QUANTIFYING THE CHANGES IN LANDSCAPE CONFIGURATION USING OPEN SOURCE GIS. CASE STUDY: BISTRITA SUBCARPATHIAN VALLEY, ROMANIA

; 10]) Name Abbrev iation Short description; Measure unit Land cover (Total surface) LC Equals the number of cells for each class based on a classified land cover matrix. The resulting values were multiplied by the cell’s value (in our case 5 meters); (ha). Landscape proportion LP The proportion of the cells from a specific class of the total number of cells of the classified raster; (%). Edge length EL Equals the total len gth of all patches from a specific class. The resulting values were, of course multiplied with the cell’s value; (m). Edge density ED The sum of the lengt hs of all edge segments involving the corresponding patch type, divided by the total landscape area; (m/ha) Patch number NP Express the number of patches identified for each class; (no.). Patch density PD Equals the number of patches of the corresponding patch type divided by total landscape area; (no./100 ha). Greatest patch area GPA The patch that sums up the highest number of cells; at the end, for showing the exact area, the number is multiplied by the value of the cell; (ha). Smallest patch area SPA The patch that sums up the smallest number of cells, the result being multiplied too; (ha). Mean patch area MPA The patch that represents the average number of cells identifie d in the entire area; (ha). Mean patch distance MPD Calculate the average Euclidean distance between all patches from the same class; (m). Landscape division D The possibility that two cells, randomly chosen form the landsc ape, to be found in the same patch; (0 ≤ D < 1). Effective mesh size m the probability that two randomly chosen cells are connected (to be included into the same patch); (ha). Splitting index S The number of patches one gets when dividing the total region into parts of equal size in such a way that this new configuration leads to the same degree of landscape division desired; (nr.). Section Ecology and Environmental Protection Tab. 4 – Landscape metrics computed for every patch type for 1986 Tab. 5 – Landscape metrics computed for every patch type for 1986 The role of the first indicator ( Total surface) is to measure the landscape composition (structure), quantifying the abundance of each class, without any reference to the spatial arrangement. The second indicator, landscape proportion , equals the fraction of a certain class from the total analyzed area. This indicator is similar to the previous one, but the percentage expression makes it easier to interpret by the users. Regarding the evolution of the total area of the 11 land cover classes, one can observe significant changes over the analyzed period (fig. 2 , 3), being highlighted the built-up area, arable land and also forests. Fig. 2: Land cover classes - 1986 Fig. 3: Land cover classes - 2006 14th SGEM GeoConference on Ecology, Economics, Education and Legislation The edge length and edge density has the same utility for the quantitative interpretation of the landscape, except that the second one reports edge length on a per unit area basis that facilitates the comparison among different sized landscapes

. By the analysis of the tables can be observed a decrease of the two indexes values for most classes, suggesting a tendency of landscape homogenization. Regarding the evolution of the patch number of each class can be noticed a significant decrease in built-up area values, from 436 to 195, which shows the agglutination of the isolated areas into larger residential areas. Also, the patch number of the arable land decreased from 436 to 342 following the tendency of association of the farmers during the last few years and hence, the transition from a subsistence agriculture to a more productive one. The same tendency can be observed in the case of grasslands, which have reduced the patch number from 546 to 384. The plots cultivated with v ineyards and orchards suffered significant cuts, not due to the increase in their surfaces, but mainly due to their abolition in favor of arable land. A different situation is recorded for the forests class, whose patch number increased significantly. The explanation lies in afforestation works during recent years and the transformation of the shrubs into forests. These facts can be explained following also the values of edge density. Edge density, with patch number and patch density are representative for establishing the fragmentation degree of the landscape. The values obtained for the fragmentation (NP and, consequently, PD and ED) reveal a decrease in the study area’s fragmentation degree, inducing a clustering tendency

. The degree of landscape fragmentation is an important environmental indicator in the fields of biodiversity and sustainable development. In addition, information on the degree of landscape fragmentation is relevant in regional planning and for decisions about infrastructure placement or removal. Its analysis on different time series show how strong the current trends are and what their direction is

