Mainly occurring in tropical forests, deforestation is a worldwide environmental threat, particularly considering associated factors of climate change and loss of biodiversity. The Brazilian MapBiomas project developed a free and open-source platform to monitor Brazilian land cover changes since 1985. To increase its analytic power with respect to land cover changes, we developed an Environmental Fragility Mapping System (EFMS), using Google Earth Engine (GEE). Environmental fragility (EF) analysis has its origins in ecodynamics, and it is composed of potential environmental fragility (PEF) and emergent environmental fragility (EEF), arising from the effects of land-use and land-cover (LULC) changes on PEF. The main advantage of EF analysis over LULC analysis is its flexibility in evaluating multiple scenarios for different aspects of environmental challenges. In EFMS, we calculate three environmental fragility indexes: (i) potential fragility index (PFI), (ii) land cover fragility index (LCFI), and (iii) emergent fragility index (EFI). Combined with LULC change analysis from the MapBiomas Project, EFMS can provide spatial analysis and generate maps and data for environmental management.
Article
Cloud-based application for generate spatial and statistical data about environmental fragility and land cover
https://doi.org/10.21203/rs.3.rs-3591460/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
Mainly occurring in tropical forests (Atlantic Rainforest and the Amazon), deforestation is a worldwide environmental threat 1, particularly considering the associated factors of climate change and loss of biodiversity. The Brazilian Annual Land-Use and Land-Cover Mapping Project (MapBiomas) 2 developed a free and open-source platform to monitor Brazilian land cover changes since 1985. With the MapBiomas platform (https://plataforma.brasil.mapbiomas.org/), it is possible to analyze annual land cover change and quantify the rate of natural vegetation loss. It is also possible to determine land cover changes. However, land cover changes are not homogeneous given a contiguous mosaic of landscapes, and the impact of change is also different, depending on physical and anthropic aspects of landscapes. Therefore, to increase the analytic power of land-use and land-cover LULC, we created the Environmental Fragility Mapping System (EFMS) linking LULC change analysis to real-time physical characteristics of the landscape. Developed from Google Earth Engine (GEE)3 and based on MapBiomas datasets2, Landsat Normalized Difference Vegetation Index (NDVI), road and hydrographic networks, and terrain metrics, EFMS is a customizable application to generate raster datasets, charts, and tabular data, embracing the environmental fragility concept4–10.
Environmental fragility (EF) analysis originates from ecodynamics4 and consists of potential environmental fragility (PEF), representing physical attributes of the landscape (e.g., terrain metrics, hydrologic networks, climate and soil) and emergent environmental fragility (EEF), arising from the effects of land-use and land-cover (LULC) changes on PEF8. EF analysis is superior to LULC change analysis by its flexibility in evaluating multiple scenarios for different aspects of environmental challenges, such as conservation of natural ecosystems5, prioritization of fragments for landscape restoration11,12, soil erosion13, natural hazards 9, and monitoring EF changes10.
In EFMS, we calculate three environmental fragility indexes: (i) potential fragility index (PFI), (ii) land cover fragility index (LCFI), and (iii) emergent fragility index (EFI). PFI represents physical fragility of the environment caused by the factors of relief and hydrography. LCFI represents land-cover fragility by LULC classes of MapBiomas, Landsat Normalized Difference Vegetation Index (NDVI) and road networks. EFI is the result of combination between PFI and LCFI.
To implement the concept of EF in EFMS, we combined the spatial analysis power of Google Earth Engine (GEE) 3 with a multi-criteria approach, using the analytic hierarchy process (AHP) 14 to assign weights to MapBiomas LULC classes considering the degree of environmental fragility. We also adopted AHP to assign default weights for spatial data through aggregating the knowledge of a multidisciplinary research team. Notably, EFMS enables the user to change all default weights and create other parametrizations, depending on need; in other words, it is customizable.
EFMS can perform spatial analysis and generate maps and data for environmental fragility analysis for multiple objectives, such as legal compliance, ecological corridor design, environmental recovery planning, soil loss, and landslide and flood fragility mapping.
