Aquífero Guarani: um imenso oceano sob nossos pés – Ecoa
Working with the Guarani Aquifer Map in Practice
The Guarani Aquifer (Aquífero Guarani) sits beneath roughly 1.2 million square kilometers across southeastern Brazil, southern Paraguay, eastern Uruguay, and northeastern Argentina. It is a sandstone aquifer within the Paraná Basin, and it matters because somewhere around 80 million people draw water from it depending on how you count. The Brazilian portion alone is roughly 840 thousand square kilometers.
I spent a while trying to line up hydrogeological cross-sections with well log data from municipal databases. The maps you find online do not always match the coordinate references those municipal datasets use. One project I was on had a vector shapefile in WGS84 (EPSG:4326) sitting next to a borehole database that was originally projected in SAD69 / UTM zone 22S. If you just drag both layers into QGIS without checking, the wells appear hundreds of meters offshore. Reclassify the database CRS explicitly, reproject on export, and only then intersect them. That step alone saved me about three hours of trying to figure out why the data looked wrong.
Where to find the mapa aquifero guarani
The official source is the Grupo de Trabalho da Bacia do Aquífero Guarani (GT HAG), which operates under COPEBRAS and the secretariat of the Intergovernmental Commission on the Aquifer Guarani. Their website hosts the downloadable geodatabase, shapefiles, and GIS-ready layers. You can also find compiled maps through the Brazilian Geological Survey (CPRM) and the national water agency (ANA). In Uruguay the relevant institution is OIRSA, and in Argentina the network is managed by provincial water authorities plus CONAE for remote sensing products.
When you go to download, look for the shapefile package that includes the stratigraphic units, potentiometric surface layers, and the thickness maps. Some portals offer a single zipped geodatabase (.gdb) instead of separate .shp files. If you are using QGIS, just load the geodatabase directly — it preserves field types better than importing individual shapefiles one by one. If you are using ArcGIS Pro, import the feature dataset and let it resolve domains rather than converting to shapefiles immediately. The domain values for lithology codes, for example, tend to drop their descriptions during an export-to-shapefile operation.
The download itself is usually under 50 MB, even when you grab the full set of thematic layers. It extracts into a folder with a clear naming convention: one layer for the aquifer extent, one for piezometric levels, one for transmissivity ranges, and a few metadata files. Read the metadata file before you start working. The coordinate system and datum shift information live there, and they matter for any analysis that involves overlaying newer satellite-derived data or regional DEMs.
I keep a shortcut on my desktop pointing to the latest official release. Every two or three years the GT HAG publishes a revised edition, and the revisions tend to fix projection issues and fill in missing thickness values for certain municipalities. The current set still shows some interpolated cells in the central Uruguay region where field data are sparse. If you are doing a thesis or a municipal study, use the latest version and cite the release year in your methodology. The older shapefiles have slightly different boundary polygons along the border with Argentina, and reviewers will ask about it.
Reading the Layers Without Getting Confused
The standard map has several overlapping layers, and each one uses a different classification system. The lithology layer uses a simplified three-category scheme for general distribution: confined, semi-confined, and unconfined. The potentiometric surface layer, though, breaks things down into drawdown cones mapped from monitoring wells. Those cones are often misread as permanent depressions. In many cases they are seasonal artifacts from pumping during the dry months. The GT HAG notes this in their documentation, but the symbol colors on the map still make them look like structural features.
Transmissivity values on the official map are presented in bands. The bands are derived from slug tests and pump tests at select wells, not from continuous measurement. A common mistake is to treat the band colors as if they were exact measurements. They are not. The published ranges span roughly 10 to over 1,000 square meters per day depending on the sector, but the uncertainty in any single cell is significant. If you need point estimates for a model, go to the well test database instead and pull the raw values.
The thickness map is probably the most useful layer for field planning. It shows the saturated thickness of the sandstone unit, not the total sedimentary column. I once tried to use the total thickness layer for a vertical flow estimate and got numbers that made no physical sense. Switching to the saturated thickness layer aligned the results with measured drawdown data within about five percent.
