Wednesday, March 7, 2018

Will Advances in Groundwater Science Force a Paradigm Shift in Sea Level Rise Attribution

In a 2002 paper, what is frequently referred to as “Munk’s enigma”, Scripps Institution of Oceanography’s senior researcher bemoaned the fact researchers could not fully account for the causes of sea level rise. He lamented, “the historic rise started too early, has too linear a trend, and is too large.” Early IPCC analyses noted about 25% of estimated sea level rise was unaccounted for. Accordingly, in 2012, an international team of prominent sea level researchers published, Twentieth-Century Global-Mean Sea Level Rise: Is the Whole Greater than the Sum of the Parts? (henceforth Gregory 2012). They hoped to balance struggling sea level budgets by re-analyzing and adjusting estimates of the contributions from melting glaciers and ice caps, thermal expansion, and the effects of dam building and groundwater extraction. However, a natural contribution from any imbalance in groundwater re-charge vs discharge was never considered. Yet the volume of freshwater stored as groundwater, is second only to Antarctica’s frozen supply, and 3 to 8 times greater than Greenland’s.

At the risk of oversimplifying, the effects of groundwater storage can be differentiated between shallow-aquifer effects that modulate global sea level on year to year and decade to decade timeframes, versus deep aquifer effects that modulate sea level trends over centuries and millennia.

Researchers are increasingly aware of natural shallow groundwater dynamics. As noted by Reager (2016) in A Decade of Sea Level Rise Slowed by Climate-Driven Hydrology, researchers had determined the seasonal delay in the return of precipitation to the oceans causes sea levels to oscillate by 17 ± 4 mm [~0.7 inches] per year. Reager (2016) also argued decadal increases in terrestrial water storage driven by climate events such as La Nina, had reduced sea level rise by 0.71 mm/year. Likewise, Cazenave 2014 had published according to altimetry data, sea level had decelerated from 3.5 mm/yr in the 1990s to 2.5mm/yr during 2003-2011, and that deceleration could be explained by increased terrestrial water storage, and the pause in ocean warming reported by Argo data.

Improved observational data suggest during more frequent La Nina years a greater proportion of precipitation falls on the land globally and when routed through more slowly discharging aquifers, sea level rise decelerates. During periods of more frequent El Niños, more rain falls back onto the oceans, and sea level rise accelerates. In contrast to La Nina induced shallow-aquifer effects, deep aquifers have been filled with meltwater from the last Ice Age, and that water is slowly and steadily seeping back into the oceans today.

Munk’s “Too Linear Trend” Enigma and Deep Groundwater Discharge

Hydrologists concerned with sustainable groundwater supplies and drinking water contamination, have been in the forefront of analyzing the volume and ages of the world’s groundwater, providing greater insight into deep aquifer effects.  Gleeson (2015) determined, “total groundwater volume in the upper 2 km of continental crust is approximately 22.6 million cubic kilometers, twice as much as earlier estimates. If all 22.6 million cubic kilometers of freshwater stored underground reached the oceans, sea level would rise 204 feet (62,430 millimeters). Via various isotope analyses and flow models, Jasechko (2017) estimated that between 42-85% of all groundwater stored in the upper 1 kilometer of the earth’s crust is water that had infiltrated the ground more than 11,000 years ago, during last Ice Age.

Clearly the earth’s groundwater has yet to reach an equilibrium with modern sea levels. With deep aquifer discharge primarily regulated by geological pore spaces (in addition to pressure heads), the slow and steady discharge of these older waters affects sea level rise on century and millennial timeframes. And, although freshwater discharge from deep aquifers may be locally insignificant relative to river runoff, deep aquifer discharge when integrated across the globe could account for the missing contribution to the sea level rise budgets.

Unfortunately quantifying the groundwater discharge contribution to sea level rise is extremely difficult, suffering from a low signal to noise problem. That difficulty is why natural groundwater contributions are often ignored or brushed aside as insignificant. Although GRACE satellite monitoring of gravity changes offers great promise for detecting changes in terrestrial groundwater storage, GRACE cannot accurately separate the relatively small discharge of deep aquifers from large annual changes in shallow groundwater. In periods of heavy rains, groundwater increases will mask deep aquifer discharge. And during a drought, any deep groundwater discharge will likely be attributed to the lack of rain.

However, estimates of groundwater re-charge via isotope analyses can provide critical information regards rates of groundwater re-charge and discharge.

