RU2406295C1 - Method of ecological monitoring of forests - Google Patents

Abstract

FIELD: agriculture.

SUBSTANCE: method of ecological monitoring of forests includes remote registration of brightness fields of forest vegetation with aerospace equipment. Remote registration of brightness fields of forest vegetation is accomplished by sensing multi-or hyperspectral sensor in the green G (450-550 nm), red R (550-670 nm) and near-infrared NIR (670-950 nm) bands of the spectrum with simultaneous obtaining of digital images for each band. Expectation value of signals (M_G, M_R, M_NIR) are calculated in each band, a matrix of the resulting image is formed by pixel adding of images of G, R, NIR. Signs of forest pathology are calculated as an index of vitality g=M_G/(M_G+M_R), lesion index R=M_R/(M_G+M_R), the normalised differential index of producing phytomass NDVI=(M_NIR-M_R)/( M_NIR +M_R), the area of reliefs of leaf canopies of image R and the resulting image, respectively, S_pR, S_p0, the average frequency of the spatial spectrums of the image R and the resulting image 0, respectively F_avR, F_av0. Degree of weakening Q of forest crop at the area S_o is determined by a calibrated standard regressional relationship of the form: Q≈0.6(NDVI g)^-1[r(l-NDVI)^1/3("П"_R/"П"₀)(D_R/D₀)^1,2, where: "П"_R, "П"₀ are estimated completeness of forest crop calculated by the reliefs squares of the corresponding matrices "П"_R=S_pR/S₀, "П"₀=S_p0/S₀; D_R, D₀ are the average tree head diametres equal to, respectively, D_r=1/F_avR, D₀=1/F_av0.

EFFECT: method provides speed and accuracy of estimation of ecological state of forest areas.

6 dwg, 2 tbl, 1 ex

Description

The invention relates to forestry and can find application in the remote monitoring of forests on vast areas.

The factors causing forest pathological phenomena can be environmental pollution as harmful emissions from industrial enterprises, leading to necrosis of photosynthetic organs (leaves, needles), and climatic anomalies and outbreaks of mass reproduction of insect pests.

Known methods for assessing the forest pathological and sanitary state of plantings during ground (field) taxation of forests, such as: visual (eye); selective measuring; enumeration method [see, for example, the Handbook, “All-Union Standards for Forest Taxation,” Cosmos, M., 1992, pp. 180-185, tables 60-63, “Scale for assessing tree condition categories,” and “Characteristics of categories of plantation status” -analog].

In the analogue method, visual signs of forest pathology are a change in the color of needles, leaves, from green to pale green and yellow, thinning of crowns, dryness, drying out of branches. According to visual signs, forests are divided into status categories: from 1 category to 6 categories.

Selective measurement and enumeration methods include recounting trees at reference sites, estimating the stock of trees in each state category and calculating the arithmetic mean stock per 1 ha for each state category.

The degree of weakening of the stands is estimated by the weighted average value of the product of the state category (from 1 to 6) by the percentage of the number of trees in the plantation by state categories (% N _i / N _Σ ).

With a weighted average of no more than 1.5, the plantation is considered healthy, up to 2.5 - weakened, up to 3.5 - very weakened and up to 4.5 - drying out.

The disadvantages of the analogue method include the high complexity and inoperability of natural taxation methods, errors in the distribution of the results of calculation of control sites to the entire taxed array.

The closest analogue is the "Method for assessing the state of forests", patent RU No. 2038001, cl. A01G 23/00, 1995 - analogue.

In the analogue method, remote spectrometric measurements are carried out, from the board of the orbital station, of the spectral brightness coefficients of the probed forest area in blue B, green G and red R sections of the visible spectrum, the values of chromatic vitality coefficients are calculated

g = G / (B + Gh-R)

and red defeat

r = R / (B + G + R),

chromatic coefficient regression is calculated

calibrate it by measuring control sites with known categories of plant conditions on them and evaluate the state of the forest (in points) by the ratio of the current calculated values of g, r. The known method has the following disadvantages:

- low spatial resolution of spectrometric means (from hundreds of meters to units of km), which does not allow to detect lesions within one hectare;

- Static instability of the results, since the assessment is carried out on the only measurable parameter-spectral brightness coefficient. The problem solved by the claimed method consists in the quantitative measurement of signs of forest pathologists that exist simultaneously in several spectral ranges of the solar radiation flux reflected from the canopy and the regression dependence between the measured signs and the standard integrated ecological indicator of the degree of weakening of stands.

