Mapping Environmental Fragility in Anthropized Landscapes using Fuzzy Logic and continuous data spatialization

doi:10.21203/rs.3.rs-3308096/v1

Download PDF

Research Article

Mapping Environmental Fragility in Anthropized Landscapes using Fuzzy Logic and continuous data spatialization

https://doi.org/10.21203/rs.3.rs-3308096/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Demographic expansion and intensive human activities have negatively impacted natural landscapes worldwide. The conversion of natural areas into anthropogenic sites often occurs without adequate planning, resulting in severe environmental and social problems. There is a constant search for the development of diagnosis and environmental planning methodologies to analyze the landscape in an integrated way. Geographic Information Systems (GIS) and the cartographic representation of different landscape attributes (i.e., soil types, rainfall regime, topography, land use/land cover, among others) are essential tools for understanding the environmental dynamics of a region. This study focuses on analyzing environmental fragility in areas near the reservoir of the Jurumirin Hydropower Plant in the state of São Paulo, Brazil, using a fuzzy logic-based approach and spatially continuous data. The results indicate a medium and high environmental fragility predominance, significantly influencing agricultural practices and land use. The fuzzy approach allowed for a more precise representation of transitions between fragility classes, enhancing the understanding complex interactions among landscape attributes. These findings significantly impact reservoir management and environmental decision-making in transformed landscapes.

Environmental Fragility

Environmental Planning

Landscape Analysis

Geographic Information Systems

Data Spatialization

Demographic expansion and the intensification of human activities threaten natural landscapes worldwide. Although some landscape configuration alterations originate from natural processes, the human factor presents itself as the primary source (Havlíček et al. 2022), aggravated by the fact that the conversion of natural areas into anthropic sites is often conducted without adequate planning (Vettorazzi and Valente 2016), resulting in multiple environmental and social issues.

Several environmental issues caused by changes in the landscape are reported, such as forest fragmentation, habitat reduction, biodiversity loss (Ferreira et al. 2019), erosion and soil loss (Fernandez et al. 2018), expansion of degraded areas, sedimentation of water bodies (Degife et al. 2019), rapid growth in the demand for water resources and the degradation of water quality (Mello et al. 2020). Several studies include their implications in essential Ecosystem Services (E.S.) (Jones et al. 2019), such as carbon regulation, pollination, pollution control, biodiversity (Hao et al. 2017), water resources (Valente et al. 2021a), and climate regulation (Maeda et al. 2021).

A relevant source of changes in the landscape configuration is related to the water supply for society's needs (Anjinho et al. 2021), dams and reservoirs are commonly used to guarantee the supply for the demand for water in its multiple uses (Padedda et al. 2017). Despite the reservoirs crucial importance, their implementation promotes significant effects on the territorial arrangement and in the hydrological dynamics of the landscape, affecting the economic, social, and environmental spheres (Wu et al. 2003; Jones et al. 2016; Havlíček et al. 2022).

During its construction, large-scale social and environmental issues are caused by flooding large areas due to river damming, and after, high anthropic pressures are observed in their contributors' watersheds (Huang et al. 2019). Such regions undergo an intense anthropization process, threatened mainly by replacing the natural soil cover with agricultural or urbanized areas, considerably affecting surface runoff and sediment production, and threatening the reservoir integrity (Anjinho et al. 2021).

To minimize those impacts, it is essential to manage human-modified landscapes better. Researchers are constantly looking for strategies to prevent, reverse or mitigate this environmental condition and reconcile socio-economic needs and environmental sustainability (Tiang et al. 2021). Strategies based on Geographic Information Systems (GIS) allow an integrated analysis of a landscape's physical, environmental, and anthropic characteristics, helping to understand the region's environmental dynamics and assisting in decision-making (Alves et al. 2021). Thus, in the last decades, GIS-based methodologies have been essential to access the driving forces that affect the E.S. and guide landscapes to high fragility levels (Oliveira-Andreoli et al. 2021).

The concept of environmental fragility concerns the susceptibility of the ecosystem to changes in its dynamic equilibrium according to its intrinsic and extrinsic characteristics and has been used to assess the susceptibility of natural ecosystems under diverse conditions (Anjinho et al., 2021). The Environmental Fragility Analysis (EFA) framework, first proposed by (Ross 1994; Massa and Ross 2012), considers a landscape's physical and climatic characteristics, such as relief, slope, soil types, and rainfall, and how they interact with the different degrees of soil protection through land use and land cover (LULC). Considering these factors, EFA's can help identify areas more susceptible to environmental degradation and guide decision-making towards more sustainable land management practices.

Two components must be considered in the EFA's framework: Potential Environmental Fragility (PEF) and Emerging Environmental Fragility (EEF). The first relates to the intrinsic fragility of the landscape, considering physical and climatic attributes, and the second represents the interaction between the different soil protection degrees promoted by its LULC with their respective PEF (Oliveira-Andreoli et al., 2021).

Traditionally, Boolean logic is used to spatialize the thematic maps required to implement overlay-based GIS methodologies as the EFA's framework (Qiu et al. 2014). However, the intrinsic simplicity of Boolean logic sometimes represents a limitation for a more credible assessment of the complex features presented in a landscape (Moonjun et al. 2020; Santos et al. 2020). In a GIS environment, a continuous data spatialization and the Fuzzy set concept (Zadeh 1965) provide considerable flexibility in options to solve this problem by supporting intermediate values between false and true, allowing for mimicking human reasoning and overcoming the problem of class boundaries provided by Boolean logic (Baalousha et al. 2021). The difference between the two logics of data spatialization stands in the membership function representing a Fuzzy set, which provides considerable flexibility of options to solve the proposed problem.