Using the abnormal levels of tritium released during nuclear testing in the 1950s, plus carbon­14 dating, researchers have categorized the time since groundwater had last left the surface into 25, 50, 75 and 100-year old age classes. As expected, the youngest water is concentrated in the shallowest aquifer layers and the proportion of young water decreases with depth. The estimated volume of 25-year-old or younger groundwater suggests global groundwater is currently recharging at a rate that would reduce sea level by 21 mm/year (0.8 inches/year). Water cycle researchers  (i.e. Dai and Trenberth) have made the dubious assumption that the amount of water transported via precipitation to the land from the ocean is balanced each year by river runoff.  But if the tritium derived estimates are valid, balancing water cycle and sea level budgets becomes more enigmatic. Clearly a significant amount of precipitation does not return for decades and centuries.

Intriguingly, comparing the smaller volume of ground water aged 50 to 100-years-old versus the volume of water 50-years-old and younger suggests 2 possible scenarios. Either ground water recharge has increased in recent decades, or if recharge rates averaged over 50 years have remained steady, then as groundwater ages a significant portion seeps back to the ocean at rates approaching 1.7 mm/year, a rate that is very similar to 20th century IPCC estimates of sea level rise.

Groundwater discharge must balance recharge or else it directly alters global sea levels. When less than 21 mm/year seeps back to the ocean, then natural groundwater storage lowers sea level. When discharge is greater than 21 mm/year, then groundwater discharge is raising sea level. Without accounting for recharge vs discharge, the much smaller estimates of all the other factors contributing to sea level rise are simply not well constrained.

Higher rates of discharge could account for the enigmatic missing sea level contributions reported by the IPCC and other researchers (i.e. Gregory 2012). More problematic, if discharge proves to significantly exceed recharge, then estimates of contributions from other sources such as melting ice and thermal expansion may be too high. What is certain, the current estimates of contributions to sea level from melting ice and thermal expansion only range from 1.5 to 2.0 mm/year, and those factors by themselves cannot offset the tritium estimated 21 mm/year of groundwater recharge. So, what is missing in our current water cycle budgets?

The Importance of Submarine Groundwater Discharge (SGD)

The recharge-discharge imbalance can be reconciled if water cycle budgets included the difficult-to-measure rates of prolific submarine groundwater discharge (SGD). Freshwater springs bubbling up from coastal sea floors have long been observed. To reliably replenish drinking water, Roman fisherman mapped their occurrences throughout the Mediterranean. Moosdorf (2017) has reviewed the locations and many human uses of fresh submarine groundwater discharge around the world.

Recent ecological studies have measured local submarine groundwater seepages to determine contributions of solutes and nutrients to coastal ecosystems. But those sparse SGD measurements cannot yet be reliably integrated into a global estimate. Rodell (2015) notes that most water cycle budgets have ignored SGD due to its uncertainty, so Rodell’s water cycle budget included a rate of SGD equivalent to 6.5 millimeters/year (~0.25 inch/yr) of sea level rise. However, that estimate is still insufficient to balance current recharge estimates.

However, with improving techniques, researchers recently estimated total submarine groundwater (saline and fresh water combined) discharges suggesting a rate 3 to 4 times greater than the observed global river runoff, or a volume equivalent to 331 mm/year (13 inches) of sea level rise. Nonetheless more than 90% of that submarine discharge is saline sea water, most of which is likely recirculated sea water, and not likely to affect sea level. Only the fraction of entrained freshwater would raise sea level. To balance the 21 mm/year ground water recharge, between 6 and 7% of total SGD must be freshwater and that amount is very likely. Local estimates of the freshwater fraction of submarine discharge range from 1 to 35%, and on average just less than 10%. If fresh submarine groundwater discharge approaches just 7% of the total SGD, it would not only balance current groundwater recharge, but would steadily raise sea level by an additional 2 mm/year, even if there was no ocean warming and no melting glaciers.

A Sea Level Rise “Base-flow” and Paleo-climate Conundrums

Hydrologists seek to quantify the aquifer contributions to river flow, otherwise known as the “base flow”. During the rainy season or the season of melting snow, any groundwater contribution is masked by heavy surface runoff and shallow aquifer effects. However, during extended periods of drought hydrologists assume the low river flow that persists must be largely attributed to supplies from deeper aquifers. Streams that dry up during a drought are usually supported by small shallow aquifers, while reduced but persistent river and stream flows must be maintained by large aquifers. Using a similar conceptual approach, we can estimate a possible “base flow” contribution to sea level.