In this article we present two EFMS use cases: one for environmental compliance with Brazilian legislation, selecting environmental recovery priority areas within permanent preservation areas (PPA), as defined by the Brazilian National Forest Code20; and other for designing ecological corridors. Both were applied to the municipality of Alfredo Wagner, in Santa Catarina, Brazil.
According to MapBiomas (https://plataforma.brasil.mapbiomas.org/), from 1985 to 2022, the state of Santa Catarina lost 999,412ha of natural habitat to farming, while 468,289ha of farming land were recovered to forest or non-forest natural formation (Table 1). This dynamic shows that Santa Catarina is still losing natural areas to anthropic activities, mainly farming.
| 2022 | ||||||||
|---|---|---|---|---|---|---|---|---|
| Class | Forest | Non- forest Natural Formation | Farming | Non- vegetated Area | Water | Not observed | Total 1985 | |
| 1985 | Forest | 3343736 | 111 | 854523 | 16578 | 14293 | 1 | 4229243 |
| Non- forest Natural Formation | 201 | 491293 | 144889 | 2751 | 1344 | 640477 | ||
| Farming | 440059 | 28230 | 3851978 | 122029 | 20869 | 4463164 | ||
| Non- vegetated Area | 707 | 1029 | 8707 | 68945 | 386 | 79774 | ||
| Water | 4125 | 1025 | 7112 | 1058 | 103760 | 117080 | ||
| Not observed | 6 | 0 | 110 | 116 | ||||
| Total 2022 | 3788828 | 521687 | 4867214 | 211360 | 140653 | 111 | 9529853 | |
The municipality of Alfredo Wagner is a representative example of ecodynamics in Santa Catarina. Alfredo Wagner is environmentally heterogeneous with altitude ranging from 400 to 1,200m above sea level. The original vegetation of Atlantic Rainforest, Araucária Forest, Grasslands and Faxinal Forests21 was replaced by pasture, temporary crops and forest plantation (non-native species, i.e., Pinnus elliotti) in the 20th century (Fig. 1). According to MapBiomas (https://plataforma.brasil.mapbiomas.org/), from 1985 to 2022, the municipality lost 6,238ha of forest formation and 3,387 of grassland for pasture, agriculture, and forest plantation. In the same period, 1,121ha of pasture, 2,238ha of agriculture and 229ha of forest plantation were recovered to forest formation, and 279ha of pasture, 40ha of agriculture and 5ha of forest plantation were recovered to grassland (Fig. 1).
Figure 1. Sankey chart of LULC (level 2) changes in Alfredo Wagner from 1985 to 2022. Source: MapBiomas (https://plataforma.brasil.mapbiomas.org/)
To illustrate the practical application of EFMS, we selected and prioritized areas (PA) for environmental restoration in Alfredo Wagner, considering two objectives: (i) to identify priority sites for environmental recovery in rural areas to be in compliance with the legal code that defines PPA and (ii) to identify PA to create ecological corridors (EC) for environmental recovery.
Objective 1
To identify priority sites for environmental recovery in rural areas to be in compliance with the legal code that defines PPA.
In EFMS, we select the Alfredo Wagner municipality as AOI and the year as 2022, followed by calculating EFI using all default values (Fig. 2).
Figure 2 . 2022 EFI map and area chart of Alfredo Wagner, SC, generated through EFMS.
The result shows that Alfredo Wagner has more than 30% of territory with high EFI (Fig. 2, Table 2). This percentage is higher (43.3%) within the PPA (Fig. 3, Table 2). To better explore the results, users can overlay EFI maps and Google´s high-resolution images and apply transparent levels of drawing (Fig. 3).
Figure 3. 2022 EFI map and area chart of Alfredo Wagner’s PPA, SC, as generated through EFMS, to identify priority sites for environmental recovery in rural areas in order to comply with the Brazilian National Forest Code. In the detailed area, we overlayed the EFI and high-resolution satellite images to show three highly fragile environmental areas: (1) deforested area with high slope; (2) agricultural activity on hilltops, and (3) construction of a small hydroelectric plant in a riparian area with a high slope.
