eingereicht von

• Chapter I   Introduction

  • Chapter I:1  Changes in the earth system and the role of land use

  • Chapter I:2 Post-socialist land use change in Eastern Europe

  • Chapter I:3 The Carpathian Mountains

  • Chapter I:4 Study Area & Research Questions

  • Chapter I:5 Approach & Specific Objectives

  • Chapter I:6 Structure of this thesis

• Chapter II   Cross-border comparison of land cover and landscape pattern in Eastern Europe using a hybrid classification technique

  • Chapter II:1  Introduction

  • Chapter II:2 Study area

  • Chapter II:3 Data and Methods

    • Chapter II:3.1  Satellite and field data

    • Chapter II:3.2 Preprocessing of Landsat data

    • Chapter II:3.3 Classification

    • Chapter II:3.4 Landscape structure

  • Chapter II:4 Results and Discussion

    • Chapter II:4.1  Land cover classification

    • Chapter II:4.2 Cross-border comparison of land cover and landscape pattern

      • Chapter II:4.2.1 Forests

      • Chapter II:4.2.2 Arable land, grassland, and shrubland

  • Chapter II:5 Conclusions

• Chapter III   Post-socialist forest disturbance in the Carpathian border region of Poland, Slovakia, and Ukraine

  • Chapter III:1  Introduction

  • Chapter III:2 Study area

  • Chapter III:3 Data and methods

    • Chapter III:3.1  Datasets used

    • Chapter III:3.2 Preprocessing of Landsat TM and ETM+ data

    • Chapter III:3.3 Forest disturbance detection

      • Chapter III:3.3.1 Separating forested and non-forested areas for 1988

      • Chapter III:3.3.2 Detecting forest disturbances for the period 1988-2000

      • Chapter III:3.3.3 Forest type stratification for changed areas

    • Chapter III:3.4 Forest fragmentation

  • Chapter III:4 Results

  • Chapter III:5 Discussion

    • Chapter III:5.1  Comparison of post-socialist forest disturbance rates among countries

    • Chapter III:5.2  Forest disturbances inside and outside protected areas

    • Chapter III:5.3 Comparison of forest disturbance rates and official statistics

    • Chapter III:5.4 Accuracy of the forest disturbance detection

  • Chapter III:6 Conclusions

• Chapter IV  Cross-border comparison of post-socialist farmland abandonment in the Carpathians

  • Chapter IV:1  Introduction

  • Chapter IV:2 Study Area 

  • Chapter IV:3 Datasets Used and Methods

    • Chapter IV:3.1  Datasets Used

    • Chapter IV:3.2 Mapping farmland abandonment using SVM change detection

    • Chapter IV:3.3 Cross-border comparison of farmland abandonment

  • Chapter IV:4 Results

  • Chapter IV:5 Discussion

    • Chapter IV:5.1  Mapping farmland abandonment using SVM

    • Chapter IV:5.2 Farmland abandonment in the border region of Poland, Slovakia, and Ukraine

  • Chapter IV:6 Conclusions

• Chapter V   Using image texture to map farmland field size in Eastern Europe

  • Chapter V:1  Introduction

  • Chapter V:2 Study area

  • Chapter V:3 Data and Methods

    • Chapter V:3.1  Datasets used

    • Chapter V:3.2 Sample of fields

    • Chapter V:3.3 Texture measures

    • Chapter V:3.4 Statistical analysis

    • Chapter V:3.5 Applying the model to images

  • Chapter V:4 Results

    • Chapter V:4.1  Statistical models

    • Chapter V:4.2 Applying the model to images

  • Chapter V:5  Discussion

    • Chapter V:5.1  Mapping parcelization using texture measures

    • Chapter V:5.2 Field size in the border triangle of Poland, Slovakia, and Ukraine

• Chapter VI   Synthesis

  • Chapter VI:1  Summary and main conclusions

  • Chapter VI:2 Future research

• Acknowledgments

• References

•  Mapping post-socialist parcelization of farmland in Eastern Europe using texture measures

  • Introduction

  • Data & Methods

  • Results & Discussion

  • Summary & Outlook

  • References

•  A method to detect and correct single-band missing pixels in Landsat TM and ETM+ data

  • Introduction

  • Background

  • Methods & Materials

    • Data

    • Methods

      • Stage I: Detection of distorted pixels

      • Stage II: Correction

  • Results and Discussion

  • Summary and Conclusions

  • References

• Table I1: Land ownership patterns and privatization strategies of the countries in the study area (Source: Lerman et al. 2004, modified).