. The greatest patch area is related to the degree of homogeneity of the landscape. As can be seen from the table, the values for 1986 are directly proportional to 2006, indicating a certain stability. Therefore, can be noticed the built-up area, with 1243, respectively 2138 ha, arable land with 8340, respectively 5292 ha, or forests, with 8178, respectively 8472 ha. Mean patch area is also higher for the categories mentioned above, while the smallest patch area show quite low values, ranging from 0,0025 ha to 6,25 ha, which contributed to the landscape’s fragmentation. Calculating mean patch distance has an important role in determining the degree of their isolation in the landscape. The higher the resulting values are, the higher is the degree of isolation of a certain class with implications on func ti onal relations of the system. The last three indicators (D, M and S) are interconnected and measure the fragmentation degree of the landscape. These indicators were introduced by

, as a result of the criticism over the simple measurements such patch number or patch density, which presents some limitations for certain phases of fragmentation process. According to

, this suite of metrics derives from the cumulative distribution of the patches and provides a series of alternative measures and more explicit on the landscape’s subdivisions. They have the advantage, unlike other conventional indicators , that any omissions or additions of other small sized patches does not influence the final result. Section Ecology and Environmental Protection The values of landscape division may vary from 0 to 1, the value 0 being recorded when the landscape is represented by a single patch, approaching to 1 when each patch of the chosen class has an area of one cell. Because the raster was created on a 5 meters resolution, the values close to 1 of the D are explicable. However, one can clearly observe the variations of the indicator between the two intervals. Alth ough landscape division and Mesh are perfectly correlated, but inversely, both metrics are included because of the differences in units and interpretation. Split is based on the cumulative patch area distribution, and is interpreted as the effective mesh number, or number of patches with a constant patch size when the corresponding patch type is subdivided into S patches, where S is the value of the splitting index

defines the splitting index as the number of patches resulted after dividing the total area into equal size parts so that this new configuration leads to the same degree of landscape division (D). When its value is 1, the landscape is represented by a single patch, the value increasing as the landscape is divided into several patches. Considering these aspects, the results interpretation must take into account the correlation of these three complementary indicators. The resulting values of the three indicators suggest different degrees of fragmentation for each class. Thus, the areas with a high degree of homogeneity are represented by arable land, forests and water bodies, having significant variations during the analyzed period. If arable land and water bodies records for 2006 a decrease of the MESH (m) index from 1251 ha to 5788 ha, to 700 ha, respectively 2876 ha, the forests keep the previously recorded trend, showing an increase of the index from 1271 ha, to 1452 ha, which indicates a slight increase in the degree of homogeneity. Of particular importance is class 1 (built-up area), which, in 2006 reduced its fragmentation degree due to a continued expansion of the constructed areas and by incorporating isolated constructions (MESH values increased from 65231 ha to 152356 ha). These facts can also be explained using the other two indicators, due to their complementarities. CONCLUSIONS In summary, one can say that landscape metrics vary with varying landscape attributes; correlate highly with one another and often provide redundant information — (which is not surprising, given they derive from a rather small set of possible attributes: area, border or edge length, distance, that one can measure), and relate differently according to the process under investigation. These results should not be surprising. The fragmentation of the landscape is a complex process that acts on a complex system and results in a wide arrangement of spatial patterns

. Altogether the results obtained in the study show the usefulness of global indicators in landscape change modeling and that this type of landscape analysis becomes increasingly important for the local actors, which must take decisions in agreement with the landscape potential of each region

. Regarding the study area landscape, there is a predominance of specific categories (built-up area or arable land), which have expanded over others, putting a high pressure over the natural conditions and local biodiversity. The latest evolution of Open Source GIS software, especially of those who can offer integrated desktop solutions such as QGIS, had created many opportunities for the development of new tools for spatial analysis. Besides of numerous plug-ins designed to ease the work of the researchers one of the most prominent feature of QGIS is it's 14th SGEM GeoConference on Ecology, Economics, Education and Legislation flexibility and capacity to interact with other packages, and by doing this the array of analyses that can be performed in spatial domain are limitless

. REF ERENCES

Pătru-Stupariu, I., Stupariu, M. S., Cuculici R. & Huzui, A. Application of the global indicators to landscape change modeling on Prahova Valley (Romanian Carpathians and Subcarpathians), International Journal of the Physical Sciences, vol. 6(3), pp. 534-539, 2011;