Another thing that trips people up: the aquifer boundary polygon is not a hard edge. It is an interpretation based on seismic lines and well control. In the southern sector near the Uruguay border, the boundary has shifted slightly between editions. The polygon itself is reliable enough for regional water budget work, but if you are zoning a specific well field, you need the underlying well logs and the local hydrogeological reports, not just the boundary shapefile.
Common Mistakes and What I Do Instead
Overlaying the Guarani map with land use data from IBGE without reprojecting both to the same CRS first is the most frequent error. I catch it by loading the data, checking the canvas CRS, and comparing the extent coordinates. If the aquifer polygon and the municipal boundaries show different coordinate values for the same geographic feature, something is off. A quick reprojection to SIRGAS2000 / UTM zone 22S usually resolves it for the Brazilian sector.
Another issue is using the piezometric contour intervals directly for quantitative head estimates. The contours are drawn at ten-meter intervals in most published versions, and the interpolation method behind them is not always disclosed in the layer metadata. For rough directional flow analysis, ten-meter contours are fine. For anything that requires a gradient calculation, sample the raw potentiometric surface point data and run your own interpolation with a method that suits your data density, such as kriging with a spherical variogram. That approach usually takes about twenty minutes on a modern laptop for the full dataset.
When I need to explain the aquifer to municipal planners or developers who do not work with GIS, I export a clean PDF map with the four essential layers visible: extent, potentiometric heads, saturated thickness, and major cities. I leave out the transmissivity bands unless someone specifically asks. The extra layer clutters the visual and introduces the band-interpretation problem I mentioned. The four-layer version is clear and takes about five minutes to generate once the base map is set up.
Offline Use and Quick Reference
If you are working in the field or in an area with poor internet, downloading the full shapefile package to a local drive is faster than querying web services. The official maps are also available through CPRM’s SigGeo platform, but the direct download avoids rate limits and session timeouts. I keep a copy on an external SSD and refresh it whenever a new GT HAG edition appears.
For quick reference without opening a full GIS project, the PDF atlas that accompanies the geodatabase is adequate. It has the main maps at 1:2,500,000 scale with annotations in Portuguese. The English version exists but is less complete on the hydrogeological cross-sections. If you need the cross-sections for a report, grab the Portuguese PDF and the underlying section shapefiles from the download package. The cross-sections themselves are stored as lines and polygons with attribute tables linking them to specific well pairs.
One practical tip: label the potentiometric flow direction arrows manually in your final map rather than relying on the automatic label placement in QGIS. The auto-placed labels overlap the arrows and make the flow paths harder to read within about ten seconds of looking at the map. Manual placement adds maybe five minutes but improves readability significantly for non-technical audiences.
What the Map Does Not Tell You
The Guarani aquifer map is a static snapshot. It does not model recharge rates dynamically, and it does not incorporate climate projection scenarios into the main layers. Recharge varies by sector and is generally higher in the outcrop areas along the eastern margin of the basin. The recharge estimates in the literature range from about 200 to 800 millimeters per year depending on land cover and rainfall, but those are spot values, not map-ready raster layers in the official dataset.
Contamination risk is another area where the map is silent. The aquifer itself is well-protected in most of its confined extent, but the unconfined and semi-confined zones near the outcrop areas are more vulnerable. I have seen municipal sewage infiltration problems in the Florianópolis region where the aquifer is shallow and unconfined. The official map will show you where those zones exist, but it will not show you current water quality. For that you need the ANA monitoring network data or the state-level environmental agency reports.
Groundwater modeling is possible with the map data as a starting framework, but you will need to bring in additional parameters: hydraulic conductivity from literature values or test data, specific yield estimates, boundary conditions defined by the regional geology, and recharge time series. A basic steady-state model built in MODFLOW using the saturated thickness and transmissivity bands as starting inputs can be set up in a couple of days by someone who already knows the software. Running a proper calibrated transient model takes months and requires a sustained monitoring network.
The download links and the most up-to-date shapefiles remain on the GT HAG portal and on CPRM’s Geobanco platform. Save the links you use. The site URL changes occasionally when they update the content management system, and the old links can break without much notice.