• Table II1: Class scheme, class descriptions, classification method and training data for the hybrid classification (* H = hybrid classification; C = ISODATA clustering; KB = knowledge-based; **number of clusters)

• Table II2: Confusion matrix for the hybrid classification (UAC = user’s accuracy, PAC = producer’s accuracy, CKA = conditional kappa; acronyms are explained in Table 1)

• Table II3: Distribution of four fragmentation components per country for the land cover types forest, arable land and grassland. Fragmentation components were calculated for three differently sized neighborhoods for the forest class (2.25ha = 5 pixels; 7.29ha = 9 pixels; 65.61ha = 27 pixels) and for two differently sized neighborhoods (2.25ha and 7.29ha) for the land cover types arable land and grassland (rows may not sum to 100% due to rounding).

• Table III1: Error matrix for the forest disturbance detection (Values represent absolute numbers of ground truth plots; UAC = user’s accuracy [%]; PAC = producer’s accuracy [%]).

• Table IV1: Accuracy assessment of the change classification.

• Table IV2: Mean patch size (Mean), area-weighted mean patch size (AMean), patch density (PD), and aggregation index (AI) for the fallow farmland and reforestation classes of the Polish, Slovak, and Ukrainian region of the study area.

• Table V1: Regression models for different combinations of texture measures and window sizes for models that included mean texture (group I models). Best models for each subgroup (one-, two-, or three-dimensional) are in bold. Acronyms: #V = number of input variables, WS = window size, adjR² = adjusted R², BIC = Bayesian Information Criterion, #BM = number of equally good best models (i.e. difference in adj. R² < 0.02 to the absolute best model); Significance: p<0.0001=***, <0.001=**, <0.01=*, <0.05=a; b indicates cases were all coefficients remained significant after Bonferonni correction.

• Table V2: Regression models for different combinations of texture measures and window sizes for models that included mean and standard deviation texture (group II models). Best models for each subgroup (one-, two-, or three-dimensional) in bold. Acronyms: #V = number of input variables, WS = window size, adjR² = adjusted R², BIC = Bayesian Information Criterion, #BM = number of equally good best models (i.e. difference in adj. R² < 0.02 to the absolute best model); Significance: p<0.0001=***, <0.001=**, <0.01=*, <0.05=a; b indicates cases were all coefficients remained significant after Bonferonni correction.

• Table V3: Example of the number of times each texture measure was included in the series of regression models containing one (n=1), two (n=2 models), or three (n=33) covariates that performed equally well (i.e. diff. in adjusted R² <0.02) for mean texture of June 2000 (window size 3, total number of variables = 78). Acronyms: range (RA), 1st-order mean (M1), variance (VA), 1st-order entropy (E1), skewness (SK), 2nd-order mean (M2), sum of squares variance (SS), homogeneity (HO), contrast (CO), dissimilarity (DI), 2nd-order entropy (E2), angular second moment (SM), correlation (CR), near infrared band (NIR), short wavelength infrared bands (SWIR1, SWIR2).

• Table V4: Mean prediction errors of mean field size (log) for the one- two-, and three-dimensional group I (mean texture) and group II (mean and standard deviation texture) models. Cross-validation was carried out for the best models per subgroup (bold models in Table 1 and Table 2) using a leave-one-out strategy and a five-fold cross-validation approach.

• Table A1: Land ownership of agricultural land and privatization strategies of the countries in the study area (Lerman et al. 2004).

• Table B1: Landsat TM images used in this study and the number of pixels affected by missing pixels distortions.

• Table B2: Statistical comparison of original and corrected spectra of artificially introduced distortions (Min / Max / Mean = minimum, maximum and average deviation of original and corrected spectra; STD = standard deviation; n = sample size; CI_L / CI_U = lower and upper limits of confidence intervals for p < 0.01; all values are given in digital numbers).

