The cure for anything is salt water—sweat, tears, or the sea.

Guest Post by Willis Eschenbach

There’s an interesting study in Science magazine, entitled “Ocean Salinities Reveal Strong Global Water Cycle Intensification During 1950 to 2000” by Durack et al. (paywalled here, hereinafter D2012). The abstract of D2012 says:

Fundamental thermodynamics and climate models suggest that dry regions will become drier and wet regions will become wetter in response to warming. Efforts to detect this long-term response in sparse surface observations of rainfall and evaporation remain ambiguous. We show that ocean salinity patterns express an identifiable fingerprint of an intensifying water cycle.

Our 50-year observed global surface salinity changes, combined with changes from global climate models, present robust evidence of an intensified global water cycle at a rate of 8 T 5% per degree of surface warming. This rate is double the response projected by current-generation climate models and suggests that a substantial (16 to 24%) intensification of the global water cycle will occur in a future 2° to 3° warmer world.

Let’s start with salinity of the ocean, and how it varies around the globe.

Figure 1. Mean salinity of the ocean in “Practical Salinity Units” (PSU). Figure 1(D) from D2012.

One thing that we can see in Figure 1 is that where there is plenty of rain, along the equator and near the poles, the ocean is less salty (lower salinity). Conversely, where there is a lot of evaporation and little rain, the ocean is saltier.

Intrigued by their thesis that “dry regions will become drier and wet regions will become wetter in response to warming”, I pulled out my Argo surface data to take a look at the salinity records, and to get some idea of how the salinity varies with time, temperature, and location. What I found agrees with my general mantra, “Nature simply isn’t that simple”.

I started by dividing the globe up into five regions: North and South Pacific, North and South Atlantic, and Indian Ocean. I like to start my investigations by looking at a large scale, and then work downwards. Here’s the records for the Indian Ocean.

Figure 2. Temperature versus salinity for the Indian Ocean. Colors indicate the year that the observations were taken. Click on the image for a larger version.

There are several things that we can see in this plot. First, there has been no obvious change salinity during the decade. The earlier records (red) are distributed very similarly to the later records (blue).

Next, the shape of the curve is interesting in that it shows a very different salinity response at different temperatures. At the coldest end of the scale, and up to about 5°C, increasing temperature yields decreasing salinity. Next, from 5°C to about 20°C, as temperature rises, salinity increases, meaning less rain.

Finally, above about 20°C, increasing temperature correlates with decreasing salinity, meaning more rain.

Now, let’s consider their claim, that with increasing temperatures “dry regions will become drier and wet regions will become wetter”. This claim rests on the reasonable assumption that the salinity is inversely related to rainfall, because the fresh rainwater dilutes the salty ocean.

But here’s the problem with the claim. Let’s take a look at two areas, both with the same salinity, say 35 PSU.

Figure 3. As in Figure 2, with a yellow line indicating ~ 35 PSU salinity.

Note that the yellow line intersects two areas, one warmer and one cooler. Now presumably, since salinity is a proxy for rainfall, the two areas are equally wet, or are equally dry.

Now, if the temperature increases, one of the areas (the one on the left) will show an increase in salinity (decreasing rain), while the other one will show a decrease in salinity (increasing rain).

But we can replicate this result at each level of salinity. At each level of salinity (and therefore rainfall), when it warms, some areas get wetter and some areas get dryer. Therefore, it is not true to say that as temperature increases “dry regions will become drier and wet regions will become wetter”. In fact, some dry regions will get wetter, and some will get dryer, and the same is true for wet regions.

This is just the Indian Ocean, however. Let’s see what the other areas show. Here’s the same graph, for the South Atlantic.

Figure 4. Temperature-salinity plot for the South Atlantic, with the years indicated by colors.

As I said above, nature simply isn’t that simple, and the South Atlantic is different from the Indian Ocean. In the South Atlantic, as the temperature increases, rainfall decreases. Instead of the wet areas getting wetter and vice-versa, all areas get drier with increasing temperature.

Next, I looked at the South Pacific:

Figure 5. As in Figure 4, for the South Pacific

In the South Pacific, we see yet another pattern. There’s not a whole lot of change in salinity as the temperature varies. How about the North Pacific?