Alfredo Wagner has 6,056ha of high and very high EFI within the PPA (Table 2). These areas are concentrated in valleys and primarily represent riparian regions that should be prioritized for environmental recovery (Fig. 3).
| EFI | Area – AOI (ha) | Area – PPA (ha) |
|---|---|---|
| Very low | 1663 | 232 |
| Low | 20562 | 1626 |
| Intermediate | 28581 | 5956 |
| High | 22058 | 6002 |
| Very high | 205 | 54 |
In some sites, such as those indicated in the detailed satellite image in Fig. 3, it is possible to identify agricultural activities on hilltops, deforested high slope, and riparian areas. To obtain detailed data from specific points, the user can click over the map. By clicking on the map, the retrieved detailed data include the Brazilian Rural Environmental Registry (CAR), the mandatory and self-declaratory registry for rural properties, altitude, slope, TPI, distance from hydrography, NDVI, LCFI, PFI and EFI. By clicking on points 1, 2 and 3, respectively, in the detailed area of Fig. 3, the detailed data are:
Point 1: Lon: -49.3093 Lat: -27.8262; Cód. CAR: SC-4200705-62912EBF2ADD4F5684BC39515CAE64BE; Altitude: 1413.00 m; Slope: 57.02 deg.; TPI: 0.42; Hydro dist.: 108.32 m; NDVI: 0.10; LCFI: 0.460; PFI: 0.798; EFI: 0.369.
Point 2: Lon: -49.3093 Lat: -27.8262; Cód. CAR: SC-4200705-8CBA8F19AC6D4CEC84B5004FE12D3D4D; Altitude: 1546.00 m; Slope: 35.32 deg.; TPI: 0.71; Hydro dist.: 190.08 m; NDVI: 0.17; LCFI: 0.467; PFI: 0.705; EFI: 0.329
Point 3: Lon: -49.3093 Lat: -27.8262; Cód. CAR: SC-4200705-A4FD716E00DA4B1BAE2CDC75540E1B74; Altitude: 638.00 m; Slope: 26.46 deg.; TPI: 0.31; Hydro dist.: 17.09 m; NDVI: 0.12; LCFI: 0.538; PFI: 0.623; EFI: 0.316
Objective 2
To identify priority areas to create ecological corridors (EC) for environmental recovery.
To achieve this objective, we change the values of PA to create connective corridors based on hydrography, slope, and topographic position. We change the values of distance from hydrography to 0-50m, slope to 30-90o and TPI to 0.5-1 to select areas within 50m of rivers, areas with more than 30o slope, and areas with upper slopes and hilltops. We also changed the weight of each physical variable to 0.35 for distance from hydrography and slope and 0.3 for TPI. Finally, we changed the weight of forest plantation LULC class from 0.05 to 0.23, the same weight as that of other agriculture and monoculture classes (Fig. 4).
Figure 4. EFMS weight customization to identify priority areas to create ecological corridors for environmental recovery in Alfredo Wagner, SC.
Considering this scenario, the areas with high and very high EFI in Alfredo Wagner total 14,669ha (Fig. 5, Table 3).
Figure 5. 2022 EFI map and area chart of Alfredo Wagner, SC, as generated through EFMS, to identify priority areas to create EC for environmental recovery.
To create an EC network in Alfredo Wagner, we would start prioritizing 8,197ha of high EFI for natural vegetation recovery (Fig. 6, Table 3). Observing the detailed area of EFI overlayed with high-resolution satellite image (Fig. 6), it is possible to perceive that environmental fragility does not represent vegetation coverage, but rather the integration of LULC and potential environmental fragility by relief and hydrography. Some areas with intermediate or low EFI can be covered by natural vegetation. On the other hand, some areas with high EFI may appear to be covered in natural vegetation, but are, instead, covered by non-native planted forests.