Fuzzy sets are applied with many objectivities in the environmental sciences, providing more explanatory results to help the decision-makers when they map priority areas (Valente et al., 2021), site suitability (Widiatmaka 2016; Chen et al. 2018), environmental fragility (Anjinho et al. 2021), and modeling ecological capability (Ghomi-Avili et al. 2020). However, in works that consider the fuzzy logic, even after applying the necessary pertinence functions on the landscape attribute maps to some of them, mainly those derived from thematic mappings, the resulting fuzzified map still carries traces of the rigid Boolean spatialization that originated them. Such situations represent a trend to overlay-based approaches since the rigid limits imposed by the discrete boolean spatialization often stand out as the smooth variation characteristic of continuous fuzzy maps (Bacani et al. 2015).

In this context, this study aimed to identify fragility areas, mainly related to soil degradation and loss, to contribute to the water reservoir management and integrity and to evaluate the performance of the fuzzy spatialization on overlay-based GIS methodologies. Thus, we structured a framework that implements the EFA according to the fuzzy logic considering continuous data spatialization on the watersheds that contribute to a water reservoir.

Study Area

The study area (Fig. 1) has approximately 4800Km² and comprehends 30 watersheds contributing to Jurumirin's reservoir supply. Located in the center-south of São Paulo state, the reservoir integrates the National Electric System Operator of Brazil and is responsible for the water accumulation to supply other reservoirs of the Paranapanema river.

The region's altitude varies from 525m to 925m, with a Cfa as the predominance climate type, according to Köppen's classification, i.e., tropical wet, characterized by a hot summer, not having a well-defined dry season and with an average temperature above 22°C. The exception is in areas where the altitude exceeds 800m, where the climate is Cfb, humid tropical but characterized by a milder summer (Alvares et al. 2014).

The drought of 2014 and 2015 that affected the Brazilian hydroelectric potential also influenced the Jurumirim reservoir negatively, reducing its water volume to the lowest level and, consequently, its energy generation too (Hunt et al. 2022). The environmental problems worsened after this period. However, we can observe anthropic pressures since the reservoir construction (Mascaro 2004), with changes and alterations in the original vegetation characterized by a transitional physiognomy between the Atlantic Forest and the Brazilian Savannah (IBGE 2012).

The anthropic pressures noted in the region reflect the current panorama observed through the São Paulo state (Calaboni et al. 2018), with forest areas being reduced since the 1960s to the detriment of agro-industrial expansion and advances in the urbanization process.

Conceptual Model

The study aims to access the PEF and EEF in the reservoir's contribution watersheds, considering continuous data spatialization and fuzzy logic. We follow the traditional EFA's framework, considering their two components, PEF, EEF. The first relates to the intrinsic fragility of the landscape and considers physical and climatic attributes. The second represents the interaction between the different soil protection degrees promoted by its LULC with their respective PEF (Oliveira-Andreoli et al., 2021).

The following items present all the methodological procedures for implementing the EFA's framework.

The procedures involve obtaining the spatial data, their continuous spatialization, applying the fuzzy membership functions, and finally, proceeding with the fragility maps aggregation to obtain the maps of potential and emerging environmental fragility.

Spatial Data and Landscape Attribute Maps

The landscape attribute maps that assess the PEF and EEF in the reservoir's contribution watersheds represent the landscape's physical, climatic, and anthropic characteristics, expressed by slope, soil erodibility, rainfall erosivity, and LULC. Its data is derived from public databases, as detailed in Table 1. The cartographic information was processed and analyzed in a GIS environment, standardized for the SIRGAS 2000 geodesic system, using the Universal Transverse Mercator (UTM) projection for zone 22 south.

Table 1

Source, description and scale of the spatial data that originated the landscape attribute maps for the watersheds contributing to the UHE Jurumirim reservoir, São Paulo, Brazil.
Spatial Data	Source	Description	Scale	Landscape Attribute Map
Pedology	São Paulo State pedological map (Rossi, 2017)	Soil classes	1:100.000	Soil Erodibility (K factor)
LULC	Sentinel − 2A Images	Land Use and Land Cover classes	10m Spatial Resolution	LULC fragility classes
Slope	ALOS PALSAR DEM	Digital Elevation Model (DEM)	12,5m Spatial Resolution	Slope (%)
Rainfall	National Water Agency	Monthly Rainfall Data	-	Rainfall Erosivity (R factor)

These landscape attribute maps follow previous works (Spörl and Ross 2004; Massa and Ross 2012; Alves et al. 2021; Anjinho et al. 2021) that implement the EFA's framework since it requires a set of data to assess the environment in an integrated manner (Anjinho et al. 2021). However, to reduce the analysis subjectivity, it was decided to use the K factor of the Universal Soil Loss Equation (USLE) to represent the pedological characteristic (Soil erodibility) and the USLE's R factor to represent the climatic characteristic (Rainfall erosivity).

The methodological procedures necessary to obtain all the landscape attribute maps are presented below.

(1) K factor: we obtained from the Pedological map of the State of São Paulo data, produced by (Rossi 2017) and published by São Paulo's Forestry Institute. Data are presented in vector format, with information about soil classes' (first taxonomic level - Order). Some have two or more classes associated (2nd taxonomic level - Suborder) or soil groups. We reclassified these classes according to their respective K values obtained from previous studies (Mannigel et al. 2002; Correchel 2003; Demarchi and Zimback 2014; Corrêa et al. 2015). The criteria to define K values consider the soil characteristics of EMBRAPA (2006) and IBGE (2007). The soil classes of the studied area with their respective K values are available in Appendix I.