The technical result is achieved by the fact that the method of ecological monitoring of forests, including remote registration of brightness fields of forest vegetation by aerospace means, and remote registration of brightness fields of forest vegetation is carried out by sensing with a multi- or hyperspectral sensor in green G (450-550 nm), red R (550 -670 nm) and near infrared NIR (670-950 nm) areas of the spectrum with the simultaneous receipt of digital images for each zone, calculate the expected value of the signals (M _G , M _R , M _NIR ) in each zone, a matrix of the resulting image is formed by pixel-by-pixel addition of G, R, NIR images, signs of forest pathology are calculated in the form of vitality index g = M _G / (M _G + M _R ), lesion index R = M _R / (M _G + M _R ), the normalized differential index of the producing phytomass NDVI = (M _NIR -M _R ) / (M _NIR + M _R ), the relief areas of the tree canopies of the image R and the resulting image, respectively, S _pR , S _p0 , the average frequency of the spatial spectra of the image R and the resulting image is 0, respectively, F _cpR , F _cp0 , and the degree is weakened Q of the stand of a site with an area of So, is determined by a calibrated reference regression dependence of the form:

Q≈0.6 (NDVIg) ^-1 [r (1-NDVI) ^1/3 (P _R / P ₀ ) (D _R / D ₀ ) ^1.2

where P _R , P ₀ - the estimated completeness of the stands, calculated through the area of the reliefs of the corresponding matrices P _R = S _PR / S ₀ , Po = S _p0 / S ₀ ;

D _R , D ₀ are the diameters of the crowns of the middle tree, respectively equal to D _R = 1 / F _cpR , D ₀ = 1 / F _cp0 .

The invention is illustrated by drawings, where:

Figure 1 - the mathematical expectation of the matrix of a digital image in the ranges of G, R, NIR; 1 - healthy forest, 2 - mixed forest in the digression stage;

Figure 2 - dependence of the relative volume of phytomass on the values of the normalized differential index NDVI;

Figure 3 - the regression function, the criterion for the dependence of the state of the forest on the ratio of the indices (g NDVT) and r (1-NDVI);

Figure 4 - dependence of the completeness of the stand on the ratio of the relief area of the canopy to the geometric area (S _p / S ₀ );

Figure 5 - envelopes of the spatial spectra of the images: the resulting F _cp0 and R range F _cpR ;

6 is a functional diagram of a device that implements the method.

The technical essence of the method is as follows

According to the analogue [p. 182, table 60 “Scale for assessing categories of tree condition”] visual signs of weakening of stands are:

- color change of needles, leaves from saturated green to pale green, yellow-green, yellow-orange;

- a decrease in the length of needles, leaf sizes, a decrease in foliage, congestion, and the volume of producing phytomass;

- openwork of tree crowns, shortened growth of branches, gaps in the canopy;

- dryness, dryness of individual individuals, thinning of the stand as a whole.

The listed visual signs of forest pathology in the claimed method are measured remotely, instrumentally, and their integral effect in the resulting indicator of the vitality of the stands is taken into account by the regression function calibrated by measuring the reference areas.

Changing the color of the plants changes the ratio between the fraction of absorbed and reflected solar radiation during photosynthesis in the spectral bands G and R. A quantitative characteristic of the photosynthesis efficiency is the calculation of the green index g and wilting r through the average values of the digital image matrices: g = M _G / (M _G + M _R ) and r = M _R / (M _R + M _R ).

