calculate the daily water consumption of vegetation in the target area as follows:

[0035]

[0036] In the formula, is the geographic latitude of the pixel; δ is the solar declination, calculated by the following formula:

[0037]

[0038] The beneficial effects of the present invention are:

[0039] The method described in the present invention mainly constructs a pixel-level vegetation water consumption calculation method based on a multidimensional feature space, which mainly provides a method based on a multidimensional feature space, combined with satellite remote sensing observation information, which can accurately observe the surface vegetation water consumption and its spatiotemporal distribution characteristics. The proposed feature space that combines multiple variables such as wind speed, air temperature, ground temperature and vegetation coverage comprehensively considers the influence of multiple factors on the evapotranspiration of surface vegetation. The evaporation ratio calculated based on this method greatly reduces the complexity of the calculation while ensuring reliable accuracy. In addition, combined with multi-source remote sensing data, pixel-level observation of surface vegetation water consumption can be achieved, and the spatial distribution characteristics of vegetation water consumption can be clearly and accurately described. The present invention realizes the provision of pixel-level precise observation of vegetation water consumption in macro-regions, which can provide support for the study of hydrological cycles in ecosystems, and can also provide a basis for the water consumption laws and irrigation needs of crops. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the overall solution flow of the present invention.

[0041] Figure 2 It is an example of input variables for the target area of ​​the present invention based on the method of the present invention.

[0042] Figure 3It is an example of the net radiation result calculated for the target area of ​​the present invention based on the method described in the present invention.

[0043] Figure 4 It is an example of soil heat flux results calculated based on the method of the present invention for the target area of ​​the present invention.

[0044] Figure 5 It is an example of the evaporation ratio result calculated based on the method of the present invention for the target area of ​​the present invention.

[0045] Figure 6 It is an example of the result of vegetation water consumption calculated based on the method of the present invention in the target area of ​​the present invention.

[0046] Figure 7 It is a scatter plot comparing the vegetation water consumption results calculated by the method of the present invention with the ground station observation results. DETAILED DESCRIPTION

[0047] The present invention will now be described in further detail with reference to the accompanying drawings.

[0048] refer to Figure 1 , is a schematic diagram of the overall solution flow of the pixel-level vegetation water consumption calculation method based on multi-dimensional feature space of the present invention, which includes the following steps:

[0049] A. Collect and organize the MODIS satellite remote sensing data of the target area from 2001 to 2020 from NASA's data center website (https: / / ladsweb.modaps.eosdis.nasa.gov / search / ). The target area is Jiangsu, China. The MODIS satellite remote sensing data include: surface albedo α, canopy cover Fc, normalized difference vegetation index NDVI, surface temperature Ts, and surface emissivity ε. The spatial resolution of the data is 500m, and the temporal resolution is 8d synthesis.

[0050] B. Collect and organize the GLDAS raster meteorological data of the target area from 2001 to 2020 from NASA's data center website (https: / / disc.gsfc.nasa.gov / datasets), including: near-surface temperature Ta, wind speed Ws. The data date corresponds to the MODIS data, and the spatial resolution of the data is 0.25°.

[0051] C. Use the resampling tool in ArcGIS software to downsample the MODIS data to a spatial resolution of 0.25° to match the GLDAS data. The relevant input data examples are as follows: Figure 2 As shown;

[0052] D. Using the raster calculator tool in ArcGIS software, combined with the surface albedo α, surface temperature Ts, and near-surface temperature Ta data, the surface net radiation Rn is calculated. The calculation results are shown in the following example: Figure 3 As shown:

[0053]

[0054] Where Gsc is the solar constant, usually 1376W / m 2 ; θ is the solar zenith angle; Z is the target area elevation; J is the Julian day; σ is the Boltzmann constant, usually 5.67×10 -8 W / m 2 K 4 ;

[0055] E. Using the raster calculator tool in ArcGIS software, the soil heat flux G is calculated by combining the surface albedo α, surface net radiation Rn, surface temperature Ts, and normalized vegetation index NDVI. The calculation results are as follows: Figure 4 As shown:

[0056]

[0057] Where t depends on the observation time. The value is 0.9 before 12:00, 1.0 between 12:00 and 14:00, and 1.1 after 14:00.

[0058] F. Use Python programming language to construct a four-dimensional feature space with four variables, namely, extract the values ​​of all pixels and put them into a four-dimensional coordinate system;

[0059] G. Using Python programming language, the canopy coverage Fc is constructed with a step size of 0.01, the near-surface temperature Ta is constructed with a step size of 0.1K, and the wind speed Ws is constructed with a step size of 0.1m / s to determine the maximum surface temperature Ts corresponding to each minimum sliding unit max and minimum surface temperature Ts min ;

[0060] H. Use the least squares method to fit each minimum sliding unit Ts separately max , Ts min The nonlinear functional relationship with the corresponding canopy coverage Fc, near-surface temperature Ta, and wind speed Ws, namely the high-dimensional dry limit and high-dimensional wet limit, is as follows:

[0061] Ts max =a1Fc 2 +b1Ta 2 +c1W 2+d1Fc×Ta+e1Ta×Ws+f1Ws×Fc+g1

[0062] Ts min =a2Fc 2 +b2Ta 2 +c2W 2 +d2Fc×Ta+e2Ta×Ws+f2Ws×Fc+g2

[0063] In the formula, parameters a1, b1, c1, d1, e1, f1, g1 and a2, b2, c2, d2, e2, f2, g2 are respectively solved by the least squares method, which can be implemented in Python programming language. The calculated values ​​in this example are: -9.76, 1.14, -5.28, -1.91, -4.27, -19.25, 329.15, -1.27, 0.95, -2.78, 2.07, 3.59, 21.33, 271.88.