• Figure I1: Simplified scheme of scale-dependency in land use decision making and the impacts of land use change on ecosystems. Underlying drivers of land use change operate at different scales and influence land use decision at the local level. This controls proximate causes which in turn affect ecosystems. Local processes may have strong global impacts when aggregated.

• Figure I2: Location of the Carpathian Mountains in Europe (altitudes range from approximately 50 to 2,650m, source: Shuttle Radar Topography Mission (SRTM) Digital Elevation Model, ESRI Data and Maps Kit).

• Figure II1: The border triangle of Poland, Slovakia and Ukraine, located in the north-eastern part of the Carpathian ridge (shaded SRTM relief).

• Figure II2: Top: Corresponding windows of the base map (shaded SRTM DEM) and raw image (ETM+ band 4) centered on a potential GCP. Bottom: Visualization of a plane of correlation coefficients calculated by correlating a 10x10 pixel-wide window centered on a potential GCP in the base map with all 10x10 sized windows within the subset of the raw image. A good GCP is represented by a high peak in the plane of correlation coefficients (x,y-axes: pixel position, z-axis: R).

• Figure II3: First principal component from 2000-06-10 before (left) and after (right) radiometric rectification and topographic correction.

• Figure II4: Classification scheme (for details compare to text; MLH = maximum likelihood classification, PCA = principal component analysis).

• Figure II5: Land cover map for the border triangle Poland, Slovakia, and Ukraine

• Figure II6: Comparison of land cover between the three countries. Top left: absolute area; Top right: proportion of land cover normalized by the total area of each country. Middle and bottom: proportions of land cover classes per altitudinal zone (acronyms are explained in Table II-1).

• Figure II7: Boxplot graphs of the distribution of elevation for each class and country (υ represent class medians; box determines the first and third quartile; whiskers represent upper and lower range, max/min values exceeding the range of ± 3 standard deviations (STD) were treated as outliers and the 3STD limit was taken instead; acronyms are explained in Table II-1).

• Figure II8: Maps of fragmentation components (left) and categorized proportions of PLC with the classes core, interior, dominant, and intermediate (normalized over the sum of these components; right). Results are based in a neighborhood size of 2.25ha for arable land and grassland and on a neighborhood size of 7.29ha for the forest class.

• Figure II9: Two-dimensional density distributions of logarithmized patch size [ha] and mean patch elevation [m] per country and for the land cover classes arable land, grassland and shrubland.

• Figure III1: Location of the study area in the Carpathian Mountain range. The study area harbors two protected areas, the trilateral East Carpathians Biosphere Reserve (ECBR) and the Skole Beskydy National Park (NP) in Ukraine (elevations range from about 100-1,700m; data sources: SRTM digital elevation model, ESRI Data and Maps Kit).

• Figure III2: Processing scheme for detecting forest disturbance in the study area (for details compare to section 3.2; BGW = Tasseled Cap brightness, greenness, and wetness; DI = disturbance index; mMLH = multitemporal maximum likelihood; DEM = digital elevation model).

• Figure III3: Forest disturbance map of the study area. The insets provide examples of disturbance patterns of the countries Poland (inset 1), Slovakia (2), Ukraine (3) and the Polish-Slovak border region (4).

• Figure III4: Yearly disturbance rates for the Polish, Slovakian, and Ukrainian region of the study area (Note: Disturbance rates before 1988 were referenced to a six year interval).

• Figure III5: Altitudinal distribution of total forest area (unchanged forest and disturbances) and disturbances for 1988, 1994, and 2000 for the three countries. (Distributions are normalized; g1 = skewness).

• Figure III6: Left: Distribution of disturbed forests among the forest types broad-leaved forest, mixed forest, and coniferous forest for disturbances mapped in 2000 and in 1994. Right: Forest fragmentation components for the years 1988 and 2000.

• Figure III7: Annual forest disturbance rates inside and outside protected areas per country and time period. Disturbance rates are given for the core zone (CZ), buffer zone (BZ), and transition zone (TZ) of the East Carpathian Biosphere Reserve, for the Skole Beskydy National Park (SK NP), and for areas outside of protected areas (Outside).

• Figure IV1: The border triangle of Poland, Slovakia, and Ukraine in the Carpathians. Farmland in the hatched region in Poland was mostly collectivized during socialism.