Figure 6. 2022 EFI map and area chart of Alfredo Wagner, SC, as generated through the EFMS, to identify priority areas to create EC for environmental recovery. In the detailed area, we overlayed the EFI results and high-resolution satellite images to show where natural vegetation recovery should be prioritized, considering connective ecological corridors and environmental fragility.
| EFI | Area – AOI (ha) | Area – EC (ha) |
|---|---|---|
| Very low | 1867 | 689 |
| Low | 26526 | 7763 |
| Intermediate | 30005 | 12745 |
| high | 14554 | 8141 |
| Very high | 115 | 56 |
To reverse the loss of natural areas in Santa Catarina, it is necessary to adopt environmental recovery policies. The focus of new policies for environmental recovery should be based on territorial planning rather than legal or socioeconomic mechanisms22. New policies should also consider all natural LULC classes as environmentally fragile and important to maintenance and recovery, not only forest formations23. In this effort, EFMS is designed to aid scientists, managers, politicians, and stakeholders in identifying and selecting priority sites and then calculating the area to be recovered.
The Brazilian National Forest Code20 defines PPA based on proximity to hydrography (riparian areas), terrain slope and topographic position. According to the Forest Code, PPA should be covered by natural vegetation in 100% of its area. In the first objective, we used the legal definition of PPA to set the default values for distance from hydrography (0-30m), slope (45-90o) and TPI (0.6-1). The resultant default values generate PA close to those defined by legislation, but the spatial resolution of 30m and the 1:50,000 scale of the raw data were unsuited for assessing PPA in larger scales.
In the second objective, we reduced the weight of distance from hydrography by distributing the weight of physical variables comprising PFI. This resulted in reducing the importance of riparian areas, as defined by Brazilian law, by considering both relief aspects and riparian areas as equally important areas for EC. At the same time, by increasing the weight of the LULC class termed as forest plantation, we defined as a monoculture that contributes to biodiversity lost.
With this new parametrization, EC should contribute to biodiversity to a degree exceeding that of legislation. EFMS reflects the importance of evaluating fragility in a designed connective network of forest fragments over the Alfredo Wagner territory.
This kind of analysis helps to plan for reforestation and the connectivity of EC by prioritizing areas, defining the species to be utilized on natural vegetation recovery, calculating the number of seedlings needed and designing the species composition that matches unique environmental characteristics, such as hydrography, slope, sunlight, shadows, and the like.
The environmental fragility concept can be widely applied 4–10. The two Alfredo Wagner, SC examples showed the potential for evaluating environmental fragility for different objectives. Combined with MapBiomas LULC change analysis, EFMS is a user-friendly tool for spatial analysis and the generation of maps and data for environmental management.
EFMS was implemented in GEE3 using JavaScript. Two versions of EFMS code were written, one in English and other in Brazilian Portuguese. The English version of EFMS is available at https://ee-vianna.projects.earthengine.app/view/efi (Fig. 7), while the Portuguese version can be found at https://ee-vianna.projects.earthengine.app/view/ife. From both links, the user manual of EFMS is also available, but only in Brazilian Portuguese. The user manual (https://doi.org/10.5281/zenodo.10033619) details the method and presents some practical examples for using the EFMS to generate maps and export charts and tabular data.
The project was developed for Santa Catarina State, Brazil (Fig. 7); however, the code is open-source and can be adopted worldwide. To do this, it is necessary to overwrite the areas of interest (AOI), the LULC map, the hydrography and the OpenStreetMap (OSM) road maps with values of the new area for the same parameters. The system is spatial scale sensitive. Larger spatial scales of input data generate more accurate outputs. For the Santa Catarina State project, the EFMS maximum scale is 1:50,000, the highest scale for Landsat products15,16.
Figure 7. English version of EFMS. The layout is divided into three parts, two side panels and map. The main panel (left) contains a brief description of the system, link to the user manual, tools for parameterizing the PA (priority area), and fragility index calculation models (PFI, LCFI and EFI). The panel on the right side shows the legends and graphs of area per mapped class. The map shows all layers generated. In this example, the 2022 LULC layer of Santa Catarina is drawn.
The method for calculating the indexes can be divided into three steps (Fig. 8), including generating terrain metrics, NDVI, and distance maps based on raw data (Step 1), min-max scaling the raster datasets to 0–1 (Step 2), and defining weights for LULC classes and for min-max scaled raster and applying map algebra to calculate indexes (Step 3).
Figure 8. Workflow of method.