(2) R factor: it was produced from data from 27 rainfall Pluviometric stations distributed throughout the study area provided by the National Water Agency (ANA). The calculation of the R factor for each station was carried out from monthly rainfall data, considering observations from 1989 to 2019, according to Lombardi Neto and Moldenhauer (1992). The spatialization was generated by interpolating each rainfall station point using the ordinary kriging interpolator as described by Oliveira et al. (2012).

(3) Slope: we produced from the interpolation of the altitude value provided by the Digital Elevation Model (DEM) from ALOS (Advanced Land Observing Satellite) satellite and PALSAR (Phased Array L-Band Synthetic Aperture Radar) sensor, which was obtained by the Alaska Satellite Facility, counting with 12,5 meters of spatial resolution.

LULC: we produced from the supervised classification of Sentinel-2A satellite images from May 5, 2019, obtained from the U.S Geological Survey database. The available bands, with a spatial resolution of 10 meters, were 2, 3, 4, and 8, which refers to blue, green, red, and near-infrared wavelengths. The images supported appropriate atmospheric and geometric corrections, and we implemented the Object-Based Image Analysis (OBIA) algorithm for classification, obtaining 90.60% global accuracy.

After the procedures described, the landscape attribute maps shown in Fig. 2 were obtained, which were analyzed individually to assign fragility levels and to apply the fuzzy membership functions according to the procedure described in the next items.

Environmental Fragility Analysis

Obtaining the PEF and EEF maps involves overlaying maps presented at various scales. Thus, landscape attribute maps were rescaled to a common scale, that varies between zero and one, from applying linear fuzzy membership functions, originating the fragility maps for each analyzed characteristic.

However, considering the fuzzy approach, the spatialization of landscape attribute maps must be continuous. The slope and R factor maps are naturally continuous. Therefore, no procedure related to their spatialization was necessary before applying the fuzzy membership functions. However, since maps referring to the K factor and LULC are naturally discretely spatialized, it was necessary to eliminate their binary character and spatialize them continuously before applying the fuzzy membership functions.

We generated a regular densified grid point and superimposed it on the K factor and LULC maps to assume its cell value. Then, using the Empirical Bayesian Kriging method (Gribov and Krivoruchko 2020; Krivoruchko and Gribov 2020), continuously generate a raster surface representing the K factor and LULC maps.

The grid point densification was around 100 meters apart. Considering the LULC map, we have a sampling point every five pixels in the raster. Thus, the interpolator smooths and assigns intermediate values in the boolean class boundaries without de-characterizing its origin.

The membership functions bring all the maps to a standard scale, varying between zero and one, with zero associated with lower fragility levels and one with higher ones. The minimum and maximum values considered for each map are described below. This procedure was implemented through the Fuzzy Membership tool in ArcGIS 10.5.

To the K factor map, 0.00 \(\text{t}.\text{h}\text{a}.\text{h}.{\text{h}\text{a}}^{-1}.{\text{M}.\text{J}.}^{-1}.{\text{m}\text{m}}^{-1}\) was associated with the value 0, and 0.065 \(\text{t}.\text{h}\text{a}.\text{h}.{\text{h}\text{a}}^{-1}.{\text{M}.\text{J}.}^{-1}.{\text{m}\text{m}}^{-1}\) was associated with the value 1. To the R factor map, 5,995.36 \(\text{M}\text{J}.\text{m}/\text{h}\text{a}.\text{h}.\text{y}\text{e}\text{a}\text{r}\) was associated with the value 0 and 6,659.45 \(\text{M}\text{J}.\text{m}/\text{h}\text{a}.\text{h}.\text{y}\text{e}\text{a}\text{r}\) with the value 1.

To slope, 0% was associated with 0 and 45% to 1, and to LULC, the fragility levels were assigned considering empirical characteristics, such as soil protection level, soil impermeability rate, presence and type of vegetation cover, and protection actions adopted in management. Such fragility levels had based on those adopted in pertinent literature by authors such as Ross (1994), Spörl and Ross (2004), Bacani et al. (2015), Barros et al. (2017), Valle et al. (2016), Storto and Cocato (2018), Campos et al. (2019), Anjinho et al. (2021), Oliveira-Andreoli et al. (2021), and Alves et al. (2021). In addition, they have displayed according to local reality.

In this context, a very low level of fragility was assumed for native vegetation, a low level for forestry, a medium level for pasture and fruit growing, a high level for agriculture, and a very high level of environmental fragility for the road network, beaches, wetlands, and urbanized areas. Notably, water-related areas were excluded from the analysis for all fragility maps.

After these steps and with all the landscape attribute maps continuously spatialized and standardized, we execute the map overlay. To obtain the PEF map, we overlay the slope, K factor, and R factor maps, and to obtain the EEF, we overlay the PEF and the LULC map. The aggregation was performed using the ArcGIS 10.5 Fuzzy Overlay tool, using the SUM operator. Figure 3 shows the synthesized procedure.

Landscape Attribute Maps

Figure 4 shows landscape attribute maps continuously spatialized in fragility scale accompanied by their respective density graphs. Density graphs help analyze fuzzy sets since they express the density of pixels associated with environmental fragility. Note that the relatively slightly uneven relief of the region contributed to the high concentration of pixels with values below 0.4 in the slope map. Regions of high fragility for this attribute are found only in rugged terrain, such as in the northern region, where the mountain range known as 'Cuesta de Botucatu' begins.

The R Factor shows a predominance of medium fragility with the highest pixel density between 0.4 and 0.6, also showing significant participation with slightly higher fragility, reaching a good concentration of pixels with values between 0.6 and 0.9. The distribution of values on the map appears visually well characterized, showing increased fragility in the South-North direction.

The map reflects this fact, where noticeable shades between light green and yellow stand out in the study area. The large concentration of median values is associated with the predominance of Oxisols in the region, which presents intermediate erodibility values. The more significant fragility regions are mainly associated with patches of more erodible Ultisols. The few regions that notice minor fragility are associated with small Nitosols patches in the study area (see supplementary material).

