Methodology for the Amazon Water Impact Index

By Cecília Gontijo Leal

Reach of impact analysis

In the Aquazonia project, we considered the part of the Amazon River Basin that lies inside Brazil—an area composing 55% of the total river basin, or five million square meters. The spatial framework adopted for the river network and definition of river basins was that of Venticinque et al. (2016).

Variable threats to the aquatic ecosystems

In order to create the Amazon Water Impact Index (AWII) we collected spatial data representing the main threats to aquatic ecosystems in the Amazon: changes in land use and cover, infrastructure, forest degradation and climate change. Other relevant threats like fishing and pesticide contamination were not included due to the lack of available data.

Data of occurrence and distribution of the threats were obtained from free public databases such as Projeto Mapbiomas (2021), The Amazonian Network of Georeferenced Social-Environmental Information (RAISG, 2021), Brazil’s National Water and Sanitation Agency (ANA, 2021), Climate Hazards Group Infrared Precipitation with Stations (CHIRPS, 2021) and from scientific publications (Funk et al.; 2015; Matricardi et al., 2020; Souza et al.; 2020). When data from more than one database were used for the same threat, we excluded the shared data to prevent duplicity.

The data used have spatial resolution of 30 m, except for rainfall, which is available in 0.05 o (approximately 5 km). We selected nine layers of data on threats to compose the index. They are: Agriculture and cattle raising (AGP), Urban areas (URB), Illegal prospecting (GAR), Industrialized mining (MIN), Hydroelectric dams (HDL), Road crossings (EST), Waterways (HID), Forest degradation (DEG), and Change in precipitation (PRE) (Table 1).

Table 1. Variables of threats included in the Amazon Water Impact Index

Category Threat Abbrev. Unit of measure Data source Resolution Year Weight
Land use Agriculture and cattle raising AGP Percentage of area (%) Mapbiomas 30 m 2020 2
Land use Urban areas URB Percentage of area (%) Mapbiomas 30 m 2020 3
Land use Illegal mining GAR Percentage of area (%) Mapbiomas, RAISG 30 m 2020 3
Land use Industrial mining MIN Percentage of area (%) Mapbiomas, RAISG 30 m 2020 2
Infrastructure Hydroelectric plants HDL Presence/absence ANA, RAISG 30 m 2021, 2020 3 e 1
Infrastructure Road Crossings EST Density (num/km2) Mapbiomas, RAISG 30 m 2020 1
Infrastructure Waterways HID Density (km/km2) Mapbiomas 30 m 2020 1
Forest degradation Forest degradation DEG Percentage of area (%) Matricardi et al. 30 m 1992-2014 1
Climate change Change in precipitation PRE mm CHIRPS 0,05° 1981-2020 2

For the variable Hydroelectric Plants (HDL), we considered hydroelectric dams, small hydropower plants (PCH) and thermoelectric plants in operation or under construction. We considered operational waterways only. To represent forest degradation (DEG) we used the data from Matricardi et al. (2020) referring to areas that had been only burned, areas that had been only clearcut, or areas that had been both burned and clearcut.

As an indicator for climate change, we used the difference in precipitation based on the precipDiff index proposed by the partners of the Synergize/CNPq project:

precipDiff = precip(2001-2020) - precip(1981-2000)

where precip(2001-2020) is the average annual precipitation for the period from 2001-2020, and precip(1981-2000) is the average annual precipitation between 1981-2000. Positive precipDiff values indicate increased average precipitation over the period and negative values indicate decrease. The average annual rainfall was calculated based on daily data available on the CHIRPS platform (Funk et al. 2015) from Google Earth Engine (GEE).

The variables of land use and cover, infrastructure and forest degradation were measured in terms of the percentage of area or density in each basin. In the case of hydroelectric plants, we chose to count not only the number of dams or area occupied by reservoirs in each microbasin, as this would highly underestimate the reach of their impact. Instead, we counted the hydroelectric plants on two spatial scales: with a weight of 3 in microbasin BL7 where its reservoir is located, and a weight of 2 in the BL5 basin.

To incorporate variables with different units of measure (i.e. % or mm) in the index, we standardized each variable to a scale of 0 to 1.

Calculation of the AWII

The definition of the spatial scale for calculating the AWII was based on the classification of the Amazon river basins proposed by Venticinque et al. (2016), Amazon GIS-Based River Basin Framework. In this database, the Amazon Basin is subdivided into 7 levels, ranging from level 1 (BL1) representing the entire basin, level 2 (BL2) for the sub-basins of the main tributaries (e.g. Madeira and Tapajós), until level 7 (BL7) with smaller or microbasins (i.e 15,269 basins from 300-1,000 km2). For the AWII, we used the BL7 micro-basins to guarantee more precision. So for the basin of the Amazon River in Brazil, we calculated the AWII for 11,216 micro-basins with an average area of 479.6 km2.

Next, we calculated the values of each threat variable for each micro-basin. In order to avoid including redundant variables which would inflate their influence on the index, we verified autocorrelation between variables according to the Pearson correlation test.

To represent the severity of the different threats, we attributed weight to the different variables: activities of greater impact were given a weight of 3, medium-impact variables were given 2, and those with the least impact, 1 (See Table 1).

The AWII value for each micro-basin was the sum of the values of each threat variable multiplied by its weight: IIAA= 3*URB + 3*GAR + 3*HDL + 2*AGP + 2*MIN + 2*PRE + EST + HID + DEG


Databases Brazilian National Water and Sanitation Agency(ANA), accessed on August 26, 2021 via the link: The Amazonian Network of Georeferenced Social-Environmental Information (RAISG), accessed on August 26, 2021 via the link: MapBiomas – Collection v.6.0 of Annual Series of Maps of Land Use and Cover in Brazil English accessed on August 26, 2021 via the link: CHIRPS: Rainfall Estimates from Rain Gauge and Satellite Observations, accessed on August 26, 2021 via the link: Amazon Forest Degradation Analytical Spatial Data, accessed on August 26, 2021 via the link:

Scientific Articles Venticinque et al. (2016). An explicit GIS-based river basin framework for aquatic ecosystem conservation in the Amazon. Earth System Science Data, 8: 651–661.

Matricardi et al. (2020). Long-term forest degradation surpasses deforestation in the Brazilian Amazon. Science, 369: 1378-1382.

Funk et al. (2015). The climate hazards infrared precipitation with stations - a new environmental record for monitoring extremes. Scientific Data, 2: 150066.

Souza et al. (2020). Reconstructing three decades of land use and land cover changes in Brazilian biomes with Landsat archive and Earth Engine. Remote Sensing, 12(17): 2735.

Return to homepage