CC BY
9
Section 1. Astronautics
DOI: http://dx.doi.org/10.20534/AJT-17-5.6-3-10
Kh osiyat Ism atova, National Aviation Academy of Azerbaijan E-mail: spaseazer@rambler.ru
CONSTRUCTION OF SPECTRAL CURVES ACCORDING TO THE
DATA OF THE WORLDVIEW2 SATELLITE AND ANALYSIS ON THEIR BASIS OF SPECTRAL REFLECTANCE OF SALINE SOILS
Abstract: The work presents the results of spectral analysis of soils and vegetation cover on the territories of Kura-Araks lowland with deferent salinization degree. Generalized spectral curves are constructed based on the processing of eight-channel image from the WorldView2 satellite and the selection of test sections of soils and vegetation cover on sections of different salinization degree. It is shown that the run of curves of saline soils and vegetation vary greatly in the wavelength range 660-900 nm, and the spectral reflectance of vegetation is much higher than in saline soils.
Keywords: spectral reflectance, run of spectral curves, spectral properties of soil and vegetation on saline lands.
Introduction
Analyzing the materials on soil salinization of Kura-Araks lowland, it can be noted that the history of salt accumulation in the lowland is most closely connected with the Quaternary history of the Caspian Sea. Conical and deluvial-proluvial salinization is developed within the sloping plains, and alluvial salinization — within the lowland areas. Summarizing the considerations about the sources of salts, ways of their migration, it is suggested that nowadays salt masses come with river waters due to primary weathering processes. It is also indicated that the main source of salts in Kura-Araks lowland is salts of secondary migration cycles. These are salts that accumulate under lagoon and coastal-marine concentrations, beginning from the upper Sarmatian stage till the modern coastal-marine accumulation [1; 2; 3; 4; 5; 6].
Figurel.Examined territory (red contour)
Undoubtedly, the methods of melioration measures will have fundamental differences for each genetic form of salinization, among which the most common forms in the Kura-Araks lowland are delu-vial, conical and alluvial. Melioration methods in the republic are mainly developed for these salinization forms and there are sufficient research materials and
design studies on them [3; 5]. Hence the interest in solving environmental monitoring tasks of soil sa-linization processes and their propagation by aerospace monitoring methods is fully justified.
The investigated territory is a pilot territory on one section of the Kura-Araks lowland (figure 1). The capabilities of the WorldView-2 satellite with ultra-high spatial resolution are considered in this paper. One of the main technical characteristics of the WorldView-2 satellite (it captures images with a resolution of 0.5 m in panchromatic mode and 2 m in the multispectral mode) is that apart from four traditional spectral channels (red, green, blue and near infrared-1), WorldView-2 spectrometer includes the following additional channels: violet (or coastal blue), yellow (yellow edge), "red edge", near infrared-2 (NIR2) (figure 2) [13].
Eight spectral channels make it possible to implement three main advantages: increase of the decoding accuracy, a high level of detalization and new unique opportunities for extraction of necessary information.
Research methodology
The task of detecting the soils that become salty is one of the most important in the process of remote soil-melioration researches. The degree and type of soil salinization, the direction of sali-nization change of rocks, salt reserves, causes of salinization are estimated when monitoring the salt regime of irrigated soils [1,5,8]. Salinization of soils is detected by remote methods as in direct appearance of salts on soil surface, as well as in the change of reflectance of crop plants as a result of extinction of individual plants, their suppression and appearance of halophytic weeds. These phenomena change the tone and pattern of the image of saline soils massifs. Such studies were widely conducted on irrigated lands in the basins of the Amu Darya and Syr Darya [7; 9; 11].
