Wykorzystanie metod geostatystycznych w badaniach środowiska przyrodniczego

Jarosław Zawadzki


The goal of geostatistics is to describe, analyze, and interpret uncertainty caused by a limited spatial sampling of a property under study. The classical statistical methods are not appropriate for this purpose because they ignore the spatial information in data sets. Geostatistics offers a variety of tools for describing spatial continuity, which is an essential feature of many phenomena. Despite constant advances in geostatistics and its many successful applications, it remains a relatively unknown field in Poland. At the same time, the application of geostatistical methods has extended intensively in other countries to many fields relating to geological and environmental studies as well as economics and sociology.This work presents the capacity to use geostatistical methods in environmental engineering and environmental protection research. The introductory part of the work presents the basic concepts, the most important geostatistical tools such as semivariogram, covariance function, correlogram, cross-se-mivariance, etc, and essential geostatistical methods. The modeling of variograms using positive definite functions is then described. The first part presents the most important type of spatial estimation method, which is known as kriging. I then explain cokriging, which is the most important tool of spatial estimation taking into account secondary information. This type of estimation is often called data integration. The cokriging method, which uses more than one variable for prediction, is particularly useful when primary variable information is sparse, and at the same time the secondary variables are plentiful or even exhaustive. Cokriging is ideally suited to spatial estimation in many fields of environmental studies such as geology, geochemistry, remote sensing, etc, where expensive or difficult measurements are required. This technique significantly improves estimation and reduces the variance of estimation error. A large number of environmental phenomena may be regarded as realizations of space-time random fields. Phenomena such as the greenhouse effect, ozone depletion in the upper atmosphere, atmospheric pollution, forest inventory and rainfall are spatial-temporal in nature. Therefore, at the end of the theoretical part, space-time models have also been briefly described The empirical part of the work presents the use of geostatistical methods for different types of environmental studies: 1. Spatial correlation analysis of vestigial element concentrations in the soil of Warsaw and its surroundings. 2. Spatial distribution analysis of Zn, Pb and Cu contamination in the soil of Warsaw and its vicinity. 3. Use of the indicative goodness of fit parameter (IGF) to decrease the number of sample points in periodical field measurement. 4. Cokriging-based analysis of underground water contamination on the foreground of the ,,Zelazny Most" reservoir. 5. Determination of semivariances calculated from Landsat TM remote sensing images of forested areas. 6. Classification of forest ecosystems using satellite imagery. Some important conclusions and results obtained from the above-listed studies are: A classical direct correlation analysis between the measured values, which are regionalized variables, such as the concentrations of elements in soil and texture of the remote sensing image, is not very useful. This is due to the omission of structured aspects of natural phenomena. Therefore, the Pearson coefficient of correlation is not an appropriate tool for the assessment of correlation among vestigial element concentrations in soil, and calculation of geostatistical measures of spatial continuity is necessary. The use of classical interpolation methods for such data such as e.g. inverse-distance weighing or least-square methods leads to non-realistic distributions of pollution. Indicator kriging, which is a sophisticated, non-parametric geostatistical method of spatial estimation, takes into account spatial correlation, and as such does not require the assumption of global stationarity or normal (log-normal) distribution of data. Probability maps showing the risk that concentration of studied metals exceeds given thresholds calculated using indicator kriging help make reasonable decisions on the extent of clean-up action. . The variogram modeling and analysis of the IGF parameter allow for a decrease in the number of sampling points in periodical field measurements without significant deterioration of the quality of obtained results. Geostatistical methods are a very useful tool both in planning complementary measurement campaigns and in an analysis of environmental pollution data. The analysis of spatial distribution of