Wyniki wyszukiwania - BazTech

1

Ocena skuteczności modeli transferu promieniowania w badaniach stanu roślinności łąk

Jarocińska A.

Teledetekcja Środowiska

|

2012

|

T. 48

1--52

EN

Vegetation analysis is an important problem in regional and global scale. Because of pollution of environment and changes in the ecosystems plant monitoring is very important. Remote sensing data can be easily used to plant monitoring. That kind of method is much faster and more reliable than traditional approaches. Spectrometry analyzes the interactions between radiation and object and it uses measurement of radiation intensity as a function of wavelength. Each object emits and absorbs different quantity of radiation, so it is possible to recognise the object and check its characteristics analysing the spectrum. The subject of the researches is Polish meadows. The human usage of the meadows determines its proper functioning. Grasslands, which consist of meadows and pastures, cover 10% of Poland. Meadows are most extensively use. In Poland the crops from meadows (hay and green forage) are very low. The meadows in Poland are floristically and morphologically very diverse. Many factors influence on this ecosystem and that is why the monitoring is very important. The aim of the researches is to study the possibility of use of the Radiative Transfer Models in modelling the state of the heterogeneous vegetation cover of seminatural meadows in Poland. Two approaches are used to canopy analysis: statistical and modelling. In the statistic approach, biophysical parameters calculated from the image are correlated with reflectance or transmittance from fi eld measurements. In second approach physically based model is used to represent the photon transport inside leaves and canopy. The Radiative Transfer Models are based on the laws of optics. Developing the model results in better understanding of the interaction of light in canopy and leaves. The Radiative Transfer Models are often applied to vegetation modelling. The Radiative Transfer Models are physically based models which describe the interactions of radiation in atmosphere and vegetation. Adjusted models can be used to fast and precise analysis of biophysical parameters of the canopy. The canopy can be described as homogeneous layer consisting of leaves and spaces. The Radiative Transfer Models are algorithms which vary by input and output parameters, the level of the analysis, kinds of plants and other modifications. Models are used on two levels: single leaf and whole canopy. The first model, which is used in this research, is PROSPECT, which describes the multidirectional refl ectance and diffusion on a leaf level. It is often employed with other models that describe whole canopy. Leaf has the same properties on both sides, the reflection from the leaves is Lambertian. The input parameters in the model are: chlorophyll and carotenoid content, Equivalent Water Thickness and dry matter content and also leaf structure parameter that describe the leaf structure and complexity. Second model, which is used in the study, is the canopy reflectance model SAIL (Scattering by Arbitrarily Inclined Leaves). It simulates the top of the canopy bidirectional reflectance and it describes the canopy structure in a fairly simple way. In this analysis the 4-SAIL model will be used. This version has few input parameters that describe plants and soil: spectrometric data – reflectance and transmittance from leaves (the output parameters form PROSPECT model), biophysical canopy parameters (Leaf Area Index, brown pigment content, mean leaf inclination angle), soil brightness parameter, reflectance geometry (solar zenith angle, observer zenith angle, relative azimuth angle), ratio of diffuse to total incident radiation and two hot spot size parameters. The SAIL model is often combined with the model on leaf level – the PROSAIL model. The PROSPECT and SAIL are very rarely used to meadows, this kind of ecosystem is normally rather heterogeneous and modelling is quite difficult. In this study two Radiative Transfer Models (PROSPECT-5 and 4SAIL) were used on single leaves and a whole canopy level. In order to acquire the input data to both, models model and reference spectrums the fi eld measurements were done. The input parameters were recalculated using fields measurements and put into the models: PROSPECT and PROSAIL. Only one leaf structure parameter was fitted for each polygon individually. The spectral reflectance obtained from the model was compared with field data. Based on the calculated Root Mean Square Error the simulation was verified. The RMSE values were calculated for whole range from 400 to 2500 nm and for specific ranges. The correctness of simulated spectra were analysed dependent on the type of meadows (cultivated meadows with reduced amount of biomass, cultivated meadows with high amount of biomass and not cultivated meadows) and the value of three different biophysical parameters (Leaf Area Index, fresh biomass content and water content). Better results were obtained using PROSPECT model than PROSAIL. In the visible light more accurate values were calculated using PROSAIL and in the infrared using PROSPECT. Generally bigger errors were noticed in the infrared, especially middle infrared range. The effectiveness of the reflectance simulation was not influenced by different kind of meadows. Apart from that, better results were obtained on meadows with higher biomass value, bigger Leaf Area Index and lower water content. Generally, the PROSPECT and PROSAIL radiative transfer models can be used to simulate the spectral reflectance of vegetation on heterogeneous meadows. The models can be used to estimate the biophysical parameters, but it is necessary to correct the values of input variables (especially water content). Meadows are very complex environment and some of the parameters should be adjusted.