When the continental ice sheets began to melt as the earth transitioned from its Ice Age maximum to our present warm interglacial, sea level began to rise from depths ~130 meters lower than today (see graph below). Melting continental ice sheets drove much higher rates of sea level rise than seen today, ranging from 10 to 40+ mm/year. Approximately 6,000 years ago, a consensus suggests the last of the continental ice sheets had melted completely, the earth’s montane glaciers had disappeared, and Greenland and Antarctic ice sheets had shrunk to their minimums. The earth then entered a long-term 5000-year cooling trend dubbed the Neoglaciation. Although sea level models forced only by growing glaciers and cooling ocean temperatures would project falling sea levels, proxy evidence enigmatically suggests global sea level continued to rise. Albeit at reduced rates, global sea level continued to rise another 4 meters (Figure 1 below). Although there is some debate regards any continued contribution from Antarctica and “ocean siphoning”, according to Lambeck 2014 about 3 meters of sea level were added between 6.7–4.2 thousand years ago. That continued sea level rise could be explained by aquifer discharge, suggesting a minimal “base flow” of ~1.2 mm/year from groundwater discharge.

Similarly, during the Little Ice Age between 1300 and 1850 AD, montane glaciers as well as Greenland and Antarctic ice sheets, grew and reached their largest extent in the last 7,000 years. Ocean temperatures cooled by about 1 degree. Yet inexplicably, most researchers estimate global sea level never dropped significantly. They report sea levels were “stable” during the Little Ice Age, fluctuating only by tenths of a millimeter. That stability contrasts greatly with the recent rising trend, that has led some to attribute the current rise to increasing CO2 concentrations. However Little Ice Age stability defies the physics of cooling temperatures and increasing water storage in growing glaciers that should have caused a significant sea level fall. However, that seeming paradox is consistent with a scenario in which a “base flow” from groundwater discharge would offset any transfer of waters to growing Little Ice Age glaciers.

Once the growth of Little Ice Age glaciers stopped, and groundwater base flow was no longer offset, we would expect sea levels to rise as witnessed during the 19th and 20th centuries. Such a scenario would also explain Munk’s enigma that sea level rise had started too early, before temperatures had risen significantly from any CO2-driven warming.

Interestingly, assuming a ballpark figure of a 1.2 mm/year groundwater base flow, unbalanced groundwater discharge could also explain the much higher sea levels estimated for the previous warm interglacial, the Eemian. Researchers estimate sea levels ~115,000 years ago were about 6 to 9 meters higher than today. That interglacial has also been estimated to have spanned 15,000 years before continental glaciation resumed. Compared to our present interglacial span of 11,700 years, an extra 3,300 years of groundwater discharge before being offset by resumed glacier growth, could account for 4 meters of the Eemian’s higher sea level.

Recent glacier meltwater contribution to sea level is likely overestimated?

In addition to a groundwater base flow driving the current steady rise in sea level, meltwater from retreating Little Ice Age glaciers undoubtedly contributed as well. But by how much? Researchers have estimated there was greater glacial retreat (and thus a greater flux of meltwater) in the early 1900s compared to now. So, current glacier retreat is unlikely to cause any acceleration of recent sea level rise. Furthermore, we cannot assume glacier meltwater rapidly enters the oceans. A large proportion of meltwater likely enters the ground, so it may take several hundred years for Little Ice Age glacier meltwater to affect sea level.

How fast can groundwater reach the ocean? Groundwater measured in the Great Plains’ Ogallala Aquifer can flow at a higher-than-average seepage rate of ~300 mm (~1 foot) in a day, or about the length of a football field in a year.  For such “fast” moving groundwater to travel 1000 kilometers (620 miles) to the sea, it would require over 10,000 years! Most ground water travels much slower. The great weight of the continental glaciers during our last ice age, applied such great pressure that it forced meltwater to into the ground at much greater rates than currently observed recharge. And that Ice Age meltwater is still slowly moving through aquifers like the Ogallala.

(However, its release to the ocean has been sped up by human pumping. Recent estimates suggest that globally, human groundwater extraction currently exceeds rates of water capture from dam building, so that groundwater depletion is now accelerating sea level rise.)