The LULC presents concentration peaks around 0.2, 0.4, 0.6, and 0.8, demonstrating a more balanced distribution of fragility due to the heterogeneity of land uses in the region. However, there is a higher concentration of values close to 0.8 due to the large concentration of agricultural regions, especially in the southern. Intermediate fragility is attributed mainly to pastures and, to a lesser extent, to fruit growing, while lower levels of fragility are associated with forestry, with a significant concentration in the Northeast region of the study area. Regions associated with lower levels of fragility are those where native forest fragments are present, presenting less fragility in their core area and a decrease for intermediate classes in bordering regions, presenting a transitional buffer area between the patch and the neighboring uses.

These buffer zones are present in all landscape attribute maps. The map's continuous spatialization provides such transitional values, assigning transitional values where, for Boolean approaches, they are abrupt breaks between classes. A desired characteristic, especially when using a fuzzy approach to perform the analysis, since it allows the association of intermediate values between zero and one, considering the value of each image pixel and capturing all the information obtained by the linear membership function.

Environmental Fragility

The first result of implementing the EFA's framework is the PEF (Fig. 5) from the overlap among the Slope, R, and K factor maps. Observing the map and following the fragility pixel density curve (Fig. 5), notice a higher concentration of fragility pixels between 0.4 and reaching its peak at 0.8. Such results demonstrate a predominance of medium and high fragilities in the region.

The landscape attribute map most present in the PEF result refers to the R factor, dictating the regionalization of fragility in its characteristic direction of South-North growth. Thus, regions of lower potential environmental fragility predominate in the southern portion of the study area, while in the northern portion, the predominance is of higher potential fragility.

The K Factor is also present in the PEF map, mainly in the more erodible patches of Red Latosols and smaller patches of Haplic Gleissolos (see supplementary material), these in regions close to the reservoirs, especially in floodplains. Slope contributed to the cracked appearance of the resulting map. Since its distribution presents lower levels of fragility, it presents little expressiveness for regionalization, except in small, steeper regions such as the chain mountainous to the north or specific locations, generally close to the drainage network.

The transitional values provided by the continuous spatialization provide coherence to the fragility regionalization since it reduces the occurrence of antagonistic very close to each other, maintaining a gradual transition of fragility values.

In a second moment, the analysis obtains the emerging environmental fragility map (Fig. 5) by overlapping the PEF and LULC maps. The results show a significant predominance of regions of high and medium fragility throughout the EEF map, while regions with low levels of fragility appear with little expression exclusively to the south of the reservoir. The pixel fragility density curve shows predominant values between 0.8 and 1.0, followed by a stretch between 0.6 and 0.8.

The result demonstrates the significant influence of LULC, as in the south of the reservoir, where the native vegetation maintains low levels of fragility where it is present. However, more fragile areas such as urbanized areas and wetlands negatively influence fragility.

Despite the significant participation of LULC, we also observe that the PEF map dictates the background of regionalization, such as the predominance of more great fragility, especially in the North region, where not even native vegetation can alleviate emerging environmental fragility. The main driving force behind this increase is the land use classes representing high fragility, mainly urbanized areas, agriculture, and wetlands. In the southern region, where lower fragility classes are predominant for EPF, agriculture drives the region's fragility to higher values. To the east of the study area, clusters of medium and low fragility can be observed, mainly linked to planted forests that are very present in that region.

The analysis carried out with the landscape attribute maps indicates that the continuous spatialization of the data, combined with the fuzzy membership functions applied, provide fragility maps consistent with the database used and effectively represent reality. Continuous data combined with fuzzy sets is highly appreciated in overlay-based analyses (Eastman 2015) since it allows greater flexibility to the applied methods than when using discrete maps conditioned to Boolean logic (Jiang and Eastman 2000).

By considering smooth variations in the level of fragility of landscape attributes, continuous spatialization not only breaks with the rigid limits imposed by Boolean classes but also brings a new level of variability to the analysis, the intra-class variability, since the buffer formed zone at its edges reflects the influence of its neighbor. The core of a class represents its intrinsic fragility, and it edges the result of the interaction with another class, which has a particular interest in maps such as LULC.

The benefits of continuously representing landscape attributes can be observed in different contexts. Guth et al. (2021) demonstrate that the geomorphometric complexity is better analyzed on continuous surfaces than discrete ones and mentions that significant practical implications are observed, mainly when the terrain is analyzed simultaneously with other biophysical aspects of the landscape. Likewise, fuzzy logic increases the usefulness of pedological data, as it treats the soil as a continuous variable in space, considering the slight variations that occur until one soil type gradually converts into another (Moonjun et al. 2020). Such characteristics are essential when, for example, these data are used in hydrological and erosion studies.

Continuous surfaces also represent a scientific advance in data representation, as is the case for rainfall values, since such an approach allows the estimation of values at unsampled points without bias and with minimal variance (Oliveira et al. 2013). Thus, the use of discrete surfaces to spatialize such a variable is inconsistent with current rainfall event mapping techniques (Nearing et al. 2017).

Notably, the LULC map (Fig. 4) shows that the fuzzy spatialization brings a transitional region between the class's boundaries, especially visible when antagonistic fragility classes confront, making a buffer zone between them. As Baalousha et al. (2021) reported, such a characteristic overcomes the boundary problems presented by the boolean maps and argues that even for a land use and land cover map, the discrete representation cannot be considered fully adequate since, ultimately, such maps are derived from remote sensing images and the spectral blending of the pixel per se represents a certain degree of uncertainty, which can be considered from the implementation of fuzzy sets.

