That bastion of careful, insightful science journalism that is The Internet is abuzz with stories of a "second Amazon river" discovered some 4,000 m (!!!) below the Original Recipe Amazon at the surface. The work was done under the auspices of the National Observatory of Brazil, which actually provides a pdf version (in English!) of Pimental and Hamza's talk here. You can find all sorts of third-party stories on a number of "news" websites, apparently because "underground river" is just the kind of evocative bullshit that gets noticed by what passes for science journalists these days. I won't link to them, because they're uniformly stupid.
To summarize their work, Pimental and Hamza report on geothermal data from 241 wells (apparently drilled by Petrobras); 185 of the wells are located in "sedimentary basins", while the remaining wells are located "outside" of those regions, presumably piercing non-sedimentary materials, although that's an assumption I'm making, not one reported in the pdf version of the talk linked above.
The wells monitored what was assumed to be advective heat transfer, which is to say the transfer of heat through the migration of fluids; in this case, deep groundwater moving through permeable rock in the subsurface. Keep in mind, I'm no hydrologist, so I don't know much about the technique of geothermal monitoring and abstracting flow characteristics from that sort of data. There must be, of course, a whole slew of geological factors that influence flow in the subsurface, ranging from sedimentological and stratigraphic attributes all the way to structural geological considerations; however, those data may not be available for the Amazon Basin, or at least weren't reported in Pimental and Hamza (2011).
Importantly, Pimental and Hamza (2011) construct a bunch of 1-D models of heat flow for their individual well sites, and make the assumption that the observed geothermal curves are a result of advective heat transfer. These 1-D thermal profiles are then extrapolated across the basin. Then they divide up there wells into "upper" and "lower" populations (upper and lower in terms of depth), and interpret the different velocities for those different depths as indicating different flow patterns. As reported by Pimental and Hamza (2011), they interpret that there is a lateral west-to-east flow transfer present in the deeper aquifers.
So far, that all seems all right to me; sure, there are problems in extrapolating 1-D data to 3-D basins goes, and maybe there are problems with the assumptions that went into the methods, but that's all fine and good and offers avenues for further investigation. And, of course, the one bright spot in all the reporting has been the idea that these results are preliminary, and that future work will be done to interrogate them further.
The problem, as far as I can see it, comes in when Pimental and Hamza start to compare the surface flows of the Amazon River with their interpretations of subsurface flow, leading the authors to make some kinda big-deal claims about the importance of this groundwater system. They have a graph, on page 23 of the pdf version of their talk, that provides this comparison. I'll wait while you take a look.
I have some problems with their numbers. First of all, I don't exactly understand what the hell they mean when they say "surface drainage". It seems like they are talking about the channelized system alone, since they label one column "Amazon River" but the width value they give is kind of off...the Amazon, at low flows, ranges from 1 to 10 km wide, but when experiencing above bankfull conditions it'll spread out to 50 km or so. Are Pimental and Hamza just sort of averaging the width to somewhere between 1-100 km? Or are they also talking about overland flow in the floodplain drainage? Who knows?
By far, though, the goofiest number is the discharge value given for the Amazon. Look, the paraphrase Douglas Adams, the Amazon River is big. Really big. You just won't believe how vastly, hugely, mindbogglingly big it is. It supplies like 20% of the freshwater to the ocean, for god's sake! But Pimental and Hamza provide a MUCH too small estimate of the Amazon's annual discharge. They state that the flow rate is 133.000 m^3/s, which is either 133 or 133,000 with that weird euro-convention of a period instead of a comma. I'll give them the benefit of the doubt and assume that they don't mean 133, which is silly, but even 133,000 is only half of the actual number. Estimates for discharge range a little closer to 210,000 m^3/s, with an upper value of +300,000 m^3/s during the rainy season. This, in comparison to their groundwater flow estimate of 3900 m^3/s. Orders of magnitude, folks. AND that's assuming that the discharge value for the groundwater stands up to further interrogation.
If you look at the number, however, the overall contribution of the groundwater value reported is ~1-2% of the discharge of the huge ass Amazon River. Yet, for some reason, the National Observatory of Brazil put out a statement that says that this is the "likely" that the low salinity found at the mouth of the Amazon is caused by this groundwater contribution. That is a stupid thing to say, National Observatory of Brazil. The reason the ocean is fresher at the mouth of the Amazon is because the Amazon is a fucking HOSE, dumping an unbelievable amount of freshwater into the sea at an ungodly rate!
Most annoyingly, of course, is the fact that for some inexplicable reason, Pimental and Hamza (2011) refer to this subsurface flow as a "River", even going so far to name it the "Rio Hamza" in honor of one of the authors. And, of course, it's the "underground river" angle that got the newsoids all excited. I honestly can't tell why Pimental and Hamza decided to do that; is it just poetical, or what? Look, people are already stupid enough, and think that the water they get out of a well taps into underground rivers in caves, or huge open water lakes deep in the earth, rather than recognizing the truth about interstitial water between grains. For the love of god, don't confuse them any further by making them think that, 4000 meters below the Amazon River there's a second river, flowing along and full of subterranean Caimans, BECAUSE THAT'S WHAT PEOPLE THINK WHEN THEY HEAR UNDERGROUND RIVER!
To sum up my view on the whole thing: Pimental and Hamza (2011) identify some tantalizing evidence that there might be some lateral mobility in the groundwater deep in the Amazon Basin, and that this water might discharge into the open ocean at the mouth of the river. This was based on a bunch of 1-D well log models of flow velocities extrapolated over the whole area of the basin, and not on any sort of tracer study or more complicated modelling. The amount of water discharged is inferred to be ~1% the value of the Amazon River's discharge, which is nothing to sneeze at, but this is going to need a LOT more data to validate such a conclusion. For some reason, the authors decide to name tiny little blebs of water squishing along through interstitial pores a "River", which is wrong and terribly confusing to people who are already shockingly ignorant about how the earth works.
The whole thing makes me mad, but now when you search "Amazon River" on the Googles, you're going to be faced with a bullshit story about a "second River", without any attempt by any of the "journalists" to point out that, well, you know, water might be moving underground and all, but it sure as shit ain't a river, and it certainly ain't on the goddamn scale of the, you know, ACTUAL RIVER sittin' up there on the surface. My prediction: you will never hear about this work again, because future work is going to DRASTICALLY reduce the actual discharge estimates from the groundwater.