The change in the mathematical expectation of image matrices depending on the change in color of the stands is shown in the graph of Fig. 1 (1 - healthy forest, 2 - mixed forest in the digression stage). The listed coefficients cover all stages of digression of plant communities; their quantitative change depending on the degree of inhibition occurs monotonously. A sign that is not taken into account in the analogue and prototype is the steepness of the function graph during the transition from the R band to the NIR. To assess this steepness, the so-called normalized differential vegetation index NDVI is used [see, for example, the Digest, “Aerospace Methods and Geoinformation Systems in Forestry and Forestry”, materials of the Second All-Russian Meeting, Moscow State University, Moscow, 1998, pp. .119-122, article, Zhirin V.M. “An approximate assessment of the phytomass of the forest cover using the vegetation index.”].

Quantitatively, NDVI characterizes the contrast of green vegetation with other natural formations, equal to

NDVI = (M _NIR -M _R ) / (M _NIR + M _R ).

The distribution of phytomass values of forest lands (tons / ha of absolutely dry matter according to the July measurements) is presented in table 1. (in parentheses is the area of vegetation type in% of the total area of the site).

Table 1 Distribution of phytomass values of forest lands (t / ha of absolutely dry matter according to July measurements) Intervals of July NDVI Values Larch plantings Natural rare larch Yernik thickets and other shrubs Swamps, water surfaces Sections of mountain tundra Bare stony placers Total t / ha (area%) 0.11-0.285 1.26 (4.2) 0.97 (7) 0.26 (8.5) 0.02 (0.4) 3.26 (25.1) 2.36 (54.8) 8.13 (100) 0.286-0.335 1.82 (6.1) 1.32 (9.5) 0.29 (9.6) 0.07 (1.3) 4.78 (36.8) 1.58 (36.7) 9.86 (100) 0.356-0.39 3.95 (13.2) 1.78 (12.8) 0.44 (14.5) 0.2 (3.6) 3.34 (25.7) 0.48 (11.2) 14.72 (100) 0.391-0.425 3.5 (11.7) 2.5 (18.0) 0.34 (11.4) 0.28 (5.0) 3.85 (29.6) 1.04 (24.3) 11.51 (100) 0.426-0.495 2.75 (9.2) 2.4 (17.3) 0.43 (14.3) 0.2 (3.6) 4.73 (36.4) 0.82 (19.2) 11.33 (100) 0.496-0.53 1.7 (5.7) 1.35 (9.7) 0.5 (16.5) 0 7.29 (56.1) 0.52 (12.0) 11.36 (100) The average value of phytomass t / ha of vegetation type 29.9 13.9 3.0 5.5 13.0 4.3

As follows from table 1, NDVI values increase with increasing volume of phytomass of the probed areas. Reduces, char (stone placers), shrubs reduce the phytomass of the probed forest areas, which is reflected in a decrease in the resulting NDVI. The change in the relative volume of phytomass of forest plots from the measured NDVI values is illustrated in the graph in FIG. 2.

In the claimed method, the calculations use the method of relations of average values of digital images I (x, y) obtained in the spectral zones G, R, NIR, synchronously, under the same external shooting conditions (solar height, viewing angle, weather conditions), which provides the elimination of systematic errors and the static stability of the measurement results. A quantitative change in the categories of planting state from the ratio of the listed indices g, g, NDVI is illustrated in the graph in FIG. 3. Obtaining synchronous measurements in zones G, R, NIR can be provided by an Astrogon-1 hyperspectrometer [see, for example, Vulkan-Astrogon Small Spacecraft with a high-resolution hyperspectrometer, draft design, RKA, NIIEM, STC Reagent ”, 2002, pp. 6-14, Astrogon hyperspectrometer for solving remote sensing problems with spacecraft] or other optoelectronic multichannel systems.

The Astrogon-1 hyperspectrometer has six spectral ranges in the range from 0.25 to 2.5 μm with a spectral resolution of 1-3 nm in each range with a spatial resolution of up to 1 m and digitization of the signal amplitude on a scale of up to 12 binary digits.

To quantify the degree of change in the structure of a weakened stand (completeness of planting, openwork), the resulting image is formed by pixel-by-pixel addition of images G, R, NIR. It is known that the variance of the sum of independent random variables is equal to the sum of the variances.

Since the image in each spectral zone G, R, NIR is obtained independently, the image arrays should be considered independent. Then, the variance of the resulting image σ _E ² = σ _G ² + σ _R ² + σ _NIR ² significantly exceeds the variance of the terms, thereby increasing the contrast of the resulting image and, as a result, the accuracy of the calculations.