underground water contamination at the eastern foreground of the ,,Zelazny Most" reservoir takes advantage of additional geoelectrical measurements and use of the cokriging method. Maps made using cokriging are more precise in comparison with those madeusing the IDW method and kriging, which both take into account only one variable. Making use of geoelectrical measurements as an additional variable causes the shape and size of those areas considered contaminated to change significantly. It has important influence on ecological risk estimation. The texture of remote sensing images of the forests can be highly complex, but at the same time it is very useful in the interpretation of these images. The analysis concerned the differences in semiva-riances calculated from transects of the Landsat TM images associated with images of young, middle- -aged, old, and natural versus planted stands, as well as compared semi variances for loblolly pine and longleaf pine in Georgia, USA. The semivariances for loblolly pine depend on age as well as on the forest stand origin (natural vs. planted). Important differences for sill value as well as for semivariance values for the distances of a few lags have been found. These parameters have proved useful for forest stand discrimination. The role of geostatistical methods in analyzing remote sensing images of forest areas have increased rapidly over the last few years following advancements in high-resolution remote-sensor technology. The results of applications of geostatistical methods for remote sensing classification of forest ecosystems are very encouraging.
Book typeMonograph
Author Jarosław Zawadzki (FEE / DISEQR)
Jarosław Zawadzki,,
- Department of Informationa Science and Environment Quality Research
Publisher name (outside publisher list) Oficyna Wydawnicza Politechniki Warszawskiej
Publishing place (Publisher address)00-644 Warszawa, ul. Polna 50
Issue year2005
Book series /Journal (in case of Journal special issue)Prace Naukowe Politechniki Warszawskiej. Inżynieria Środowiska, ISSN 1234-4338
Keywords in Polishbadanie środowiska przyrodniczego, metody geostatystyczne, techniki satelitarne
Abstract in PolishW pracy zaprezentowano szeroko możliwości zastosowania metod geostatystycznych w badaniach środowiska przyrodniczego. Metody geostatystyczne zostały wykorzystane w szczegółowych badaniach korelacji przestrzennych zawartości wybranych pierwiastków śladowych w glebach Warszawy i okolic oraz w analizie rozkładu przestrzennego zanieczyszczenia tych gleb cynkiem, miedzią i ołowiem przy zastosowaniu krigingu wskaźnikowego. Wynikami tych obliczeń są mapy rozkładów przestrzennych zawartości badanych pierwiastków oraz mapy prawdopodobieństwa przekroczenia wybranych wartości progowych zanieczyszczenia. Takie rozkłady stanowią precyzyjną i wygodną informację dotyczącą zanieczyszczenia badanego obszaru. Metody geostatystyczne pozwalają na znaczną redukcję kosztów badań środowiska naturalnego. Zaprezentowano to na przykładzie wykorzystania własności wskaźnika dopasowania IGF w celu ograniczania liczby punktów opróbowania w cyklicznych kampaniach pomiarowych, stosując dane dotyczące zanieczyszczenia gleb. Warszawy i okolic. Ograniczenia kosztów badań środowiska i możliwość uzyskania z tych samych danych bardziej precyzyjnych informacji wynikają również z faktu, że metody geostatystyczne umożliwiają wykorzystywanie i integrację różnorodnych typów informacji przestrzennej. Przedstawione w pracy badania rozkładu zanieczyszczeń na przedpolu zbiornika "Żelazny Most" przy użyciu metody kokrigingu wykazały możliwość połączenia tanich i wygodnych pomiarów elektrooporowych z pomiarami chemicznymi pochodzącymi ze statycznego i kosztownego systemu monitoringu składającego się z systemu studni drenażowych i piezomet-rów przystudziennych. Na przykładzie badań rozkładu zanieczyszczeń na przedpolu zbiornika "Żelazny Most" pokazano również, że wykorzystanie pomiarów geooporowych jako zmiennej dodatkowej powoduje zmianę zarówno kształtu, jak i rozmiaru obszarów, które można uznać za zanieczyszczone, co ma kluczowe znaczenie dla oceny rzeczywistego zagrożenia ekologicznego. Inny obszar zastosowań metod geostatystycznych przedstawiono na przykładzie badań lasów za pomocą technik satelitarnych. W pracy omówiono wykorzystanie metod geostatystycznych do klasyfikacji ekosystemów leśnych przy użyciu technik satelitarnych, jak również zaprezentowano semiwarian-cje tekstury zdjęć satelitarnych obszarów leśnych sondy Landsat TM.
Score (nominal)12
Publication indicators GS Citations = 13.0
Citation count*13 (2020-09-26)
Share Share

Get link to the record

* presented citation count is obtained through Internet information analysis and it is close to the number calculated by the Publish or Perish system.
Are you sure?