Modelling Potential Biophysical Impacts of Climate Change in the Atlantic Forest: Closing the Gap to Identify Vulnerabilities in Brazil

Abstract

The lack of information on the climate-related vulnerability of territories, populations, and productive sectors in Brazil—and the ecosystems that they are dependent upon—is a serious constraint to identifying possible strategies for local and regional adaptation to climate change. In order to mainstream adaptation efforts in sectors’ planning and budget demands, the existing knowledge gap between what climate models project and what impacts this might have at the local level must be filled. This is particularly significant in the Atlantic Forest region, where over 70% of Brazil’s population currently resides. In the context of the Biodiversity and Climate Change project in the Atlantic Rainforest, this article focusses on the main results of a spatially explicit climate-impact model, which forecasts trends in soil humidity, natural vegetation, agro-climatic zoning, erosion, flooding, landslides, and animal disease vectors, according to optimistic and pessimistic climate change scenarios until 2100. This study is intended as an advisory tool to support the definition of robust and proactive adaptation actions for several sectors in Brazil, framed into the new National Adaptation Plan.

Annex 1

Flood Model

Results describing physical susceptibility to flood occurrence were produced by crossing data about major relief characteristics such as: Declivity; Stream Order; And Vertical and Horizontal Distance to Stream (Strahler 1964; Rennó et al. 2008), as displayed in Fig. 3.17.

Fig. 3.17

Floods submodel flowchart

Extreme precipitation events are closely linked to flood occurrence (Parr et al. 2015; Alfieri et al. 2015; Apurv et al. 2015; Yoon et al. 2015). The indicators of climate extremes selected as model inputs are: Annual count of days with precipitation greater than 10 mm (R10); Annual rainfall during days when precipitation is greater than or equal to the 95th percentile of all humid days of the year, with rainfall values higher than or equal to 1 mm (R95p); Maximum number of consecutive wet days a year (CWD); And Maximum annual rainfall on 5 consecutive days (Rx5 day).

The model results were calibrated and validated by using data from the Sistema Integrado de Informações sobre Desastres (SEDEC) and Atlas Brasileiro de Desastres Naturais (CEPED 2012).

Rainfall Erosivity Model

Figure 3.18 displays the model of rainfall erosivity (i.e., erosive force of rain), based on the equations suggested by Silva (2004); da Silva (2001); Rufino et al. (1993); Neto and Moldenhauer (1992); and Val et al. (1986). Rainfall data include monthly and annual precipitation averages.

Fig. 3.18

Potential soil erosion by water submodel flowchart

Landslides Model

Soil slope and erodibility, together with rainfall erosivity, were used to generate an indicator of the physical sensitivity of the system to landslides. The slope was calculated using the Digital Elevation Model (DEM) for the Topodata-INPE project. The erodibility was classified according to Silva and Alvares (2005) and Lino (2010). The rainfall erosivity characterizes the exposure as described in section “Rainfall Erosivity Model” (Fig. 3.19).

Fig. 3.19

Landslide submodel flowchart

All the information was normalized and crossed using a weighted average: erosivity and erodibility were given a weight 0.3 and slope was given a weight of 0.4, since gravity plays an important role in landslides.

The calibration and validation of the model’s output used records provided by the Civil Defence, CEMADEN and newspaper.

Dengue Vector Distribution Model

In general, the model projection of tropical disease vector distribution was constituted from the geostatistical relationships between the distribution of exposure and climatic extremes observed in periods of high infestation, and the data indicating the presence of the vector in the territory. Thus, it is possible to draw up a spatial representation of potential distribution of the tropical disease vector in the face of climate change, including future exposure data. The result is shown by a composed index pointing the susceptibility of the region to this vector (Aedes aegypti). The first indicator presents the probability of occurrence generated by the MAXENT algorithm (Phillips et al. 2004). The second reports on the likely number of vector generations according to temperature limits (Fernandes et al. 2006). The third indicator sets the transmission potential, also related to temperature data (Lambrechts et al. 2011). The fourth indicator designs positivity for hatching eggs, associated with rainfall, temperature and relative humidity (Vianello et al. 2006). In addition, the fifth indicator is known as General Susceptibility Index, a methodology for quantitative assessment of susceptibility (Brasil 2005). The indicators were crossed as the same weight by map algebra and thus it was possible to draw up a spatial representation through a normalized ratio vector distribution of the potential impact of tropical diseases according to climate change as shown in the following Fig. 3.20.