2

Analiza zależności między zawartością wody w roślinach zmierzoną w terenie a teledetekcyjnymi wskaźnikami roślinności

Niedzielko J., Milczarek M., Szepietowska M., Pokrzywnicka M., Boral B., Łach G., Kaźmierczak M., Jarocińska A.

Teledetekcja Środowiska

|

2012

|

T. 47

43--57

EN

Monitoring the plant moisture has a significant role in geographical research. It may be used, among the others, for climate modelling, agricultural predicting, rational water management, drought monitoring and determining vulnerability to the occurrence of the fire. Traditional methods, based on field measurements, are the most accurate, but also time-consuming. Therefore these methods can be applied only in a limited area. In order to explore bigger areas remote sensing methods are useful. To analyse plant condition and water content vegetation indices can be used. Their calculations are based on the reflectance in different bands. Despite many studies conducted on the development of remote sensing indices, still there is a need for verification of their accuracy and usefulness by comparing the results obtained through remote sensing tools with the results of field measurements. In this paper three indices are used: Moisture Stress Index (MSI), Normalized Difference Infrared Index (NDII) and transformation Tasseled Cap (the Wetness band). The aim of this study was to compare the value of vegetation indices calculated using images from Landsat 5 Thematic Mapper with the results of field measurement from five test areas of different type of land cover: cereal crops, non-cereal crops, forests, meadows and pastures. Research was carried out in province Ontario (Canada) and consisted of two stages. The first stage was the fi eld measurements, where the specified number of plant samples was collected and water content was calculated. The second stage consisted of the preparation of relevant satellite images (atmospheric correction and making the mosaic) and the calculation of vegetation indices. The study has shown, that statistical relationships between data sets obtained through remote sensing indices and calculated on the basis of field measurements are diverse for different indices. MSI and NDII values are significantly correlated with the water content in plants (R= -0.62 and 0.56, respectively). The correlation of TCW was rated as moderate (R=0.30). Spatial distribution of water content based on maps created using NDII and MSI is similar. It was noticed that TC Wetness transformation overestimates water content in cereal plants (smaller water content) and underestimates it in natural green plant ecosystems, which generally have higher water content. As a result, the range of water content values obtained from TCW is more narrow (dominates the class of 60-70% water in plants) than the range of values calculated using NDII and MSI. Both indices have more uniform distribution dominated by the classes of moderate water content (50-60%), rather wet plants (60-70%) and very wet plants (70-80%). Each index is characterized by different distribution of the water content. In general values calculated on the basis of NDII and MSI are higher than calculated using TCW. In order to perform more accurate analysis between values calculated using satellite images and the results of field measurements, the values of particular types of land cover should be compared.

3

Modelowanie charakterystyk spektralnych heterogenicznych zbiorowisk trawiastych przy użyciu modelu transferu promieniowania

Jarocińska A.

Teledetekcja Środowiska

|

2011

|

T. 46

29--42

EN

Meadows are important ecosystems and should be protected. Also, in Poland organic agriculture and farming, where crops from meadows are used, is getting more popular. That is why meadows monitoring and predicting crops is important issue. Much information can be calculated from spectrum of plants and that is why remote sensing data are very useful tool. Two approaches are used to calculate biophysical variables: statistical and modelling. In statistical, values from field measurements have to be compared with images. In modelling, radiative transfer models are used. RTM are physical models based on the fundamental equation of radiative transfer. After all necessary adjustments, models can give the description of the canopy with fewer field measurements. In this paper model on leaf level was chosen. PROSPECT uses only five input variables: chlorophyll and carotenoid content, water content, dry matter and leaf structure parameter. Model is normally used to homogeneous canopy, like corn. In this paper, PROSPECT was used to simulate spectrum for heterogenic meadows using field measurements. Biophysical variables were collected during field measurements in the Bystrzanka catchment in the Low Beskid Mountains. In the same time more than 10 samples of spectrum were collected using ASD FieldSpec 3 FR and then averaged. The minimum size of polygon was 100m2. All input parameters for every polygon were included into the model and spectrum was modelled. Then spectrum was compared with measured samples of each polygon. In the end the vegetation indices were calculated using two kinds of spectrum and compared. All used vegetation indices are describing plant condition or crop monitoring: Normalized Difference Vegetation Index, Red Edge Normalized Difference Vegetation Index, Photochemical Reflectance Index, Normalized Difference Nitrogen Index, Normalized Difference Lignin Index, Cellulose Absorption Index, Carotenoid Reflectance Index, Water Band Index and Moisture Stress Index. Researches shows, that it is possible to simulate spectra for heterogeneous meadows using PROSPECT. The average RMSE value for all polygons was 0,0346, which mean the spectra are well modelled. The biggest mistake was for near infrared range, where is the strongest influence of dry matter content. The differences between measured and modelled spectrum were also noticed on the part of visible light – 400-500nm. For most calculated vegetation indices values were similar for both kinds of spectra. Values of NDVI,WBI and NDLI were very close. The biggest differences were noticed form PRI and CRI.