The parameter, taking into account the “openwork” of crowns of weakened stands, is the distribution of trees in the stand according to the diameters of crowns. This information is contained in the envelope of the spatial spectrum of the image matrix I (x, y). Software envelope calculation is carried out by two-dimensional Fourier transform, which is an integral part of the specialized software MATHCAD.

On the applicability of the software calculation of the crown diameters of tree canopies [see, for example, Patent RU No. 2242867, 2004, “Method for calculating the stock of forests”]. The envelopes of the spatial spectra of the resulting image and the R image are illustrated by graphs in figure 5.

Decrease in concentration, dryness, dryness of crown reduce the size of the diameter of crowns, which affects the shift of the spatial spectrum towards high frequencies. The crown diameter of the middle tree is related to the spatial frequency D _cp = 1 / F _cp . The height of the middle tree H _cp = 7 (D _cp ) ^1.2 and the planting stock [m ³ / ha] depend on the diameter of the crown.

Planting stock also depends on the number of trees in the plot, i.e. from the fullness of planting.

The completeness of the stands is calculated by software, by calculating the relief area of the canopy [see Patent RU No. 92294622, 2007, “A method for determining the completeness of stands”].

The graph of the function of the completeness of the stand P from the ratio of the relief area of the canopy S _p to the geometric area of the plot S ₀ (equal to the product of the number of pixels in the matrix by the spatial resolution of one pixel Δ) is shown in Figure 4. The completeness of the stand of the matrix of the resulting image P _o and the completeness of the stand of the matrix R of the image of P _{R are} calculated.

In accordance with the "All-Union Standards for Forest Taxation" [analog. Handbook, M., Kolos, 1992, p. 188] the degree of ecological weakening of the stand Q is calculated as the product of the state category (graph in Fig. 3.) by the coefficient α of the ratio of the stock of weakened trees P _R D _R ^1.2 to the total stock of the site P ₀ D ₀ ^1,2 , α = P _R / P ₀ (D _R / D ₀ ) ^1,2 .

Obtaining the reference regression function in the form of a normative indicator of the weakening of Q stands as a function of the listed forest pathological features is presented in the implementation example.

An example implementation of the method.

The claimed method can be implemented on the basis of the device according to the scheme of Fig.6. Functional diagram of the device contains a space observation platform 1, such as a laboratory module 77 KML, docked with the international space station (ISS). A multi- or hyperspectral optical-electronic camera of high spatial resolution 2 is installed on the space platform, for example, an Astrogon-1 hyperspectrometer that takes pictures of scaffolds 3 by commands from the onboard control complex (BKU) 4, based on programs laid down in the BKU from flight control center 5 via command line radio 6.

Forest images obtained synchronously in the three spectral zones G, R, NIR are stored in the buffer memory 7. According to the commands of the control centers in the radio-visibility zones of the space platform and ground receiving points of information PPI 8, the data is reset via the mobile communication channel 9. After the preliminary processing of the frames according to official features (turn number, time of shooting, plot coordinates) on PPI 8 information is transmitted through the PMC to the thematic processing center 10, where through the communication channel 11 it is sent to the thematic processing PC 12, to the PC Math processing includes: a processor 13, an operational memory 14, a hard drive 15, a display 16, a printer 17, a keyboard 18. The results of processing the ecological state of the forest areas are displayed on the web server 19.

Initial for environmental monitoring are estimates of the same forest area, simultaneously obtained in three zones of the G, R, NIR spectrum.

In the process of flight testing of a hyperspectrometer on an aircraft carrier, a series of images of forest massifs containing reference (test) sites, respectively, categories 1, 2, 4, was obtained. Assessment of the state of the stands on the reference sites was carried out according to the operations of the analogue method.