[0064] I. Using the raster calculator tool in ArcGIS software, integrate the surface albedo α, canopy coverage Fc, surface temperature Ts, near-surface temperature Ta, and wind speed Ws data that are downsampled to the same resolution, and calculate the evaporation ratio ER of the target pixel according to the following formula. The calculation results are shown in the following example: Figure 5 As shown:

[0065]

[0066] L. Using the raster calculator tool in ArcGIS software, the daily vegetation water consumption at the pixel level in the target area was calculated based on the assumption that the daily evaporation ratio is constant (the results are shown in Figure 6 (shown in the manner shown):

[0067]

[0068] In the formula, is the geographic latitude of the pixel; δ is the solar declination, calculated by the following formula:

[0069]

[0070] The vegetation water consumption data measured by the ground flux tower from 2001 to 2020 were compared with the calculation results of the method described in the present invention, proving that the method has good accuracy (such as Figure 7 shown).

[0071] The present invention is not limited to the above-mentioned embodiments. On the basis of the technical solution disclosed in the present invention, technicians in this field can make some substitutions and deformations to some technical features therein according to the disclosed technical content without creative labor, and these substitutions and deformations are all within the protection scope of the present invention.

Claims

1. A pixel-level vegetation water consumption calculation method based on multidimensional feature space, characterized in that: The following steps are involved: S1. Download and organize MODIS satellite remote sensing data of the target area for 30 days or more, including: surface albedo α, canopy cover Fc, normalized difference vegetation index NDVI, surface temperature Ts and surface radiation rate ε; S2, download and organize the GLDAS grid meteorological data of the target area with the same time range and matching the MODIS data date, including: near-surface temperature Ta, wind speed Ws; S3, using the resampling tool in ArcGIS software to downsample the above two types of data to the same spatial resolution; S4. Calculate the surface net radiation Rn using the surface albedo α, surface temperature Ts, and near-surface temperature Ta; S5. Calculate soil heat flux G using surface albedo α, surface net radiation Rn, surface temperature Ts, and normalized vegetation index NDVI; S6, construct a four-dimensional feature space using four variables: canopy coverage Fc, surface temperature Ts, near-surface temperature Ta, and wind speed Ws; S7, canopy coverage Fc with a step size of 0.01, near-surface temperature Ta with a step size of 0.1K, wind speed Ws with a step size of 0.1m / s to construct the minimum sliding unit, and determine the maximum surface temperature Ts corresponding to each minimum sliding unit max and minimum surface temperature Ts min ; S8, use the least squares method to fit Ts max , Ts min The nonlinear functional relationship between the canopy coverage Fc, the near-surface temperature Ta, and the wind speed Ws is solved to obtain the high-dimensional dry limit and high-dimensional wet limit; S9, using the surface temperature Ts of different pixels in the target area, combined with the high-dimensional dry boundary and high-dimensional wet boundary solved above, calculate the evaporation ratio ER of the target pixel; S12. Calculate the daily water consumption of vegetation in the target area by using the evaporation ratio ER and the assumption that the daily evaporation ratio is constant.

2. The method for calculating vegetation water consumption at pixel level based on multidimensional feature space according to claim 1, characterized in that: In step S4, the formula for calculating the surface net radiation Rn using the surface albedo α, the surface temperature Ts, and the near-surface air temperature Ta is as follows: Where Gsc is the solar constant, which is 1376W / m 2 ; θ is the solar zenith angle; Z is the target area elevation; J is the Julian day; σ is the Boltzmann constant, which is 5.67×10 -8 W / m 2 K 4 .

3. The method for calculating vegetation water consumption at pixel level based on multidimensional feature space according to claim 1, characterized in that: In step S5, the soil heat flux G is calculated using the surface albedo α, surface net radiation Rn, surface temperature Ts, and normalized vegetation index NDVI. The calculation formula is as follows: Where t depends on the observation time. The value is 0.9 before 12:00, 1.0 between 12:00 and 14:00, and 1.1 after 14:

00.

4. The method for calculating vegetation water consumption at pixel level based on multidimensional feature space according to claim 1, characterized in that: In step S8, the least squares method is used to fit Ts max , Ts min The nonlinear functional relationship with canopy coverage Fc, near-surface temperature Ta, and wind speed Ws is as follows: Ts max =a1Fc 2 +b1Ta 2 +c1Ws 2 +d1Fc×Ta+e1Ta×Ws+f1Ws×Fc+g1 Ts min =a2Fc 2 +b2Ta 2 +c2Ws 2 +d2Fc×Ta+e2Ta×Ws+f2Ws×Fc+g2 In the formula, the parameters a1, b1, c1, d1, e1, f1, g1 and a2, b2, c2, d2, e2, f2, g2 are solved by the least square method respectively and are dimensionless. Among them, a1, b1, c1, a2, b2, c2 are the coefficients of the quadratic term, d1, e1, f1, d2, e2, f2 are the coefficients of the linear term, and g1 and g2 are constant terms.

5. The method for calculating vegetation water consumption at pixel level based on multidimensional feature space according to claim 1, characterized in that: In step S9, the surface temperature Ts of different pixels in the target area is used, combined with the high-dimensional dry boundary and high-dimensional wet boundary solved above, and the formula for calculating the evaporation ratio ER of the target pixel is as follows:

6. The method for calculating vegetation water consumption at pixel level based on multidimensional feature space according to claim 1, characterized in that: In step S12, the evaporation ratio ER is used, combined with the assumption that the daily evaporation ratio is constant, to calculate the daily water consumption of vegetation in the target area as follows: In the formula, is the geographic latitude of the pixel; Gsc is the solar constant, which is 1376W / m 2 ; δ is the solar declination, calculated as: Where J is the Julian day.

Images