4

Zmienność wilgotności w dolinie górnej Narwi w okresie 20 lat na podstawie transformacji Tasseled Cap i wskaźników wilgotności

Jarocińska A., Nasiłowska S.

Teledetekcja Środowiska

|

2009

|

T. 41

51--57

EN

Wetness monitoring is very important issue especially on wetlands ecosystems, because they are very vulnerable to changes, particularly those made by human. The upper Narew valley with eminence was analyzed. Described area is in north-eastern Poland and covers the valley from Tykocin to Łazy. This area is unique wetland habitat in Europe. In natural part is an anastomosing river system, whereas second part is covered by agricultural areas (wetlands which were drained in ‘70 of XX century). The aim of this paper is to demonstrate quantitative multitemporal analyses of changes in this environment by using various wetness indices and comparing them. To investigate the amount of changes the images from Landsat were used: from TM and ETM+ scanner (available from http://glovis.usgs.gov/). They were from two time series: the end of XX century (1989, 1992, 1993 and 1994) and the beginning of XXI century (2006 and 2007). All of the images were from the beginning or the middle of the vegetation season. In addition, meteorological data were used (from www. tutiempo.es), to detect the precipitation influence on analyzed indices. NDVI was calculated using image from the 2006, then the mask was created to remove all apart from the vegetation (everything under 0,4). After that the Tasseled Cap transformation was made to obtain Wetness band (TCW). Values under -37 on image from 1993 were masked to eliminate cloudy areas. In next step two wetness indices were calculated: Normalized Difference Infrared Index (NDII) and Moisture Stress Index (MSI). TCW is based on visual, near-infrared and-middle infrared electromagnetic radiation, because of that it could depend on atmospheric conditions. NDII and MSI are calculated only from 4th and 5th Landsat bands. Scattering from aerosols in that part of wavelength is weaker and doesn’t have big impact on indices values. Three describing indices are used when atmospheric correction isn’t possible or needed. Values of the three parameters were mapped by dividing into four classes: higher, medium, lower and the lowest wetness. Maps were averaged in the two time series (end of XX and beginning of XXI century). They were reclassified into tree difference maps to show the differences in wetness conditions and between various indices. Three maps showing changes in wetness were classified into five categories: much more wet, more wet, no changes, drier and much drier. These set of data could be compared. The results show that about 55% of analyzing area is stable. Table 3 present that about 2% of all changes were big. About 30% of total amount of transformation are connected with drainage areas. Areas which were more wet cover about 10%. Drained areas are getting extremely wet based on TCW, but opposite tendency can be noted on MSI and NDII maps. Big discrepancy between the maps of changes was discovered. TCW showed that the natural valley is getting drier and eminences are getting wet, but the results are different for the other two analyzed indices. Apart from that, some of the results are different for the parameters. In further research this kind of analysis should be compared with land cover and field measurements.

5

Remote sensing tools for analyzing state and condition of vegetation

Jarocińska A., Zagajewski B.