The presence of salts in the soil in most cases helps to increase the level of light reflection by soils [15; 16; 17]. However, in real natural conditions
the presence of these substances can change other properties of soils, and, consequently, indirectly influence the coloring. For example, hygroscopic salts such as magnesium and calcium chlorides cause moisture absorption and thereby contribute to the increase of soil moisture. Consequently, when such salts are present, the soil has a darker color, not a lighter one [16; 19]. The influence of carbonates, as a rule, is reduced to a linear increase of the soil reflection coefficient with the growth of salt content in soils, the area and dimensions of which are determined by the instrument resolution. Therefore, the measured indicator of soil spectral brightness is already an averaged value. A special feature of the spectral information of additional channels is the red edge channel (705-745 nm,), where, as we see in the image (figure 2), it is possible to detect the suppressed vegetation by decrease of the reflectance in this channel.
On the basis of properties of 8-channel satellite, the research methodology consisting of several stages is proposed in this work:
1. Spectral analysis of soils by a space image WordView2;
2. Calculation of spectral parameters and construction of spectral curves in 8 channels of the satellite;
3. Comparison of spectral curves of soils with different salinization degrees and spectral curves of vegetation that grow on saline and non-saline soils;
The spectral analysis was carried out using the software ENVI version 5.2.
A visual analysis of spectral channels with the aim of selecting the optimal combination of channels identified the following optimal combinations for selection of soils with different salinization degrees: the combination of channels: 7,5,3, and 7,3,2 provide good contrast of vegetation and soils of different types, a combination of channels 8,7,6 is very different from other combinations, as it mainly indicates the content of water in soils, and, finally, the combination of channels 8,4,1 shows the
greatest contrast in soils of different salinization and humidity, as well as vegetation.
Study of spectral curves of soils and vegetation on lands with different salinization degree according to the data ofWorldView-2 satellite
Fundamental researches on spectrometry were carried out by E. L. Krinov as far back as in the 1940 s. Having laid the foundation for works on the optics of landscapes, he developed the first spectrometric classification, which eventually became classical. The largest amount of experimental data was obtained on the spectral brightness in the visible range. Apart from integral reflection, the reflection maximum and minimum values are sometimes calculated, and also the inflection points of spectral curve are determined, for which purpose we calculated the first derivative dp/dl and determined wavelengths that correspond to extremum points of the function p=f(l), where l — wavelength of electromagnetic spectrum (EM), p — spectral reflectance of Earth's surface objects [17].
Figure 2. Base values The system of basic indicators of soil spectral reflectance, proposed in 1995 by Orlov and co-authors, includes the following indicators (figure 2) [15; 17]:
• spectral reflectance coefficient at \ = 750nm,
P750;
• reflection coefficient — pe;
• absolute value of the inflection located in the middle part of the spectral curve, Ap = p650 - p480;
• Height of inflection;
• Location of inflection (length of wave of semi-inflection), \ ;
• The angle of the spectral curve slope and its individual sections.
The method for diagnostics of soil cover according to remote data includes the conduct of a space survey, collection of thematic cartographic materials and conduct of selective ground-based studies, and also a multi-zone survey in three channels of visible and near infrared subspectra is performed with the aim of increasing the effectiveness of decoding. The obtained results are processed by the cluster analysis method, on the basis of which the soil map is compiled [4; 8; 9; 11; 12; 15; 17; 19].
Unlike the survey in the visible and infrared zone, this work analyzes spectral curves of 8 channel WorldView-2 satellite.
Spectral curves and histograms were constructed along the examined territory using the software package ENVI 5.2, and also spectral characteristics such as average value, maximum and minimum values, scatter of brightness values within the selected test fragments of soils with different salinization degree and vegetation were calculated. The enumerated spectral characteristics and curves for each selected test fragment are given in the table 1 and table 2. Then a comparative analysis of the obtained curves and curves from early studies was made [5; 6; 8; 9].
Tables 1 and 2 demonstrate fragments of the image, where we selected test sections with different soil salinization and sections of saline lands covered with vegetation respectively and their spectral curves, obtained using the software package ENVI 5.2. According to spectral curves, extreme points in the behavior of curves of spectral reflectance of saline soils and vegetation cover are points with wavelengths: 400 nm, 480 nm, 510 nm, 610 nm, 660 nm, 720 nm, 830 nm (Fig. 4, 5, 6, 7). The values of spectral reflectance of soils at these points differ depending on the salinization degree, humidity, salinization type and wavelength. The properties of soils were analyzed on
the basis of ground-based measurements and literature sources [1; 2; 4; 8; 11; 12; 13; 14; 18].