The importance of considering a buffer zone between LULC classes is justified in several contexts. Concerning forest fragments, edge effects that negatively affect the structure, microclimate, biodiversity, and protection of these environments are widely reported in the literature (Ferreira et al. 2019), so the level of environmental fragility is naturally higher at their edges than at their core.

The negative influence of urbanized areas on the environment is significant up to 40Km away from its source (Du et al. 2019), so it is of interest to our environmental analysis that this effect, in a certain way, manifests itself in the buffer zone provided by the fuzzy approach.

By overcoming the boolean problems with the borders, the fuzzy logic allowed better detailing in the mapping and a better performance considering the heterogeneity of the landscape attributes, Baalousha et al. (2021) consider this one of the main reasons that fuzzy logic performs better in vulnerability mappings, for example.

A small participation of pixels with low or very low levels of fragility (values between 0.0 and 0.4) was identified, demonstrating that the occurrence of regions where the association of flat reliefs, poorly erodible soils, and lower erosive power of rains occur is rare. However, it is remarkable that the concentration of such regions in the southern portion of the study area is visibly determined mainly as a function of the R Factor.

Here, one must consider that the R Factor reflects long-term erosion rates reasonably on average (Nearing et al. 2017). However, Oliveira et al. (2013) demonstrate that significant variations in the severity of this factor are only observed on a macroscale analysis, so it can be said that the significant participation of this factor in the PEF represents an overestimation of the contribution of rainfall erosivity to environmental fragility.

The low variability of rainfall data at the watershed scale is yet reported (Lombardi Neto and Moldenhauer 1992; Oliveira et al. 2013; Nearing et al. 2017). In studies where this variable is considered in conjunction with other biophysical characteristics, the authors use different strategies to make its inclusion in decision-making processes appropriate. Such strategies can range from the simplest ones, such as considering rainfall as a constant for the entire area (Valle et al. 2016; Alves et al. 2021) or the option for not including it (Campos et al. 2019), to more complex solutions that involve assigning different weights to the landscape attribute maps. In this context, according to Spörl and Ross (2004), the predominance of the medium fragility class can be credited to the equal weighting between the landscape attributes, which can attenuate the results and may mask the identified fragility.

In similar scenarios, previous studies also achieved similar results, Anjinho et al. (2021) and Alves et al. (2021) observed very high PEF clusters in mountainous relief associated with Ultisols and Neossols, in both cases with little significant contribution, 0.88 and 0.2% respectively. Miguel et al. (2021) state that Neosols are naturally susceptible to erosive processes, mainly when found in steep regions since they are new, shallow soils and have a sandy-loam texture. To Castro (2016), the low drainage capacity of Ultisols, especially from 40 cm in depth, increases the possibility of developing erosive processes, especially in steep or poorly managed regions. Several studies of fragility and vulnerability report the high susceptibility to erosive processes in mountainous regions (Aires et al. 2022), soil loss is a fundamental factor for this risk, and when associated with regions with little vegetation, fragility increases considerably (Campos et al. 2019).

When the LULC map is added to the analysis to obtain the EEF, the first response observed is a substantial increase in environmental fragility, so an extensive portion of the study area has high EEF fragility values. It is noted in the southern region, where for the EPF, the very low and low fragility predominates, whereas, in the EEF, the medium fragility predominates. This result is related to the high agricultural activities present in the region, characterized by annual croplands on the conventional cropping system, with the use of synthetic fertilizers, pesticides, and mechanization.

According to Calaboni et al. (2018), this production system has expanded in São Paulo since the 1960s, accompanied most of this period by a decrease in forest cover, including in areas considered less suitable, such as sloping regions and more sensitive soils. The constant movement and exposure of the soil provided by this production system are responsible for the increase in environmental fragility in these regions, according to authors such Fushita et al. (2010) and Oliveira-Andreoli et al. (2021).

The low hydraulic conductivity can cause severe problems in terms of runoff and soil erosion (Lucas-Borja et al., 2019), mainly caused by the transit of large agricultural machinery, especially in large-scale grain production areas, as is the case of activities developed in the region. The significant influence of the Factor R is also noted for the Fuzzy approach, with greater participation related to the highest values, resulting in a predominant region of high and very high fragility in the northwest and north edges of the studied region, standing out over the other landscape attributes.

Throughout the map, it is possible to notice that in regions where there is native forest cover, small clusters with low fragility are formed, demonstrating the performance of an essential role in reducing anthropic pressures by the forest patches, especially along with the hydrographic network, reducing a load of nutrients and the runoff of sediments into the rivers (Mello et al. 2017).

Campos et al. (2019) also reported this, the authors observed fragility reduction in regions covered with planted forests, crediting this fact to the ideal protection of the soil provided by this cover, mainly reducing runoff. Lamb (2018) considers that timber plantations provide such ecosystem services but with low association with biodiversity improvement.

In this case, it must be considered that most native forest fragments found in the study area are made up of riparian vegetation, following the drainage network following the Brazilian legislation on permanent preservation areas.

However, this characteristic originates in elongated fragments with little nuclear area. The literature widely reports that neighboring land uses significantly affect elongated fragments with small nuclear areas and cannot fully perform the ecosystem services provided by native vegetation.

The elongated shape in forest patches favors the susceptibility to edge effects that affect the structure and biodiversity (Magnago et al. 2015) and affects its resilience to adverse phenomena such as storms, fires, and biological invasions, in addition to being greatly influenced by neighboring land use and by human activities as logging and hunting (Ferraz et al. 2014).

In a way, the Fuzzy approach allows considering such characteristics, considering that the presence of a forest fragment alone does not necessarily represent the power to increase local fragility. This effect is observed with greater consistency as the fragment has a shape more regular and more extensive nuclear area it is possible to observe lower fragility clusters.