Roczniki Geomatyki

|

2009

|

T. 7, z. 2

47-53

EN

Hyperspectral data, which are characterized by very high spectral, spatial and radiometric resolutions, allow the analysis of the biometric properties of plants in different wavelengths of the electromagnetic spectrum. This kind of data can be applied to interpretation of vegetation, land cover forecast biomass and crops and also for analyzing plant condition, because vegetation cover is a very good indicator of environmental condition. All the spectral characteristics of plants can be measured and analyzed quantitatively using different vegetation indices, which are a mathematical combination of various bands. The most frequently used regions of the spectrum are visible, red-near infrared edge, near and middle infrared. In these regions it is possible to measure chlorophyll, carotenoids and other pigment content, fresh and dry biomass, water and nutrient content, internal leaf structure, soil moisture and plant surface temperature. In this study, four of the vegetation indices have been analysed: Normalized Difference Vegetation Index (Rouse et al., 1973; Griffith et al., 2002), Soil Adjusted Vegetation Index (Huete, 1988), Leaf Area Index (Surlock, 2001; Haboudane et al., 2004) and fAPAR - fraction of Absorbed Photosynthetically Active Radiation (Moreau, Li, 1996). These indices measure the condition of plants and estimate the quantity of biomass. Correctly calculated indices offer much information about the functionality of an ecosystem. Such vegetation indices are broadly used for vegetation monitoring. The main purpose of the research was an analysis of plant condition using remote sensing methods. Maps of spatial distribution of the NDVI, SAVI, LAI and fAPAR were prepared using ground and airborne measurements (DAIS 7915 products were corrected and verified by field measurements). Indices from airborne and ground level measurements were also correlated. The studies took place in the Low Beskid Mountains., which constitute one of the most natural parts of the Polish Carpathian Mountains (Fig. 3). The area extends from 49o34'- 49o41'N to 21o01'-21o09'E, with an altitude range of 400-750 m. The study area focuses on the Bystrzanka catchment around the town of Szymbark. This catchment has an area of around 13.5 km2. The largest part of the area, 40%, is covered by forest. Meadows and pasture comprise 28% of the area. A small fragment of the area is covered by arable land. The area is defined as a natural and seminatural environment. The human influence is relatively low and natural processes are not disturbed, so that vegetation can be used here as an indicator of other ecosystem components (soils, microclimate etc.)

PL

Techniki teledetekcyjne umożliwiają prowadzenie monitoringu przyrodniczego roślinności, w tym dokładną analizę fizjologii oraz właściwości biometrycznych. W artykule przedstawiony jest sposób badania kondycji roślinności wykorzystujący teledetekcyjne wskaźniki roślinności oraz związki mię dzy wskaźnikami mierzonymi z poziomu naziemnego i pułapu lotniczego. Badania były prowadzone na terenach naturalnych i ekstensywnie wykorzystywanych rolniczo zlewni Bystrzanki w Beskidzie Niskim. W badaniach wykorzystano dwa rodzaje danych: wartości wskaźników NDVI, SAVI, LAI i fAPAR pobranych na poziomie terenowym oraz obraz hiperspektralny ze skanera lotniczego DAIS 7915. Pobrano dane z poziomu terenowego. Następnie utworzono obrazy wskaźników w dwóch progra mach ATCOR i ENVI 4.3 (obraz wskaźnika NDVI). Obrazy wskaźników SAVI, LAI i fAPAR uzyskane z pierwszego programu były w jednostkach niezgodnych dla wskaźników, dlatego wymagały dalszych transformacji. Pobrano wartości wskaźników z obrazów. Następnie przeprowadzono analizy staty styczne porównując wartości z obrazów z danymi terenowymi, uzyskując równania regresji, których użyto do transformacji obrazów. Ostatnim etapem było utworzenie map przestrzennego rozkładu czterech wskaźników oraz mapy kondycji roślinności biorącej pod uwagę wartości wskaźników SAVI, LAI i fAPAR. Stwierdzono, że użycie teledetekcyjnych wskaźników roślinności ułatwia pozyskiwanie informacji o stanie roślinności i obiektywizuje te dane. Zanotowano korelacje między wskaźnikami NDVI i LAI oraz NDVI i fAPAR, są one zdecydowanie silniejsze na poziomie lotniczym. Na ścisłość korelacji wpływa sposób pobierania danych oraz sposób użytkowania terenu. Techniki hiperspektralne stwa rzają dodatkowe możliwości pozyskiwania informacji przez analizę krzywej odbicia spektralnego, a nie jedynie jej wycinków, tak jak w przypadku technik wielospektralnych. Wykorzystując tak utworzo ne wskaźniki możliwa jest dokładniejsza analiza roślinności. Stwierdzono, że na badanym terenie wskaźniki NDVI, SAVI, LAI i fAPAR mają wysokie wartości. Na podstawie mapy kondycji roślinności stwierdzono, że na przeważającym obszarze roślinność była w dobrym stanie.

6

Korelacje naziemnych i lotniczych teledetekcyjnych wskaźników roślinności dla zlewni Bystrzanki

Jarocińska A., Zagajewski B.

Teledetekcja Środowiska

|

2008

|

T. 40

100--124