Spectral analysis of soils with different salini-zation degree
Test fragments were taken on the territory of sol-onchaks (1, 2 fragments in table 1), solonchak soils covered with a small percentage of halophyte vegetation (fragment 3), in areas that were former cultivated, but highly saline (fragment 5), and also in soils on veg-
etable plots (fragment 4). Fragments of vegetation cover were taken on lands, which were long-term cultivated but subjected to salinization of one or another degree. The analysis of spectral curves of soils with different salinization degree was conducted (figure 3) and for comparison the curves were summarized in a single chart (figure 4), which shows that soils of different salinization degree vary in the values ofspectral reflectance and the run of a spectral curve (Fig. 3, 4).
Figure 3. The run of a spectral curve of highly saline soils: long-term cultivated soils (the first) and the run of a spectral curve of low salinity, medium humus cultivated lands (the third), and the second one is the run of a spectral curve in solonchak soils (solonetz-solonchak)
Comparison of the curves (figure 5) shows:
1. All soil curves have an inflection in the range 610-720 nm at 660 nm.
2. In the visible range the run of curves is the same for all soil test fragments (table 1), there are differences in the values of spectral reflectance depending on salinization degree and soil composition. The greatest reflectance was in solonchak soils (solonetz-solonchak), according to the image in fragment 1 (table 1) the soils are white.
3. Highly saline (according to ground-based measurements 5.364%) long-term cultivated (apparently readily soluble salts under secondary sali-
nization) soils have the greatest reflectance in comparison with all other curves (figure 1) in the range 550nm-610nm. In the picture these soils are dazzling white. Also, the curve has two inflections: the first — when passing from the red-edge zone to the near infrared 1 (NIR1) and the second one — when passing from NIR 1 to NIR2.
4. The greatest difference between soil curves of different salinization degrees is observed in the wave range from 660 nm to 900 nm. (i. e., in red edge, near infrared 1 and near infrared 2 zones). Here the soil curve with vegetable plots has the lowest reflectance (fragment 4 in table1).
Endmember Collection Spectra
Î50
250
/ \ //хч^-^
\\ v 1 ro1 — -
\\ R01 n
\W ROI #3
V '"V"' ROI #4
Wavelength (rtm)
Figure 4. Generalized chart of spectral curves of soils of different salinization degrees
Thus, the run of a curve and the values of spectral reflectance on soils with different salinization degrees vary in the range of 660-90 nm (figure 4).
Spectral analysis of vegetation cover in saline soils
The vegetation cover in saline soils is mainly herbaceous, as the survey was conducted in March and the vegetation shoots were insignificant. Table 2 shows spectral curves of the vegetation (herbaceous) cover in soils with different salinization degree and in the last line — the curve in vegetation plots of slightly saline lands.
The first line shows the curve of vegetation on solonchaks, the second curve — dense vegetation on medium saline cultivated lands, the third one — the curve of sparse vegetation on medium saline lands and the fourth curve is the vegetation cover on vegetable plots, slightly saline lands. The analysis
of spectral curves in the figure 6 shows that spectral reflectance of vegetation on long-term cultivated medium saline soils is much higher than that of vegetation on cultivated non-saline, medium humus soils both in the visible range and in the infrared band.
Three curves summarized in the single chart (figure 5) show:
1. Reflectance of herbaceous vegetation cover on solonetz-solonchak, heavily saline lands in the red edge zone (705-745) is lower than that of densely cultivated vegetation and sparse vegetation on medium saline cultivated lands.
2. Run of the curve of dense cultivated vegetation in NIR1 increases and passes into a straight line in NIR2, and on the contrary, the run of the curve of sparse vegetation decreases in NIRland passes into a monotonically increasing straight line in NIR 2.