Pătroescu, Maria, Toma, Simona, Sasaki, L. And Apostol, G. Priorities in the rehabilitation and re-naturation of rural landscape of the Roumanian Plain, southern Romania, Analele Universităţii Bucureşti – Geografie, XLIX, 2000;

Farina A. Principles and methods in Landscape Ecology. Towards a Science of Landscape, Springer, 412 p, 2006;

Turner M.G. & Garnder R.H. (edit.). Quantitative methods in landscape ecology, Springer, New York, 1991;

McGarigal K. And Marks B. FRAGSTATS: Spatial pattern analysis program for quantifying landscape structure. General Technical Report PNW-GTR-351. USDA Forest Service, Pacific Northwest Research Station, Portland, Oregon, USA, 1995;

Tur ner M., Gardner R., O’ Neill R. Landscape Ecology in Theory and practice. Pattern and process, Springer, 403 p, 2001;

McGarigal, K., SA Cushman, Mc Neel, & E Ene. - FRAGSTATS v3: Spatial Pattern Analysis Program for Categorical Maps. Computer software program produced by the authors at the University of Massachusetts, Amherst. http://www.umass.edu/landeco/research/fragstats/fragstats.html , 2002.;

Perrera A.H., Baldwin D.J.B. And Schnekenburger F. LEAP II: A Landscape Ecological Analysis Package for Land Use Planners and Managers. Forest Research Report no. 146, Ontario, Canada, 1997;

Jaeger J.A.G. Landscape division, splitting index, and effective mesh size: new measures of landscape fragmentation. Landscape ecology 15.2, pp. 115-130, 2000;

Rutledge D. Landscape indices as measures of the effects of fragmentation: can pattern reflect process , Doc Science Internal Series 98, 27 p, 2003;

Jaeger J.A.G., Esswein H., Schwarz-Von Raumer H.G. Measuring Landscape Fragmentation with the Effective Mesh Size meff, Zurich, http://gpe.concordia.ca/documents/faltblatt_engl.pdf , 2006;

Roșca B., Chelaru D. -A., Pleșcan S. The Analysis Of Landscape Morphology In Lower Bistrita Valey Using Grass And Quantum Gis, 13th International Multidisciplinary Scientific GeoConference SGEM 2013, Conference Proceedings - Informatics, geoinformatics and remote sensing, pp. 951 – 959, 2013.

; 10]) Name Abbrev iation Short description; Measure unit Land cover (Total surface) LC Equals the number of cells for each class based on a classified land cover matrix. The resulting values were multiplied by the cell’s value (in our case 5 meters); (ha). Landscape proportion LP The proportion of the cells from a specific class of the total number of cells of the classified raster; (%). Edge length EL Equals the total len gth of all patches from a specific class. The resulting values were, of course multiplied with the cell’s value; (m). Edge density ED The sum of the lengt hs of all edge segments involving the corresponding patch type, divided by the total landscape area; (m/ha) Patch number NP Express the number of patches identified for each class; (no.). Patch density PD Equals the number of patches of the corresponding patch type divided by total landscape area; (no./100 ha). Greatest patch area GPA The patch that sums up the highest number of cells; at the end, for showing the exact area, the number is multiplied by the value of the cell; (ha). Smallest patch area SPA The patch that sums up the smallest number of cells, the result being multiplied too; (ha). Mean patch area MPA The patch that represents the average number of cells identifie d in the entire area; (ha). Mean patch distance MPD Calculate the average Euclidean distance between all patches from the same class; (m). Landscape division D The possibility that two cells, randomly chosen form the landsc ape, to be found in the same patch; (0 ≤ D < 1). Effective mesh size m the probability that two randomly chosen cells are connected (to be included into the same patch); (ha). Splitting index S The number of patches one gets when dividing the total region into parts of equal size in such a way that this new configuration leads to the same degree of landscape division desired; (nr.). Section Ecology and Environmental Protection Tab. 4 – Landscape metrics computed for every patch type for 1986 Tab. 5 – Landscape metrics computed for every patch type for 1986 The role of the first indicator ( Total surface) is to measure the landscape composition (structure), quantifying the abundance of each class, without any reference to the spatial arrangement. The second indicator, landscape proportion , equals the fraction of a certain class from the total analyzed area. This indicator is similar to the previous one, but the percentage expression makes it easier to interpret by the users. Regarding the evolution of the total area of the 11 land cover classes, one can observe significant changes over the analyzed period (fig. 2 , 3), being highlighted the built-up area, arable land and also forests. Fig. 2: Land cover classes - 1986 Fig. 3: Land cover classes - 2006 14th SGEM GeoConference on Ecology, Economics, Education and Legislation The edge length and edge density has the same utility for the quantitative interpretation of the landscape, except that the second one reports edge length on a per unit area basis that facilitates the comparison among different sized landscapes