Fig. 3.20

Vector diseases submodel flowchart

The results of this model were calibrated using data from the international repository DRYAD and validated by comparison to a study from the University of Oxford (Kraemer et al. 2015).

Soil Moisture Model

Soil moisture is typically incremented through rainfall events and lost through runoff, infiltration to deeper soil layers and groundwater, and evapotranspiration. In turn, the evapotranspiration is controlled by vegetation and atmospheric conditions. The dynamics of water in the soil depends mainly on these processes. Based on this logic, we apply soil equations describing these dynamics (Porporato et al. 2004) to understand how changes in climate conditions may affect soil moisture. Soil moisture estimations are made using a numerical model which implements these equations (Fig. 3.21).

Fig. 3.21

Water availability in soil submodel flowchart

This modeling seeks to simplify the large number of processes that make up the dynamics of water in the soil. These hydrological phenomena are accompanied by a high degree of spatiotemporal nonlinearity. The potential evapotranspiration (amount of transpiration which would occur if sufficient water were available), one of the input variables, is obtained from an algorithm described by FAO (Food and Agriculture Organization of the United Nations) (Allen et al. 1998) that uses minimum, average and maximum temperature related to the extraterrestrial solar radiation latitude.

The result generated by the model is the average predicted soil moisture taking into account climatic, pedological and vegetation conditions. The final product is a prediction of volumetric soil moisture (mm³ of water/mm³ of soil).

The soil moisture maps consistently reproduced the standards expected for tropical regions, with satisfactory agreement in the temporal variability on climate change validated from INPE (National Institute for Space Research) expert judgement (Gevaerd and Freitas 2006) and results design by MODIS project.

Vegetation Type Model

The projections were made by Maximum Entropy model approach—MAXENT (Phillips et al. 2004). The entropy concept refers to the uncertainty of a probability distribution. The MAXENT is an algorithm that seeks to calculate and minimize this uncertainty, finding the nearest future distribution of real uniform distribution, based on the constraints of environmental variables available. The algorithm seeks this new distribution by minimizing a measure of divergence between actual and simulated results, given the same set of constraints. The actual phytophysiognomy data, together with historical environmental variables, are used as model inputs to calibrate and set the constraints. Subsequently, the restrictions set by the MAXENT algorithm are used to reference the likelihood occurrence of vegetation type in relation to the constraints identified in the spatial distribution of future climate variables. Therefore, input data for the model were: occurrences of phytophysiognomies; soil moisture; topography information; historical climatic exposure and weather extremes for calibration and future data for the generation of projections of potential distributions. The data resulting from this analysis indicates the potential distribution of vegetation type according to future climate conditions, as shown in Fig. 3.22.

Fig. 3.22

Vegetation type submodel flowchart

Because of the low availability of supporting data, the results of changes in the distribution of vegetation types were validated by Report No. 6: Climate Change and possible changes in South America Biomes (Marengo 2007).

Agro-climatic Zoning Model

This impact was modeled from the variation of climatic characteristics, which mainly influence the phases of germination and grain filling, crucial to a good crop yield. The form and extent of climate change, related to the physical aspects of the environment, affect the optimal conditions of agricultural productivity of a region can be analyzed from these results. The site is considered to have low susceptibility to climate impacts if the chance of successful harvest is at least 80%.

The Water Requirement Satisfaction Index (WRSI), the ratio between the actual and the maximum level of agricultural evapotranspiration, was used as a measure of the crop water stress. The modeling was performed by using the data of the soil moisture submodel as an input to the Eagleman (1971) equation to define the real coefficient of culture. To the maximum coefficient, tabulated data are listed per culture. The air temperature rasters and extraterrestrial solar radiation are used as inputs in the evapotranspiration sub-model. These data reported above are the input components in the FAO methodology notebook 56 (Allen et al. 1998) to define the actual and maximum evapotranspiration, which is used to calculate the WRSI as shown in Fig. 3.23.

Fig. 3.23

Agro-climatic zoning submodel flowchart

The crops considered in the preparation of this study (maize, soybean, beans, cotton, sugarcane and rice) were those viewed as strategic for the economy and food security of Brazil. The results are shown sorted by a WRSI ranging from 0 to 1, with higher values suggesting lower degrees of water stress. The information generated by the model was validated using experimental results of the study by Embrapa (Assad et al. 2013), for maps of the Ministry of Agriculture, Livestock and Supply (MAPA) and results of the project (MODIS).