• Figure IV2: Farmland abandonment from 1986 to 2000 in the study area.

• Figure IV3: Comparison of fallow land and reforestation rates (1986/88 – 2000) among the Polish, Slovak, and Ukrainian portions of the study area.

• Figure IV4: Rates of fallow land and reforestation (1986/88 – 2000) by elevation class (50m elevation increase per class, histogram bars are stacked).

• Figure IV5: Rates of fallow land and reforestation (1986/88 – 2000) by slope class (5% slope per class; histogram bars are stacked).

• Figure IV6: Comparison of farmland abandonment rates (1986/88 – 2000) of lands managed by the state during socialism and lands that were never collectivized in the Polish region of the study area.

• Figure V1: Top: Study area in the border region of Poland, Slovakia, and Ukraine in the Carpathians. Bottom: Example of land use pattern in the three countries Poland, Slovakia, and Ukraine (Landsat Enhanced Thematic Mapper Plus image from 30th September 2000; band combination: red = band 4, green = band 5, blue = band 3).

• Figure V2: Example of an area characterized by heterogeneous land use pattern in high-resolution Quickbird data (left), June 2000 Landsat ETM+ data (1st principal component, middle), and image texture derived from the Landsat image (1^st-order entropy of band 7 calculated at a window size of 3 pixels, right). The subset is centred on the village of Bezovce in Slovakia (21.15E, 48.63N).

• Figure V3: Examples of the relationship between prediction accuracy (R² of field size vs. texture measures) and window size used to calculate texture measures. Mean field size was estimated using mean texture (left) and standard deviation of texture (right).

• Figure V4: Field size map of the border region of Poland, Slovakia, and Ukraine for the year 2000. Top: map derived using the best two-dimensional mean texture model; Bottom: map derived using the best three-dimensional mean texture model. The three-dimensional map is shown using the color scheme of the two-dimensional map for better comparison. (Coordinate System: UTM / Zone 34N; Ellipsoid/Datum: WGS84)

• Figure V5: Distribution of field sizes for the Polish, Slovak, and Ukrainian region of the study area. Whiskers indicate the 90^th and 10^th percentiles.

• Figure V6: Distribution of field sizes per elevation zone and country. Boxplot whiskers extend to 1.5 times the interquartile range.

• Figure A1: Changes in the parcelization of farmland in the border region of Poland, Slovakia, and Ukraine (UTM reference system with WGS84 datum and ellipsoid).

• Figure B1: Three different kinds of distortions occurring in Landsat TM data. Left: only a single pixel is affected; middle: several pixels affected and random pattern of distorted pixels can be observed; right: distorted pixels are clustered and the affected show a detector pattern (all distortions occurred in a Landsat 5 TM image, acquired 4th July 1994).

• Figure B2: Three examples of distorted areas occurring in a Landsat TM image (left), flagged pixel values after the detection procedure (middle) and the corrected image after locating similar pixels using a spectral matching operation and then substituting the erroneous band values (right). Band combinations for red, green, and blue are: 3/2/1 (top), 4/5/3 (middle), and 4/3/2 (bottom).

• Figure B3: Examples of spectra with erroneous band values before and after the correction. The deviation between uncorrected and corrected spectra is substantial and the correction algorithm results in more useful spectra (for details refer to text). All spectra were taken from a Landsat (5) TM image acquired 4th July 1994.

• Figure B4: Comparison of raw image data, NDVI, and Tasseled Cap indices for uncorrected (left) and corrected (right) images. The missing pixel distortions are not visible in the corrected images, thus allowing for better visual interpretation. All operations were carried out on a subset of a Landsat (5) TM image, acquired 4th July 1994, that displayed single-band distortions.

• Figure B5: Change maps (no-change / change) for different change detection methods before (bottom) and after (top) the correction of single-band missing pixels. All analyses were carried out on the image from 7th June 1994 that had ~65,400 distorted pixels. Image ratioing (left), NDVI difference image (middle), and the disturbance index (right) were calculated for a forested region in the Ukrainian Carpathians. The forest change maps differ considerably for different change detection methods. However, the missing-pixels distortions appear as pseudo-change for all approaches when relying on uncorrected images. The correction method removes the pseudo-change from the change maps (for details refer to text).