. By the analysis of the tables can be observed a decrease of the two indexes values for most classes, suggesting a tendency of landscape homogenization. Regarding the evolution of the patch number of each class can be noticed a significant decrease in built-up area values, from 436 to 195, which shows the agglutination of the isolated areas into larger residential areas. Also, the patch number of the arable land decreased from 436 to 342 following the tendency of association of the farmers during the last few years and hence, the transition from a subsistence agriculture to a more productive one. The same tendency can be observed in the case of grasslands, which have reduced the patch number from 546 to 384. The plots cultivated with v ineyards and orchards suffered significant cuts, not due to the increase in their surfaces, but mainly due to their abolition in favor of arable land. A different situation is recorded for the forests class, whose patch number increased significantly. The explanation lies in afforestation works during recent years and the transformation of the shrubs into forests. These facts can be explained following also the values of edge density. Edge density, with patch number and patch density are representative for establishing the fragmentation degree of the landscape. The values obtained for the fragmentation (NP and, consequently, PD and ED) reveal a decrease in the study area’s fragmentation degree, inducing a clustering tendency

. The degree of landscape fragmentation is an important environmental indicator in the fields of biodiversity and sustainable development. In addition, information on the degree of landscape fragmentation is relevant in regional planning and for decisions about infrastructure placement or removal. Its analysis on different time series show how strong the current trends are and what their direction is

. The greatest patch area is related to the degree of homogeneity of the landscape. As can be seen from the table, the values for 1986 are directly proportional to 2006, indicating a certain stability. Therefore, can be noticed the built-up area, with 1243, respectively 2138 ha, arable land with 8340, respectively 5292 ha, or forests, with 8178, respectively 8472 ha. Mean patch area is also higher for the categories mentioned above, while the smallest patch area show quite low values, ranging from 0,0025 ha to 6,25 ha, which contributed to the landscape’s fragmentation. Calculating mean patch distance has an important role in determining the degree of their isolation in the landscape. The higher the resulting values are, the higher is the degree of isolation of a certain class with implications on func ti onal relations of the system. The last three indicators (D, M and S) are interconnected and measure the fragmentation degree of the landscape. These indicators were introduced by

, as a result of the criticism over the simple measurements such patch number or patch density, which presents some limitations for certain phases of fragmentation process. According to

, this suite of metrics derives from the cumulative distribution of the patches and provides a series of alternative measures and more explicit on the landscape’s subdivisions. They have the advantage, unlike other conventional indicators , that any omissions or additions of other small sized patches does not influence the final result. Section Ecology and Environmental Protection The values of landscape division may vary from 0 to 1, the value 0 being recorded when the landscape is represented by a single patch, approaching to 1 when each patch of the chosen class has an area of one cell. Because the raster was created on a 5 meters resolution, the values close to 1 of the D are explicable. However, one can clearly observe the variations of the indicator between the two intervals. Alth ough landscape division and Mesh are perfectly correlated, but inversely, both metrics are included because of the differences in units and interpretation. Split is based on the cumulative patch area distribution, and is interpreted as the effective mesh number, or number of patches with a constant patch size when the corresponding patch type is subdivided into S patches, where S is the value of the splitting index