To run EFMS, the user first selects the area of interest (AOI), defines the parameters for mapping the priority areas (PA), selects the year of analysis and then weights the land cover classes and variables that constitute the indexes. The system executes all calculations for indexes each time the "Calculate EFI" button is pressed. The processing speed is proportional to the AOI. Large areas slow down the processing.
Raw data. We used the Shuttle Radar Topographic Mission (SRTM) digital elevation model (DEM)17, Landsat 5, 7 and 8, and LULC data from the Brazilian Annual LULC Mapping Project (MapBiomas Collection 8; https://brasil.mapbiomas.org/colecoes-mapbiomas/) from 1985 to today. The 30m spatial resolution of raw data defines the output spatial resolution of indexes. We also used the 1:10,000 hydrography network18 and the Open Street Map (https://www.openstreetmap.org/) road network for Santa Catarina State to calculate Euclidean distance.
Step 1. In the first step, two terrain metrics, NDVI, two Euclidean distances and the PA mask raster datasets are generated. The system clips the data for the selected AOI. From SRTM-DEM, the system generates slope and Topographic Position Index (TPI)19 raster datasets. From Landsat images, it generates the median NDVI, using Landsat 5 data for years between 1985 and 2011, Landsat 7 data for 2012, and Landsat 8 data for years beyond 2012. From the MapBiomas raster dataset, the system selects the year and clips the LULC raster dataset. From the hydrography network and road network maps, the system calculates two Euclidean distance raster datasets. The system also generates the PA mask raster dataset according to values defined by the user. By default, the values for altitude are 0-1900m, the range found in the state of Santa Catarina. The parametrization of elevation enables the user to calculate the indexes by range of altitude. The default parametrization values are also 0-30m for distance of hydrography, 0-45o for slope and 0.6-1 for TPI. These are the values that best represent the permanent preservation areas (PPA), as defined by the Brazilian National Forest Code20. However, no parametrization values are fixed, allowing the user to customize masks. At the end of the first step, seven variables are generated and added as eight layers to the map: altitude, slope, TPI, distance from hydrography, distance from roads, NDVI, land cover and land cover within PA.
Step 2. In this step, the raster datasets are min-max scaled for 0–1, according to the environmental fragility of each variable. Min-max scaling is linear and considers the minimum and maximum values of each variable within the AOI, which makes the index sensitive to the scale defined by the AOI. Min-max scaling is calculated as
$${x}_{scaled}=\frac{x-{x}_{max}}{{x}_{max}-{x}_{min}}$$
In scaled variables, values close to 0 represent lower environmental fragility, while values close to 1 indicate higher fragility. The variables Slope and TPI are directly proportional to fragility. This means that higher values of slope or TPI correlate with higher environmental fragility. The variables NDVI, distance from hydrography and distance from roads are inversely proportional to environmental fragility. Lower values of NDVI and distances from hydrography correlate with higher environmental fragility. After min-max scaling, NDVI and distances are inverted, multiplying by -1 and adding 1. This is necessary to adjust values for the adopted scale.
Also in this step, MapBiomas land cover classes are weighted. The weight of each class is directly proportional to environmental fragility. The default weights were defined by a multidisciplinary team of researchers and experts, applying the Analytic Hierarchy Process (AHP)14. In Table 4, we show the weight of each LULC class. The natural LULC classes were considered as the lowest environmental fragility and weighted to zero.
