**3. Methodology**

We used cartographic techniques that combine remote sensing and GIS, with ecological analysis at different spatial scales. Both methods allowed us to determine the landscape changes and to detect pressures acting upon some areas included in the PNA during the last decades (urban and transport infrastructures development, etc). We reviewed i) previous studies focused on photo-interpretation techniques [30-34]; ii) land uses changes and territorial dynamics in different environmental conditions [35-40] and iii) definition of land uses categories [41-44]. We chose two working scales: the first analyzed the variability of the PNA as a whole (1:50,000), and the second, that use a more detail scale, permited us to recognize different ecological processes that occurred in the territory (1:12,500).

Cartography of Landscape Dynamics in Central Spain 231

development of cartography. It was considered of interest to note the state or degree of consolidation of urban or under urban development, as this is one aspect that has changed

We then crossed the two land-use maps by means of techniques of overlapping and digital layer intersection, thus comparing the information from both years. The result was a new map on which each patch showed the land use observed in both years. Thus, we identified all the types of changes that had taken place in the PNA during the 35-year period. Then it was possible to know if each type of vegetation or land use in each part of the territory has changed or not. Results allowed us to identify a set of changes occurred in the PNA during

Then we calculated the percentage of the area that has changed in the whole PNA and for each type of dynamic [28, 48]. These percentages allowed us to establish categories of dynamism that distinguished zones of more or less changing and map its spatial distribution. Finally, we used categories of dynamism map (1:12,500 scale) to select three locations. We conducted a more detailed study about natural dynamic of territory [49-51], using a set of ecological parameters as indicators of structure and organization of territory.

A more detailed study allowed us to recognize and map 23 types of land uses. These were the result of consider at the same time land uses and vegetation units. Using this criterion

In each selected locations, we calculated relative frequencies of each land use type using their abundance, and graphed their relative frequency profile in 1975 and 2009. Each profile has been defined by three parameters: the richness of land uses, R(u), calculated as the number of different categories of land uses in the corresponding year [52]; Shannon's

Evenness is the proportion between observed diversity value and maximum diversity value that would be possible to reach with the registered R(u). High values of E(u) indicate an even land use distribution, that means that there are not a dominating land use in the territory. To the contrary, low values of E(u) are indicating that one or a small set of land

In order to complete the study of evolution of territorial structure and organization, we analyzed ecological connectivity and fragmentation in each selected location in 1975 and 2009. We calculated ecological connectivity (c) of territory according to a set of permeability values assigned to each land uses. These permeability values were established taking into account matter and energy flows (genes, seeds, species, etc.) through two adjacent land uses [55-57], and assessing the quality of each land uses according to landscape functionality,

Finally, we calculated fragmentation of territory (f) according to the variation in patches number occurred between 1975 and 2009 at each selected locations. We focused on

we obtained categories as Dehesas *of Quercus ilex* subsp. *ballota* i.e.

diversity, H'(u), expressed in bits [27]; and Pielou's evenness, E(u) [53].

boundary effect caused by human land uses. Results were mapped.

in the whole Madrid Region from the 70's of last century.

these 35 years.

uses are more frequent [54].

forestry value, cultural value, among others.

We used the aerial photography of Spanish Air Force of 1975 and the orthophotography of 2009 (Plan Nacional de Ortofotografía Aérea). ERDAS Imagine 9.1 software was used for processing analogical information (1975 flight, Spanish Air Force). Previously, we had tested ER-MAPPER and ArcGis 9.10 methods to this end. We scanned each of 150 photograms at 600 dpi. Mosaicing was conducted using Mosaic Tool extension. This tool showed the best balanced colour result. Each photo was improved with a Root Mean Square (RMS) and its tolerance was less than 0.5. Distortions of the photos were corrected using a Digital Terrain Model (DTM) at 1:5,000 scale. Re-sampling method applied was the nearest neighbour algorithm employing at least a cubic polynomial fit using 15 Control Ground Points and at least 10% of overlapping areas. The result was a continuous image of the study area with a 5 x 5 m resolution.

Phenological changes mean different colours that depend on season and this colour difference causes errors in the photo-interpretation so, it was necessary to balance colours for the ortophotography of 2009 (2.5 x 2.5 m resolution).. This problem is more evident in agricultural areas. Likewise, existing vegetation and land use maps of Madrid Region were reviewed [45, 46]. All layers were managed in a format compatible with ArcGIS 9.3 (shapefile, coverage or GRID) referred to the WGS84 ellipsoid and UTM coordinates Zone 30 N.

The photo-interpretation considered an accuracy of the reference map unit depending on the type of area, establishing a minimum of 0.5 ha (1:12,500 scale), similar to that used in other studies with this orthophotography scale in forest formations [47] and of 0.61 ha (1:50,000 scale) using cartographic techniques that merge patches (dissolve ie.) Also on the orthophotography were required some data about the phytostructure and percentages of canopy cover. If necessary, we worked with DGPS techniques (GPS Trimble Nomad 6GB) to refine the patches shape.

We employed a touch screen Wacom Cintig 12WX joined to ArcGis editor. Topological errors were processed using ARCEDIT. Finally, we assigned a topology to each layer using ARC/INFO 9.1. Database was designed with three fields: land use, connectivity value and surface area (ha).