Figure 6. As in Figure 5, for the North Pacific. 

Again, the change in salinity is much smaller that in e.g. the Indian Ocean. However, the same thing is true—for every place that is dry that will get drier if it warms, there is another place that is dry that will get wetter if it warms.

Finally, for complexity, nothing matches the North Atlantic.

Figure 7. As in Figure 6, for the North Atlantic.

Once again, we see that there are dry areas that would get wetter, and wet areas that would get drier, with a temperature increase. However, there are a lot of areas in the North Atlantic that seem not to be following any general trend … complex nature strikes again.

Having looked at how the temperature is related to the salinity for large areas, I decided to look at how the salinity and temperature changed with time for smaller areas. I started with the Pacific, and I picked an area where I could look at ten-degree latitudinal bands. Figure 8 shows those bands.

Figure 8. Salinity map as in Figure 1. Yellow boxes show the delineation of the areas analyzed.

First I looked at the Northern Hemisphere in the Pacific.

Figure 9. North Pacific salinities by ten-degree bands. Colors indicate from coldest to warmest for each individual band. Purple dotted line shows the average for the entire North Pacific region. Black line shows a 200-point gaussian average of the salinity. “Sal. Chg.” is the salinity change (expressed as a change per 50 years for comparison with the D2012 study, along with the “p-value” for the trend rounded to three digits. “Temp. Chg.” is the temperature change (expressed as a change per 50 years for comparison with the D2012 study, along with the “p-value” for the trend rounded to three digits. “Sal. Anom.” is the salinity anomaly expressed in relation to the area salinity average (purple dashed line). “PA” is the “pattern amplification” discussed in the paper, which is the salinity change divided by the salinity anomaly. Note that some of the changes are not statistically significant (p greater than 0.05).

The claim made in the paper is that when the salinity is high (positive salinity anomaly), the salinity change should be positive with increasing temperature (dry gets drier), as well as the reverse—when salinitiy is low, the salinity change should be negative with increasing temperature (wet gets wetter).

However, in four of the six areas shown above, this is not the case. Nature is simply not that simple.

Next, Figure 10 shows the corresponding chart for the South Pacific:

Figure 10. As in Figure 9, for the South Pacific.

Once again, nature is not cooperating. First, in many areas there is little change in salinity. From 50°S to 40°S, the annual swing in temperatures is 14°C, but there is almost no annual change in salinity. In addition, the overall change is in the wrong direction. For another example, look at the band from 20°S to 10°S. Salinity is above the area average, but despite that, salinity is higher during the cool part of the year. In addition, temperature dropped, but contrary to predictions, the previously high salinity increased …

I append the corresponding charts for the Atlantic Ocean, which show much the same thing as we see in the Pacific—a confusing mix of responses.

My conclusions? Well, my main conclusion is that there is no general “wet get wetter and dry get drier” changes. For every dry area that is getting wetter with increasing temperature, there is another area which is just as dry that is getting drier with increasing temperature.

My second conclusion is that different parts of the ocean react very differently to increasing temperature. In some areas, neither annual nor decadal changes in temperature make much difference to the salinity. In others they are positively correlated, and in yet others, they are negatively correlated.

Does this make the D2012 paper incorrect? I don’t know, because they didn’t archive the data that they used for the 50 year period 1950-2000. It does, however, indicate that as is usual with the climate, generalizations are hard to draw. Humans always want things reduced to simple relationships like “if temperature goes up, wet areas get wetter and dry areas get drier.” Like Aesop, we prefer simple morals for our fables. Unfortunately, nature is nowhere near that simple.

My best to all,

w.

PS—I have not commented on the use of a combination of traditional salinity measurements and Argo float measurements, and I do not find any comments by the authors of D2012 regarding the topic. However, it would seem that it should be discussed and the two measurements compared where they overlap in time and space.

APPENDIX 1: Salinity charts by latitude band for the Atlantic Ocean. See the captions for Figures 8 and 9 for details.

Appendix 2. Data and code (in the computer language “R”) are here as a zipped archive (WARNING: 32 Mb archive). Data is in an R “save” file called “argo temps.tab”. WARNING 2: The code is not “user-friendly” in any sense, and might best be termed “user-aggressive”. It is NOT designed to be run as a single piece.