| LULC classes | Weight |
|---|---|
| Forest Formation | 0 |
| Savanna Formation | 0 |
| Mangrove | 0 |
| Wooded Sandbank Vegetation | 0 |
| Wetland | 0 |
| Grassland | 0 |
| Salt Flat | 0 |
| Rocky Outcrop | 0 |
| Herbaceous Sandbank Vegetation | 0 |
| Other Non-forest Formations | 0 |
| Pasture | 0.15 |
| Agriculture | 0.23 |
| Temporary Crop | 0.23 |
| Soybean | 0.32 |
| Sugar Cane | 0.23 |
| Rice | 0.42 |
| Cotton | 0.23 |
| Other Temporary Crops | 0.23 |
| Perennial Crop | 0.09 |
| Coffee | 0.09 |
| Citrus | 0.09 |
| Other Perennial Crops | 0.09 |
| Forest Plantation | 0.05 |
| Mosaic of Uses | 0.23 |
| Beach, Dune, and Sand Spot | 0 |
| Urban Area | 0.83 |
| Mining | 1 |
| Other Non-vegetated Areas | 0.68 |
| River,Lake and Ocean | 0 |
| Aquaculture | 0.54 |
| Not Observed | 1 |
Step 3. In the last step, the system calculates PFI, LCFI and EFI. PFI is calculated by the weighted average among the scaled datasets of slope, TPI and distance from hydrography. By default, the adopted weights are 0.4 for distance from hydrography, 0.35 for slope and 0.25 for TPI. LCFI is calculated by the weighted average among the scaled datasets of MapBiomas LULC, NDVI and distance from roads. By default, the adopted weights are 0.4 for the MapBiomas LULC, 0.35 for NDVI and 0.25 for distance from roads. EFI is the IFP multiplied by the IFSC. The result is a raster dataset with values between 0 and 1 where 0 is the lowest fragility and 1 the highest. The scale of fragility is represented in Table 5.
Table 5: Weight of variables, scale, and colors of PFI, LCFI and EFI.
EFI classification:
0– 0.1: EFI very low
0.1– 0.2: EFI low
0.2 – 0.25: EFI intermediate
0.25 – 0.4: EFI high
0.4 – 1: EFI very high
Data availability
All of the raw data used by EFMS are from public datasets. The Shuttle Radar Topographic Mission (SRTM) digital elevation model (DEM)17 and the Landsat 5, 7 and 8 images are available from GEE datasets (https://developers.google.com/earth-engine/datasets), that can be accessed by any user with an account.
Land use land cover data from the Brazilian Annual LULC Mapping Project (MapBiomas Collection 8; https://brasil.mapbiomas.org/colecoes-mapbiomas/) from 1985 to today can be accessed by GEE dataset link (https://code.earthengine.google.com/?asset=projects/mapbiomas-workspace/public/collection8/mapbiomas_collection80_integration_v1)
Spatial and statistical data about environmental fragility and land cover, from 1985 to today, for the Santa Catarina State and other political and environmental subdivisions are available in EFMS English (https://ee-vianna.projects.earthengine.app/view/efi), or Portuguese version (https://ee-vianna.projects.earthengine.app/view/ife). Data generated by EFMS can be analyzed and viewed in the system or exported as a table (csv) or chart (png, svg). Since each analysis generates a specific dataset, EFMS itself can be considered a data repository.
Code availability
All of our codes are available from Zenodo (https://doi.org/10.5281/zenodo.10018501) under Creative Commons Attribution 4.0 International. In the Zenodo repository, users will find the freely available codes in JavaScript and a user manual. AOI, hydrography and roads data are for Santa Catarina State, Brazil. To run the code for other regions, it is necessary to overwrite these data for local ones in GEE assets. We suggest using data at higher scale of 1:50,000.
Acknowledgements
This study was financed by the Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina (FAPESC) – Finance Code TO 2021TR001397. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We thank the editor and reviewers whose helpful comments and suggestions consistently helped to improve the clarity and robustness of the data and analyses presented in this manuscript.
Author Contributions
L.F.N.V. designed and implemented the method, write the JavaScript code and wrote the first version of the manuscript. F.M.Z. reviewed the literature and contributed on definition of default weights for land-use-land-cover classes and environmental factors by analytical hierarchy process.
Competing Interests
The author(s) declare no competing interests.
- Brondízio, E. S., Settele, J., Díaz, S. & Ngo, H. T. (eds). IPBES (2019), Global assessment report of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Debating Nature’s Value (2019).
- Souza, C. M. et al. Reconstructing three decades of land use and land cover changes in brazilian biomes with landsat archive and earth engine. Remote Sens (Basel) 12, (2020).
- Gorelick, N. et al. Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sens Environ 202, 18–27 (2017).
- Tricart, J. Ecodinâmica. (IBGE, 1977).