The photointerpretation was completed with fieldwork. It took over 2,000 panoramic photographs for checking on field the areas that were more complex during the development of cartography. It was considered of interest to note the state or degree of consolidation of urban or under urban development, as this is one aspect that has changed in the whole Madrid Region from the 70's of last century.

230 Cartography – A Tool for Spatial Analysis

We used cartographic techniques that combine remote sensing and GIS, with ecological analysis at different spatial scales. Both methods allowed us to determine the landscape changes and to detect pressures acting upon some areas included in the PNA during the last decades (urban and transport infrastructures development, etc). We reviewed i) previous studies focused on photo-interpretation techniques [30-34]; ii) land uses changes and territorial dynamics in different environmental conditions [35-40] and iii) definition of land uses categories [41-44]. We chose two working scales: the first analyzed the variability of the PNA as a whole (1:50,000), and the second, that use a more detail scale, permited us to

We used the aerial photography of Spanish Air Force of 1975 and the orthophotography of 2009 (Plan Nacional de Ortofotografía Aérea). ERDAS Imagine 9.1 software was used for processing analogical information (1975 flight, Spanish Air Force). Previously, we had tested ER-MAPPER and ArcGis 9.10 methods to this end. We scanned each of 150 photograms at 600 dpi. Mosaicing was conducted using Mosaic Tool extension. This tool showed the best balanced colour result. Each photo was improved with a Root Mean Square (RMS) and its tolerance was less than 0.5. Distortions of the photos were corrected using a Digital Terrain Model (DTM) at 1:5,000 scale. Re-sampling method applied was the nearest neighbour algorithm employing at least a cubic polynomial fit using 15 Control Ground Points and at least 10% of overlapping areas. The result was a continuous image of the study area with a 5

Phenological changes mean different colours that depend on season and this colour difference causes errors in the photo-interpretation so, it was necessary to balance colours for the ortophotography of 2009 (2.5 x 2.5 m resolution).. This problem is more evident in agricultural areas. Likewise, existing vegetation and land use maps of Madrid Region were reviewed [45, 46]. All layers were managed in a format compatible with ArcGIS 9.3 (shapefile, coverage or

The photo-interpretation considered an accuracy of the reference map unit depending on the type of area, establishing a minimum of 0.5 ha (1:12,500 scale), similar to that used in other studies with this orthophotography scale in forest formations [47] and of 0.61 ha (1:50,000 scale) using cartographic techniques that merge patches (dissolve ie.) Also on the orthophotography were required some data about the phytostructure and percentages of canopy cover. If necessary, we worked with DGPS techniques (GPS Trimble Nomad 6GB) to

We employed a touch screen Wacom Cintig 12WX joined to ArcGis editor. Topological errors were processed using ARCEDIT. Finally, we assigned a topology to each layer using ARC/INFO 9.1. Database was designed with three fields: land use, connectivity value and

The photointerpretation was completed with fieldwork. It took over 2,000 panoramic photographs for checking on field the areas that were more complex during the

GRID) referred to the WGS84 ellipsoid and UTM coordinates Zone 30 N.

recognize different ecological processes that occurred in the territory (1:12,500).

**3. Methodology** 

x 5 m resolution.

refine the patches shape.

surface area (ha).

We then crossed the two land-use maps by means of techniques of overlapping and digital layer intersection, thus comparing the information from both years. The result was a new map on which each patch showed the land use observed in both years. Thus, we identified all the types of changes that had taken place in the PNA during the 35-year period. Then it was possible to know if each type of vegetation or land use in each part of the territory has changed or not. Results allowed us to identify a set of changes occurred in the PNA during these 35 years.

Then we calculated the percentage of the area that has changed in the whole PNA and for each type of dynamic [28, 48]. These percentages allowed us to establish categories of dynamism that distinguished zones of more or less changing and map its spatial distribution. Finally, we used categories of dynamism map (1:12,500 scale) to select three locations. We conducted a more detailed study about natural dynamic of territory [49-51], using a set of ecological parameters as indicators of structure and organization of territory.

A more detailed study allowed us to recognize and map 23 types of land uses. These were the result of consider at the same time land uses and vegetation units. Using this criterion we obtained categories as Dehesas *of Quercus ilex* subsp. *ballota* i.e.

In each selected locations, we calculated relative frequencies of each land use type using their abundance, and graphed their relative frequency profile in 1975 and 2009. Each profile has been defined by three parameters: the richness of land uses, R(u), calculated as the number of different categories of land uses in the corresponding year [52]; Shannon's diversity, H'(u), expressed in bits [27]; and Pielou's evenness, E(u) [53].

Evenness is the proportion between observed diversity value and maximum diversity value that would be possible to reach with the registered R(u). High values of E(u) indicate an even land use distribution, that means that there are not a dominating land use in the territory. To the contrary, low values of E(u) are indicating that one or a small set of land uses are more frequent [54].

In order to complete the study of evolution of territorial structure and organization, we analyzed ecological connectivity and fragmentation in each selected location in 1975 and 2009.

We calculated ecological connectivity (c) of territory according to a set of permeability values assigned to each land uses. These permeability values were established taking into account matter and energy flows (genes, seeds, species, etc.) through two adjacent land uses [55-57], and assessing the quality of each land uses according to landscape functionality, forestry value, cultural value, among others.

Finally, we calculated fragmentation of territory (f) according to the variation in patches number occurred between 1975 and 2009 at each selected locations. We focused on boundary effect caused by human land uses. Results were mapped.