This study assessed the performance of the EFA framework within the influence zones of a water reservoir. The framework's evaluation utilized spatialized landscape attribute maps produced in the fuzzy logic context. The framework demonstrated its suitability for identifying pairs of environmentally protected areas when considering the interplay of distinct landscape attributes. It also stood out for its simplicity and utilization of readily available input data, making its methodology accessible and straightforward for implementation across various locations globally.

However, it became evident that the methodology's sensitivity is pronounced concerning the specific landscape attributes integrated into the analysis. Therefore, carefully selecting these attributes is crucial to ensure the resultant map accurately represents all environmental variables. The seamless integration of continuous spatialization and protection attribution through fuzzy membership functions fosters a harmonious synergy among landscape attributes. This synergy enhances the coherence of regionalization, thus facilitating the planning of management strategies, informed decision-making in environmental governance, and the effective execution of public policies.

For future research, we recommended exploring diverse landscape attributes, particularly those related to variables representing rainfall or the erosive potential of rainfall. Additionally, investigating spatialization methods that yield uninterrupted environmental variable surfaces and employing distinct fuzzy membership functions to comprehend control levels in landscape attribute maps would be valuable areas of investigation.

Statements & Declarations

This work was supported by the Coordination for the Improvement of Higher Education Personnel - Brazil (CAPES) – Finance Code 001. The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Milton Vinícius Morales and Roberta Averna Valente. The first draft of the manuscript was written by Milton Vinícius Morales and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Aires A de A, Costa JD da, Bezerra JM, Rêgo ATA do (2022) Análise Multicritério da Suscetibilidade Erosiva da Microbacia Hidrográfica da Barragem de Pau dos Ferros/RN. Revista Brasileira de Geografia Física 15:1128–1141
Alvares CA, Stape JL, Sentelhas PC, et al (2014) Köppen ’ s climate classification map for Brazil. Meteorologische Zeitschrift 22:711–728. https://doi.org/10.1127/0941-2948/2013/0507
Alves J, Demetrius C, Michel S, et al (2021) Environmental fragility and land use capacity as instruments of environmental planning , Caratinga River basin , Brazil. Environ Earth Sci 1–13. https://doi.org/10.1007/s12665-021-09553-2
Anjinho P da S, Barbosa MAGA, Costa CW, Mauad FF (2021) Environmental fragility analysis in reservoir drainage basin land use planning: A Brazilian basin case study. Land use policy 100:104946. https://doi.org/10.1016/j.landusepol.2020.104946
Baalousha HM, Tawabini B, Seers TD (2021) Fuzzy or non-fuzzy? A comparison between fuzzy logic-based vulnerability mapping and DRASTIC approach using a numerical model. a case study from qatar. Water (Switzerland) 13:. https://doi.org/10.3390/w13091288
Bacani VM, Sakamoto AY, Luchiari A, Quénol H (2015) Sensoriamento Remoto e sig aplicados à avaliação da fragilidade ambiental de bacia hidrográfica. Revista Mercator 14:119–135. https://doi.org/10.4215/rm.v14i2.1651
Barros A De, Bacani VM, Borges TA (2017) Análise da fragilidade ambiental da bacia hidrográfica do córrego guanandy, Aquidauana - MS. Revista Eletrônica da Associação dos Geoógrafos Brasileiros 26:16
Calaboni A, Tambosi LR, Igari AT, et al (2018) The forest transition in São Paulo, Brazil: Historical patterns and potential drivers. Ecology and Society 23:. https://doi.org/10.5751/ES-10270-230407
Campos JA, Aires URV, DA SILVA DD, Calijuri ML (2019) Environmental fragility and vegetation cover dynamics in the Lapa Grande state park, MG, Brazil. An Acad Bras Cienc 91:1–16. https://doi.org/10.1590/0001-3765201920170940
Castro RA de (2016) Fragilidade ambiental na bacia do córrego água branca, oeste maranhense. Inter Espaço 2:221–234
Chen J, Wang J, Baležentis T, et al (2018) Multicriteria approach towards the sustainable selection of a teahouse location with sensitivity analysis. Sustainability (Switzerland) 10:1–17. https://doi.org/10.3390/su10082926
Corrêa EA, Moraes IC, Pinto S dos AF (2015) Estimativa da Erodibilidade e Tolerância de Perdas de Solo na Região do Centro Leste Paulista. Geociencias 34:848–860
Correchel V (2003) Avaliação de índices de Erodibilidade do Solo através da Técnica da Análise da Redistribuição do “fallout” do Cs. Dissertação (Doutorado) 79
Degife A, Worku H, Gizaw S, Legesse A (2019) Land use land cover dynamics, its drivers and environmental implications in Lake Hawassa Watershed of Ethiopia. Remote Sens Appl 14:178–190. https://doi.org/10.1016/j.rsase.2019.03.005
Demarchi JC, Zimback CRL (2014) MAPEAMENTO , ERODIBILIDADE E TOLERÂNCIA DE PERDA DE SOLO NA SUB-BACIA DO RIBEIRÃO DAS PEROBAS. Revista Energia na Agriultura 29:102–114