Figure 5. The run of spectral curves of herbaceous vegetation: 6 - curve on solonetz-solonchak, medium-saline soils, occasionally covered with vegetation, 7 - curve of dense herbaceous vegetation on medium saline cultivated lands, 8 - curve of sparse herbaceous vegetation on medium-saline long-cultivated lands
400
300
Endmember Collection Spectra
ROI #5
; — ROI #6
: — ROI #7 if------ _ -H
i; / 1
500
600
700
soo
900
Wavelength (nm)
Figure 6. Spectral curves of soils and vegetation on soils of different salinization degrees, summarized in a single chart
All curves have an inflection in the interval 610 -720 nm at the point \ = 660 nm. In the section with dense vegetation (curve 7 in table 2), when passing from the red edge to NIR 1, the angle of the spectral curve slope is > 90°1.
Comparing all curves of herbaceous vegetation cover on heavily and medium saline sections and the run of vegetation curve from the section with low saline soils (curve 9 in table 2) shows that the spectral reflectance in NIR 1 and NIR 2 is higher in curve 9 and increases from NIR to NIR 2.
Conlusion
Many authors detected a spectral channel 770-880-nm (out of 13 examined ranges from 410 to 1250 nm) with the most expressed contrast of spectral brightness between open soil and vegetation cover [14; 15; 16; 17; 18; 19].
When in our researches we had the opportunity to study in detail spectral characteristics in eight spectral channels ofWorldView-2 satellite, we summarized all spectral curves in a single chart (figure 6), according to which we can make the following conclusion:
1. The spectral curves ofvegetation and soil with different salinization have an inflection in the range 610-720 nm at the point at \ — 660 nm;
2. The run of spectral curves in the range 410-660 nm is practically the same both in soils
and vegetation on sections with different salinization degree, the only difference is in the values of spectral reflectance: it is higher in soils than in vegetation;
3. The greatest difference in the run of spectral curves in soils and vegetation on sections with different salinization degree is observed in the range 660-900 nm.
4. In the range 660-900 nm the spectral reflectance of vegetation cover is much higher than in saline soils;
5. In the red edge zone, the spectral reflectance of vegetation on solonetz-solonchak soils is less than that of vegetation on medium and low saline long-term cultivated soils;
6. The angle of curve slope of a vegetation cover on saline lands is > 90 degrees, and in saline soils it is < 90 degrees when passing from the red edge zone to the near infrared at A= 720 nm. Hence the curve of soils decreases to NIR 1, the curve of vegetation cover increases to NIR 2.
The researches on this subject were carried out within the framework of the project "Science Development Fund" under the President of the Republic of Azerbaijan, code EIF-2013-9 (15) -46/17/1-M-27. The images are provided by OJSC "Azerkosmos" within the framework of this project.
Table 1. - Spectral characteristics of saline soils on test sections
№
Test fragment
Spectral curve of test fragment
Note
The section is white. Sierozem-meadow strongly saline, solonetz-solonchak, the content of sodium is more than 15-20%, magnesium is more than 50%, soils salinized by a dense residue (more than 2%) [1, 8]. According to ground-based measurements, salinization is 1.785.364 g/l._
Test section is light gray. Sierozem-meadow, solonetz-solonchak, the content of sodium is > 15-20%, magnesium is > 50%, soils salinized by a dense residue (more than 2%) [1, 8]. According to ground-based measurements, salinization is 1.78-3.64 g/l._
1
2
Table 2. - Spectral characteristics of vegetation on saline soils
№
Test fragment
Spectral curve of test fragment
Note
Test section 6 is gray-brown, sierozem-meadow, solonchak-solonetz, moistened, covered with vegetation.
Test section 7 — plots on cultivated meddium saline lands, with dense vegetation shoots
Test section 8 — plots on cultivated medium saline soils, with sparse vegetation shoots
Test section 9 — vegetable plots on little saline lands, with vegetation shoots
6
7
8
9
References:
1. Приклонский В. А. Формирование грунтовых вод в засушливых областях на примере Кура-Арак-синской низменности, Из-во АН СССР серия геологическая, - 1946, № 4.