- França, L. C. de J. et al. Environmental Fragility Zoning Using GIS and AHP Modeling: Perspectives for the Conservation of Natural Ecosystems in Brazil. Conservation 2, 349–366 (2022).
- Rosa, M. R. & Ross, J. L. Aplicação De Sig Na Geração De Cartas De Fragilidade. Revista do Departamento de Geografia 13, 77–106 (1999).
- de Jesus França, L. C. et al. Ahp Approach Applied To Multi-Criteria Decisions In Environmental Fragility Mapping. Floresta 50, 1623–1632 (2020).
- Ross, J. L. S. Análise Empírica da Fragilidade dos Ambientes Naturais Antropizados. Revista do Departamento de Geografia 8, 63–74 (1994).
- Roberts, N. J., Nadim, F. & Kalsnes, B. Quantification of vulnerability to natural hazards. Georisk 3, 164–173 (2009).
- Xia, M. et al. Spatio-temporal changes of ecological vulnerability across the Qinghai-Tibetan Plateau. Ecol Indic 123, 107274 (2021).
- De Matos, T. P. V., De Matos, V. P. V., De Mello, K. & Valente, R. A. Protected areas and forest fragmentation: sustainability index for prioritizing fragments for landscape restoration. Geology, Ecology, and Landscapes 5, 19–31 (2021).
- Valente, R. A., de Mello, K., Metedieri, J. F. & Américo, C. A multicriteria evaluation approach to set forest restoration priorities based on water ecosystem services. J Environ Manage 285, (2021).
- Guo, L. et al. Multiscale spatiotemporal characteristics of landscape patterns, hotspots, and influencing factors for soil erosion. Science of the Total Environment 779, 146474 (2021).
- Saaty, R. W. The analytic hierarchy process-what it is and how it is used. Mathematical Modelling 9, 161–176 (1987).
- Welch, R., Jordan, T. R. & Ehlers, M. Comparative evaluations of the geodetic accuracy and cartographic potential of Landsat-4 and Landsat-5 Thematic Mapper and image data. Photogramm Eng Remote Sensing 51, 1249–1262 (1985).
- Cintra, J. P., César, J., Politécnica, E. & Paulo, D. S. Avaliação da exatidão cartográfica de documentos atualizados com imagens orbitais e sistemas de informação geográfica. in O Espaço Sem Fronteiras 7 (Sociedade Brasileira de Cartografia, 1994).
- Farr, T. & Kobrick, M. The shuttle radar topography mission. Rev. Geophys. 45, 1–33 (2007).
- SDS. Aerolevantamento do estado de Santa Catarina - Relatório de Produção Final. (2012).
- Jenness, J. TPI Documentation. Topographic Position Index (TPI) v. 1.2 16, 1–42 (2006).
- Brasil. Lei n. 12.651, de 25 de maio de 2012. Dispõe sobre a proteção da vegetação nativa; altera as Leis nos 6.938, de 31 de agosto de 1981, 9.393, de 19 de dezembro de 1996, e 11.428, de 22 de dezembro de 2006; revoga as Leis nos 4.771, de 15 de setembro de 196. Câmara dos Deputados, 38 Preprint at http://www2.camara.leg.br/legin/fed/lei/2012/lei-12651-25-maio-2012-613076-normaatualizada-pl.pdf (2012).
- Klein, R. M. R. M. Mapa fitogeográfico do estado de Santa Catarina. Flora Ilustrada Catarinense (Herbário Barbosa Rodrigues, 1978).
- Souza, J. M. de, Morgado, P., Costa, E. M. da & Vianna, L. F. de N. Predictive Scenarios of LULC Changes Supporting Public Policies: The Case of Chapecó River Ecological Corridor, Santa Catarina/Brazil. Land (Basel) 12, 181 (2023).
- Souza, J. M. de, Morgado, P., Costa, E. M. da & Vianna, L. F. de N. Modeling of Land Use and Land Cover (LULC) Change Based on Artificial Neural Networks for the Chapecó River Ecological Corridor, Santa Catarina/Brazil. Sustainability 14, 4038 (2022).
No competing interests reported.