The mathematical expectation of the matrix signals in a quantization scale of 0 ... 255 levels of sections 1.2 and 4 of the status categories and the calculated values of the indices g, r, NDVI are presented in table 2.

table 2 Status Category M _g M _R M _BIC g r NDVI 1-NDVI Etalon.plosch. one 38 22 196 0.7 0.3 0.8 0.2 2 43 47 190 0.51 0.49 0.6 0.4 four fifty 77 182 0.4 0.6 0.41 0.59

To obtain the state regression function (k) in the form of a multi-parameter dependence on forest pathological indices, the Mathematical Institute named after Steklov recommended monotone power functions of the type k = a b ^x c ^y ... We will look for the regression function in the form of the dependence k = a (g NDVI) ^x [r (1-NDVI] ^y .

According to table 2, the initial system of equations will take the form:

1 = α (0.7 0.8) ^x (0.3 0.2)

2 = α (0.51 0.6) ^x (0.49 0.4)

4 = α (0.4 0.41) ^x (0.6 0.59)

Logarithm system of power equations is reduced to linear. The solution of the system was carried out according to the Cramer-Sarrius rule. The following unknown values were obtained: a = 0.6; x = -1; y = 0.3; which does not contradict logic, because the lower the vitality index (g * NDVI), the more weakened the planting is (i.e., inversely proportional). Thus, the monotonous regression function of the categories of plantation status, calibrated from the images of the reference areas, is presented in an analytical form (graph in Fig. 3) as

k = 0.6 / (g NDVI) ^-1 [r (1-NDVI] ^0.3

Further, as an example, we consider the results of software processing images of a plot of the 2nd category, they are illustrated by the graphs of Fig. 4, Fig. 5.

According to the graph of Figure 4, we determine the average frequency of the spatial spectra of the resulting matrix F _cp0 = 0.75 and the matrix F _cpR = 1.25.

The estimated completeness of the stands, calculated by the corresponding matrices P ₀ = 0.6; P _R = 0.5

Coefficient α = П _R / П ₀ (D _R / D ₀ ) ^1.2 = 0.5 / 0.6 (D _R / D ₀ ) ^1.2

According to the normative indicator, the state of forest stands Q = (3 * 0.45), which is <1.5. Planting should be considered healthy.

The claimed method can be implemented on the existing technical base of the Russian Segment of the International Space Station (PC ISS) after docking with it the laboratory module 77 KML in 2010, developed by RSC Energia named after SP Koroleva on the basis of domestic spacecraft for remote sensing of the Earth (DEZ), such as Resurs-DK, on the basis of foreign systems of DEZ.

The effectiveness of the method is determined by the productivity, efficiency and accuracy of the assessment of the ecological status of forests. Efficiency, productivity and globality are ensured by the features of satellite imagery, and accuracy is ensured by a set of measured forest pathological features that are not implemented in analogues.

Claims (1)

A method of ecological monitoring of forests, including remote registration of brightness fields of forest vegetation by aerospace means, and remote registration of brightness fields of forest vegetation is carried out by sensing with a multi - or hyperspectral sensor in green G (450-550 nm), red R (550-670 nm) and near infrared NIR (670-950 nm) areas of the spectrum with the simultaneous receipt of digital images for each zone, calculate the mathematical expectation of the signals (M _G , M _R , M _BIC ) in each zone, a matrix of the resulting image is formed by pixel-by-pixel addition of images G, R, NIR, signs of forest pathology are calculated in the form of vitality index g = M _G / (M _G + M _R ), lesion index R = M _R / (M _G + M _R ), the normalized differential index of the producing phytomass NDVI = (M _BIC -M _R ) / (M _BIC + M _R ), the relief areas of the wood canopies of the image R and the resulting image, respectively, S _pR , S _p0 , the average frequency of the spatial spectra of the image R and the resulting image 0, respectively, F _cpR , F _cp0 , and the degree of weakening Q of the stand stand of an area of So, is determined by the calibrated reference regression dependence of the form:
Q≈0.6 (NDVIg) ^-one [r (1-NDVI) ^1/3 (P _R /P ₀ ) (D _R / D ₀ ) ^1/2 ,
where P _R , P ₀  - the estimated completeness of the stands, calculated through the area of the reliefs of the corresponding matrices P _R = S _pR / S ₀ , P ₀ = S _p0 / S ₀ ;
D _R , D ₀  the diameters of the crowns of the middle tree, respectively, equal to D _R = 1 / F _cpR , D ₀ = 1 / F _cp0 .

Images