2. Саваренский Ф. П. Кура-Араксинская низменность и её грунтовые воды и процессы их засоления. «Почвоведение», - 1929 г., - № 1-2.
3. Абдулов М. Р. Почвы с делювиальной формой засоления и вопросы их мелиорации. Из-во «Озан», -Баку, - 2003, - С. 269.
4. Волобуев В. Р. Основные вопросы генезиса и мелеорации засоленных земель Азербайджана. Известия АН Азерб.ССР, вып.Ш, - 1946, - № 12.
5. Микаилов Н. К. Природно-географические особенности и экологические особенности засоления почв Кура-Араксинской низменности, проблема мелиорации и оценки их плодородия. И-во «Озан», - Баку, - 2000, - С. 375.
6. Гасанов А. М., Исматова Х. Р., Нагиев П. Ю., Алиева М. Картографирование сельскохозяйственных угодий Кура-Араксинской низменности по результатам цифровой обработки видеоизображений. ИКИ РАН, сб. научных статей «Современные проблемы дистанционного зондирования», вып. 3, том.2, - 2006, - С. 300-306.
7. Книжников Ю. Ф., Кравцова В. И. Принцип множественности в современных аэрокосмических методах и способы дешифрирования серии снимков при сельскохозяйственных исследованиях
//Аэрокосмические методы в почвоведении и их использование в сельском хозяйстве. - М.: Наука, - 1990. - С. 47-54.
8. Орлов Д.С, Воробьёва Л. А., Суханова Н. И. Количественные параметры спектральной отражательной способности почв. Вестник МГУ, серия Почвоведение, - 1995 г., - No 42, - С. 35-42.
9. Панкова Е. И. Дистанционная диагностика засоления почв под культурой хлопчатника//Аэрокос-мические методы в почвоведении и их использование в сельском хозяйстве. - М.: Наука, - 1990. -С. 175-182.
10. Черный С. Г., Абрамов Д. А. Определение параметров линии почв черноземов правобережной Украины с помощью спектральных спутников Николаевский Национальный Аграрный Университет. ISSN 1684-9094. Gruntoznavstvo. - 2013. - Vol. 14, - No. 3-4, - С. 106-113.
11. Черный С. Г. Мониторинг гумусового состояния чернозема южного с использованием многоспектральных снимков спутника Landsat 7//Почвоведение и агрохимия. - 2013. - № 1 (50). - С. 97-105.
12. Черепанов А. С., Дружинина Е. Г. Спектральные свойства растительности и вегетационные индексы. Ж. Геоматика, - 2009, - № 3, - С. 28-33.
13. Дворкин Б. WorldView-2: революционное изменение подхода к получению, анализу и использованию данных космической съемки. Ж. ГЕОМАТИКА - № 3, -2011, - С. 27-29.
14. Huete A. R., Jackson R. D., Post D. F., - 1984, "Soil spectral effects on 4-space vegetation discrimination", Remote Sens. Environ., - No. 15, - P. 155-165.
15. Михайлова Н. А., Орлов Д. С. Оптические свойства почв и почвенных компонентов. - М. Наука, -1986, - С. 119.
16. Орлов Д. С. Цвет и диагностика почвы. Изд-во МГУ. Науки о Земле, - 1997 г., - С. 45-51.
17. Караванова Е. И. Оптические свойства почв и их природа, - 2001, - С. 152.
18. Garey A., Fox G. A., Sabbagh G. J., Searcy S. W., Yang C., - 2004, "An Automated Soil Line Identification Routine for Remotely Sensed Images", Soil Sci. Soc. Am. J., - No. 68, - P. 1326-1331.
19. Orlov D. S., Sukhanova N.I, Rozanov M. S., - 2001, "Spectral reflectivity of soils and their components", -Moscow, MGU, - 176 p.