Du J, Fu Q, Fang S, et al (2019) Effects of rapid urbanization on vegetation cover in the metropolises of China over the last four decades. Ecol Indic 107:105458. https://doi.org/10.1016/j.ecolind.2019.105458
Eastman RJ (2015) Terrset Manual
EMBRAPA (2006) Sistema Brasileiro de Classificação de Solos, 2^o. Centro Nacional de Pesquisa de Solos, Rio de Janeiro - RJ
Fernandez P, Delgado E, Lopez-Alonso M, Poyatos JM (2018) GIS environmental information analysis of the Darro River basin as the key for the management and hydrological forest restoration. Science of the Total Environment 613–614:1154–1164. https://doi.org/10.1016/j.scitotenv.2017.09.190
Ferraz SFB, Ferraz KMPMB, Cassiano CC, et al (2014) How good are tropical forest patches for ecosystem services provisioning ? Landsc Ecol 29:187–200. https://doi.org/10.1007/s10980-014-9988-z
Ferreira IJM, Bragion G da R, Ferreira JHD, et al (2019) Landscape pattern changes over 25 years across a hotspot zone in southern Brazil. South For 81:175–184. https://doi.org/10.2989/20702620.2018.1542563
Fushita ÂT, Camargo-Bortolin LHG, Arantes EM, et al (2010) Fragilidade ambiental associada ao risco potencial de erosão de uma área da região geoeconômica médio mogi guaçu superior (SP). Revista Brasileira de Cartografia 63:477–488
Ghomi-Avili A, Akbarinia M, Hosseini SM, et al (2020) Fuzzy and Boolean operation based modelling for evaluation of ecological capability in the hyrcanian forests. J For Sci (Prague) 66:170–184. https://doi.org/10.17221/130/2019-JFS
Gribov A, Krivoruchko K (2020) Empirical Bayesian kriging implementation and usage. Science of the Total Environment 722:137290. https://doi.org/10.1016/j.scitotenv.2020.137290
Guth PL, Van Niekerk A, Grohmann CH, et al (2021) Digital elevation models: Terminology and definitions. Remote Sens (Basel) 13:1–19. https://doi.org/10.3390/rs13183581
Hao R, Yu D, Liu Y, et al (2017) Impacts of changes in climate and landscape pattern on ecosystem services. Science of the Total Environment 579:718–728. https://doi.org/10.1016/j.scitotenv.2016.11.036
Havlíček M, Dostál I, Pavelková R (2022) Water Reservoirs as a Driver of Anthropogenic Changes in Landscape and Transport Networks: The Czech Republic Experience. Water (Switzerland) 14:. https://doi.org/10.3390/w14121870
Huang C, Huang X, Peng C, et al (2019) Land use/cover change in the Three Gorges Reservoir area, China: Reconciling the land use conflicts between development and protection. Catena (Amst) 175:388–399. https://doi.org/10.1016/j.catena.2019.01.002
Hunt JD, Nascimento A, Caten CS ten, et al (2022) Energy crisis in Brazil: Impact of hydropower reservoir level on the river flow. Energy 239:1–28. https://doi.org/10.1016/j.energy.2021.121927
IBGE (2012) Manual Técnico da Vegetação Brasileira
IBGE (2007) Manual Técnico de Pedologia, 2°. Rio de Janeiro - RJ
Jiang H, Eastman JR (2000) Application of fuzzy measures in multi-criteria evaluation in GIS. International Journal of Geographical Information Science 14:173–184. https://doi.org/10.1080/136588100240903
Jones IL, Bunnefeld N, Jump AS, et al (2016) Extinction debt on reservoir land-bridge islands. Biol Conserv 199:75–83. https://doi.org/10.1016/j.biocon.2016.04.036
Jones SK, Boundaogo M, DeClerck FA, et al (2019) Insights into the importance of ecosystem services to human well-being in reservoir landscapes. Ecosyst Serv 39:100987. https://doi.org/10.1016/j.ecoser.2019.100987
Krivoruchko K, Gribov A (2020) Distance metrics for data interpolation over large areas on Earth’s surface. Spat Stat 35:100396. https://doi.org/10.1016/j.spasta.2019.100396
Lamb D (2018) Undertaking large-scale forest restoration to generate ecosystem services. Restor Ecol 26:657–666
Lombardi Neto F, Moldenhauer WC (1992) Rainfall erosivity - its distribution and relationship with soil loss at Campinas, Sao Paulo. Bragantia 51:189–196. https://doi.org/10.1590/s0006-87051992000200009
Lucas-Borja ME, Zema DA, Antonio Plaza-álvarez P, et al (2019) Effects of different land uses (abandoned farmland, intensive agriculture and forest) on soil hydrological properties in Southern Spain. Water (Switzerland) 11:1–14. https://doi.org/10.3390/w11030503
Maeda EE, Abera TA, Siljander M, et al (2021) Large-scale commodity agriculture exacerbates the climatic impacts of Amazonian deforestation. Proc Natl Acad Sci U S A 118:. https://doi.org/10.1073/pnas.2023787118
Magnago LFS, Rocha MF, Meyer L, et al (2015) Microclimatic conditions at forest edges have significant impacts on vegetation structure in large Atlantic forest fragments. Biodivers Conserv 24:2305–2318. https://doi.org/10.1007/s10531-015-0961-1
Mannigel AR, Carvalho MDP e, Moreti D, Medeiros LDR (2002) Fator erodibilidade e tolerância de perda dos solos do Estado de São Paulo. Acta Scientiarum 24:1335–1340. https://doi.org/10.4025/actasciagron.v24i0.2374
Mascaro SDA (2004) Evolução Espaço-Temporal Do Uso E Cobertura Do Solo Nas Estâncias Turísticas De Avaré E Paranapanema , No Reservatório De Jurumirim (SP). Master’s Degree, Universidade Estadual Paulista “Júlio de Mesquita Filho” - UNESP
