More Flooding in Paraguay

Lower Paraguay River Basin

It's helpful to refresh ourselves on the geography of this region. Figure 1(b) from our paper shows the Lower Paraguay River Basin; the capital city Asunción is shown with the letters ASU.

One relevant feature of this region is that it's extremely flat. This is shown by the contours (strictly speaking they show \(\log_{10}\) of elevation, in meters). As a consequence, rainfall that falls here tends to take a long time to drain once the soils are saturated. This means that above-average rainfall over a long time can cause flooding, even if none of the individual storms are particularly heavy.

Rainfall Patterns

Now we're ready to explore what happened during this event.

Let's start by plotting the rainfall observed over the past several months. This figure shows the rainfall, averaged in space over the region defined in the red box in the map above. We can see that between 14 September and 30 October of 2018, the region observed persistent and repeated heavy rainfall events. For context, the average rainfall in this region is just under 4 mm per day.

Now let's see what it looks like if we look on a map. Since the time from 14 September to now seems to be the most important (at time of writing they have data through 29 October), let's composite (average) over that time period. This data shows rainfall anomalies; this is a word I will be using a lot here and just means the difference between what was observed and the average expected given the seasonal cycle [1]. We can see that the region feeding the _Lower_ Paraguay River Basin got far hit especially hard [2].

Direct Drivers

It's worth thinking a bit about this rainfall pattern. In our paper (DossGollin et al., 2018), we found (consistent with many other studies) that intense rainfall in Paraguay is typically driven by the "South American Low-Level Jet", which channels moisture and energy (both are needed for rain storms) from the Amazon to South East South America. This low-level jet can either break past 25 degrees S, in which case it will favor rainfall in Northern Argentina and Uruguay ("Chaco jet event") (Salio et al., 2002), or it can turn to the East, in which case it will favor rainfall over Paraguay and SW Brazil ("No-Chaco Jet Events") (Vera et al., 2006). If you look at our figure 6, available on my GitHub page, you can see that the rainfall observed over the past six weeks looks a lot like the weather type (we labeled it number 4) that we identified as a key driver of the 2015-16 flooding.

To get a better sense of how the low-level jet behaved during this period, we can look at climate anomalies that persisted during this time period. The most interpretable variable to look at is the wind field. This plot shows the wind at 850, which is the lower part of the atmosphere; this part of the atmosphere carries most of the moisture and energy in the low-level jet, so this data tells us a lot about large-scale moisture transport by the atmosphere.

The most obvious feature here is that the region around (60W, 17.5S) shows strong wind anomalies -- this means that the low-level jet was much stronger during this period. This makes sense what we saw in the rainfall time series plot -- it's reasonable to assume that during most or all of the peaks in the rainfall time series, the low-level jet was active on that day and/or the previous day.

It's also interesting to note a few other features in this plot. First, if we look a little bit to the South-East of Paraguay -- say around (52.5W, 27.5S) -- we see that the wind anomaly, though week, is pointing towards Paraguay. This means that (on average) the circulation didn't allow the low-level jet to push through to Uruguay (this would be the "Chaco" jet event defined above). Further, air moving in this direction towards the Lower Paraguay River Basin supports convergence. Simply put, when two parcels of air near the bottom of the atmosphere run into each other, they will tend to go up (since the ground is below them they can't go down) and upward motion favors rainfall. There also seem to be some interesting things going on in the mid-latitudes; there appears to be a strong persistent low centered around (82.5W, 42.5S) that could be relevant here.

Indirect Drivers

That wasn't an exhaustive analysis (this is a blog!) but at least we saw that the rainfall we found was consistent with observations [3] of the low-level jet. In our Journal of Climate paper we found some interesting links between a few large-scale climate indices and rainfall in this region. These links encompass a lot of variability; in other words, there is a lot of noise between the large-scale factors we found and rainfall in the Lower Paraguay River Basin, which makes sense since it's a very small region.

The first place to look for signal on these sub-seasonal to seasonal time scales is the ocean. Since the specific heat (amount of energy required to raise the temperature) of water is much larger than the specific heat of air, heating anomalies in the ocean can drive persistent atmospheric circulations on these relatively short time scales [4].

There's lots that one could unpack here, but I'm going to focus on one particular interesting feature. In our paper we hypothesized that a dipole pattern in the South Central Atlantic -- we defined it as going from 30W to 10W and 15S to 40S -- could favor "No-Chaco" jet events over "Chaco" jet events and thus increase the probability of heavy rainfall in the Lower Paraguay River Basin.

Although we hypothesized specifically that this could happen during El Niño years during the summer (December-February), a dipole pattern like the one we identified was active during the current floods. The dipole during the current floods appears to be shifted somewhat to the South and East of our sketch (no more than five degrees or so). Nonetheless, it may have contributed to the rainfall that we observed.

Wrapping Up

Thanks for reading this far! A few more points in case you're interested