defines the splitting index as the number of patches resulted after dividing the total area into equal size parts so that this new configuration leads to the same degree of landscape division (D). When its value is 1, the landscape is represented by a single patch, the value increasing as the landscape is divided into several patches. Considering these aspects, the results interpretation must take into account the correlation of these three complementary indicators. The resulting values of the three indicators suggest different degrees of fragmentation for each class. Thus, the areas with a high degree of homogeneity are represented by arable land, forests and water bodies, having significant variations during the analyzed period. If arable land and water bodies records for 2006 a decrease of the MESH (m) index from 1251 ha to 5788 ha, to 700 ha, respectively 2876 ha, the forests keep the previously recorded trend, showing an increase of the index from 1271 ha, to 1452 ha, which indicates a slight increase in the degree of homogeneity. Of particular importance is class 1 (built-up area), which, in 2006 reduced its fragmentation degree due to a continued expansion of the constructed areas and by incorporating isolated constructions (MESH values increased from 65231 ha to 152356 ha). These facts can also be explained using the other two indicators, due to their complementarities. CONCLUSIONS In summary, one can say that landscape metrics vary with varying landscape attributes; correlate highly with one another and often provide redundant information — (which is not surprising, given they derive from a rather small set of possible attributes: area, border or edge length, distance, that one can measure), and relate differently according to the process under investigation. These results should not be surprising. The fragmentation of the landscape is a complex process that acts on a complex system and results in a wide arrangement of spatial patterns

. Altogether the results obtained in the study show the usefulness of global indicators in landscape change modeling and that this type of landscape analysis becomes increasingly important for the local actors, which must take decisions in agreement with the landscape potential of each region

. Regarding the study area landscape, there is a predominance of specific categories (built-up area or arable land), which have expanded over others, putting a high pressure over the natural conditions and local biodiversity. The latest evolution of Open Source GIS software, especially of those who can offer integrated desktop solutions such as QGIS, had created many opportunities for the development of new tools for spatial analysis. Besides of numerous plug-ins designed to ease the work of the researchers one of the most prominent feature of QGIS is it's 14th SGEM GeoConference on Ecology, Economics, Education and Legislation flexibility and capacity to interact with other packages, and by doing this the array of analyses that can be performed in spatial domain are limitless

. REF ERENCES

Pătru-Stupariu, I., Stupariu, M. S., Cuculici R. & Huzui, A. Application of the global indicators to landscape change modeling on Prahova Valley (Romanian Carpathians and Subcarpathians), International Journal of the Physical Sciences, vol. 6(3), pp. 534-539, 2011;

Pătroescu, Maria, Toma, Simona, Sasaki, L. And Apostol, G. Priorities in the rehabilitation and re-naturation of rural landscape of the Roumanian Plain, southern Romania, Analele Universităţii Bucureşti – Geografie, XLIX, 2000;

Farina A. Principles and methods in Landscape Ecology. Towards a Science of Landscape, Springer, 412 p, 2006;

Turner M.G. & Garnder R.H. (edit.). Quantitative methods in landscape ecology, Springer, New York, 1991;

McGarigal K. And Marks B. FRAGSTATS: Spatial pattern analysis program for quantifying landscape structure. General Technical Report PNW-GTR-351. USDA Forest Service, Pacific Northwest Research Station, Portland, Oregon, USA, 1995;

Tur ner M., Gardner R., O’ Neill R. Landscape Ecology in Theory and practice. Pattern and process, Springer, 403 p, 2001;

McGarigal, K., SA Cushman, Mc Neel, & E Ene. - FRAGSTATS v3: Spatial Pattern Analysis Program for Categorical Maps. Computer software program produced by the authors at the University of Massachusetts, Amherst. http://www.umass.edu/landeco/research/fragstats/fragstats.html , 2002.;

Perrera A.H., Baldwin D.J.B. And Schnekenburger F. LEAP II: A Landscape Ecological Analysis Package for Land Use Planners and Managers. Forest Research Report no. 146, Ontario, Canada, 1997;

Jaeger J.A.G. Landscape division, splitting index, and effective mesh size: new measures of landscape fragmentation. Landscape ecology 15.2, pp. 115-130, 2000;

Rutledge D. Landscape indices as measures of the effects of fragmentation: can pattern reflect process , Doc Science Internal Series 98, 27 p, 2003;

Jaeger J.A.G., Esswein H., Schwarz-Von Raumer H.G. Measuring Landscape Fragmentation with the Effective Mesh Size meff, Zurich, http://gpe.concordia.ca/documents/faltblatt_engl.pdf , 2006;

Roșca B., Chelaru D. -A., Pleșcan S. The Analysis Of Landscape Morphology In Lower Bistrita Valey Using Grass And Quantum Gis, 13th International Multidisciplinary Scientific GeoConference SGEM 2013, Conference Proceedings - Informatics, geoinformatics and remote sensing, pp. 951 – 959, 2013.