Massa EM, Ross JLS (2012) Aplicação De Um Modelo De Fragilidade Ambiental Relevo-Solo Na Serra Da Cantareira, Bacia Do Córrego Do Bispo, São Paulo-Sp. Geography Department, University of Sao Paulo 24:57–79. https://doi.org/10.7154/rdg.2012.0024.0004
Mello K de, Randhir TO, Valente RA, Vettorazzi CA (2017) Riparian restoration for protecting water quality in tropical agricultural watersheds. Ecol Eng 108:514–524. https://doi.org/10.1016/j.ecoleng.2017.06.049
Mello K de, Taniwaki RH, Paula FR de, et al (2020) Multiscale land use impacts on water quality: Assessment, planning, and future perspectives in Brazil. J Environ Manage 270:110879. https://doi.org/10.1016/j.jenvman.2020.110879
Miguel P, Dalmolin RSD, Moura-Bueno JM, et al (2021) Mapping of erodibility and potential soil erosion in a hillside watershed. Engenharia Sanitaria e Ambiental 26:1–9. https://doi.org/10.1590/s1413-415220190235
Moonjun R, Shrestha DP, Jetten VG (2020) Fuzzy logic for fine-scale soil mapping: A case study in Thailand. Catena (Amst) 190:104456. https://doi.org/10.1016/j.catena.2020.104456
Nearing MA, Yin S qing, Borrelli P, Polyakov VO (2017) Rainfall erosivity: An historical review. Catena (Amst) 157:357–362. https://doi.org/10.1016/j.catena.2017.06.004
Oliveira PTS, Rodrigues DBB, Sobrinho TA, et al (2012) Spatial Variability Of The Rainfall Erosive Potential In The State Of Mato Grosso Do Sul, Brazil. Engenharia Agricola 32:69–79. https://doi.org/10.1590/S0100-69162012000100008
Oliveira PTS, Wendland E, Nearing MA (2013) Rainfall erosivity in Brazil: A review. Catena (Amst) 100:139–147. https://doi.org/10.1016/j.catena.2012.08.006
Oliveira-Andreoli EZ, Moraes MCP de, Faustino A da S, et al (2021) Multi-temporal analysis of land use land cover interference in environmental fragility in a Mesozoic basin, southeastern Brazil. Groundw Sustain Dev 12:. https://doi.org/10.1016/j.gsd.2020.100536
Padedda BM, Sechi N, Lai GG, et al (2017) Consequences of eutrophication in the management of water resources in Mediterranean reservoirs: A case study of Lake Cedrino (Sardinia, Italy). Glob Ecol Conserv 12:21–35. https://doi.org/10.1016/j.gecco.2017.08.004
Qiu F, Chastain B, Zhou Y, et al (2014) Modeling land suitability/capability using fuzzy evaluation. GeoJournal 79:167–182. https://doi.org/10.1007/s10708-013-9503-0
Ross JLS (1994) Análise Empírica Da Fragilidade Dos Ambientes Naturais Antropizados. Geography Department, University of Sao Paulo 63–74
Santos AR dos, Araújo EF, Barros QS, et al (2020) Fuzzy concept applied in determining potential forest fragments for deployment of a network of ecological corridors in the Brazilian Atlantic Forest. Ecol Indic 115:106423. https://doi.org/10.1016/j.ecolind.2020.106423
Spörl C, Ross JLS (2004) Análise comparativa da fragilidade ambiental com aplicação de três modelos. GEOUSP Espaço e Tempo 15:39–49
Storto C, Cocato GP (2018) Análise de Fragilidade Ambiental a partir de Técnicas de Geoprocessamento: Área de Influência da Hidrelétrica de Mauá - PR. Revista Brasileira de Geografia Física 11:1694–1708
Tiang DCF, Morris A, Bell M, et al (2021) Ecological connectivity in fragmented agricultural landscapes and the importance of scattered trees and small patches. Ecol Process 10:. https://doi.org/10.1186/s13717-021-00284-7
Valente RA, de Mello K, Metedieri JF, Américo C (2021a) A multicriteria evaluation approach to set forest restoration priorities based on water ecosystem services. J Environ Manage 285:. https://doi.org/10.1016/j.jenvman.2021.112049
Valente RA, de Mello K, Metedieri JF, Américo C (2021b) A multicriteria evaluation approach to set forest restoration priorities based on water ecosystem services. J Environ Manage 285:. https://doi.org/10.1016/j.jenvman.2021.112049
Valle IC, Francelino MR, Pinheiro HSK (2016) Mapeamento da Fragilidade Ambiental na Bacia do Rio Aldeia Velha, RJ TT - Mapping of Environmental Fragility in the Aldeia Velha River Basin, State of Rio de Janeiro, Brazil. Floresta e Ambiente 23:295–308. https://doi.org/10.1590/2179-8087.107714
Vettorazzi CA, Valente RA (2016) Priority areas for forest restoration aiming at the conservation of water resources. Ecol Eng 94:255–267. https://doi.org/10.1016/j.ecoleng.2016.05.069
Widiatmaka (2016) Integrated use of GIS , AHP and remote sensing in land use planning for tropical high altitude vegetable crops. Journal of Applied Horticulture 18:87–99
Wu J, Huang J, Han X, et al (2003) Three-Gorges Dam--Experiment in Habitat Fragmentation? Science (1979) 300:1239–1240. https://doi.org/10.1126/science.1083312
Zadeh LA (1965) Fuzzy Sets. Information and Control 8:338–353. https://doi.org/10.1061/9780784413616.194

No competing interests reported.

Appendix1.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Mapping Environmental Fragility in Anthropized Landscapes using Fuzzy Logic and continuous data spatialization

Status:

Version 1

Abstract

Figures

Introduction

Material and Methods

Study Area

Conceptual Model

Spatial Data and Landscape Attribute Maps

Environmental Fragility Analysis

Results

Landscape Attribute Maps

Environmental Fragility

Discussion

Conclusion

Declarations

Statements & Declarations

Author Contributions

References

Additional Declarations

Supplementary Files

Status:

Version 1