How much of the current meltwater can we expect to transit to the ocean via a slow groundwater route? That’s a tough question to answer. However, thirteen percent of the earth’s ice-free land surface is covered by endorheic basins as illustrated by the gray areas shown in the illustration below. Endorheic basins have no direct outlets to the ocean. Water entering endorheic basins only return to the sea via evaporation, or by the extremely slow route of groundwater discharge. Any precipitation or glacial meltwater flowing into an endorheic basin could require centuries to thousands of years to flow back to the oceans.

For example, in 2010-2011, researchers reported that a La Nina event had caused global sea level to fall by the equivalent of 7mm/year (~0.3 inches/year). That dramatic drop happened despite concurrent extensive ice melt in Greenland and despite any base flow contribution.  As described by Fasullo (2013), GRACE satellite observations detected increased groundwater storage caused by higher rates of rainwater falling on endorheic basins, primarily in Australia. Although satellite observations suggested much of the rainwater remained in the Australian basin, sea level resumed its unabated rise as groundwater base flow contribution would predict.

To balance their sea level budgets, researchers assert melting glaciers have added ~0.8 mm/year to recent sea level rise. The 20th century retreat of most glaciers is undeniable, but we cannot simply assume all 20th century glacier meltwater immediately reached the oceans. The greatest concentration of ice, outside of Greenland and Antarctica, resides in the regions north of India and Pakistan, in the Himalaya and Karakoram glaciers. Most melt water flowing northward enters the extensive Asian endorheic basins. Likewise, some of the Sierra Nevada meltwater flows into Nevada’s Great Basin, and some Andes meltwater flows into the endorheic basins of the Altiplano and Lake Titicaca as well as the Atacama Desert. It is very likely much of the current glacial meltwater will then take decades to millennia to reach the ocean and has yet to impact modern sea levels. If the glacial melt water contribution to sea level is overestimated, then, the unaccounted-for contribution to sea level rise becomes much larger than initially thought.

Accurate Attribution of Groundwater Discharge and Recharge Will Constrain Sea Level Contributions

Using a combination of GRACE gravity data that measured changes in ocean mass, altimetry data that measured changes in ocean volume and Argo data that measured heat content, Cazenave (2008) used 2 different methods and both estimated the contribution from increased ocean heat to be about 0.3 to 0.37 mm/year. Jevrejeva (2008) calculated a similar heat contribution. Other researchers suggest thermal expansion contributes 1.2 to 1.5 mm/year (i.e. Chambers 2016). Such large discrepancies reveal contributing factors to sea level rise are not yet reliably constrained.

One of the great uncertainties in sea level research are glacial isostatic adjustments.
Researchers have subjectively adopted various Glacio-isotatic adjustment models with recommended adjustments ranging from 1 to 2 mm/year. For example, although GRACE gravity estimates had not detected any added water mass to the oceans, Cazenave (2008) added a 2 mm/year adjustment, as illustrated from her Figure 1 below. Other researchers only added a 1 mm/yr adjustment.

In the Gregory (2012) paper Twentieth-Century Global-Mean Sea Level Rise: Is the Whole Greater than the Sum of the Parts? researchers suggested the sea level budget could be balanced and the IPCC’s unaccounted for contribution to sea level rise could be explained by making 5 assumptions:

1)    Assume the contribution from glacier melting was greater than previously estimated.
But greater melting rates were documented for the 30s and 40s, and the likelihood that some glacier meltwater is still trapped as groundwater, suggests the glacier meltwater contribution has been overestimated.
2)    Assume an increased contribution from thermal expansion.
But Argo data suggests the contribution from thermal expansion has been decreasing and plateauing.
3)    Assume Greenland positively contributed to sea level throughout the entire 20th century.
Greenland has undoubtedly contributed to episodes of accelerating and decelerating sea level changes, but the greatest rate of Greenland warming occurred during the 1920s and 30s. Previous researchers suggested Greenland glaciers have oscillated during the 20th century but had been stable from the 60s to 1990s.  Although there was increased surface melt in the 21st century, culminating in 2012, that melt rate has since declined. And according to the Danish Meteorological Institute, Greenland gained about 50 billion tons of ice in 2017 which should have lowered sea level in 2017.  Clearly Greenland cannot explain the enigmatic steady 20th century sea level rise.

4)    Assume reservoir water storage balanced groundwater extraction.
But net contributions from groundwater extraction vs water impoundments and other landscape changes are still being debated. For the period 2002–2014 landscape changes have been estimated to have reduced sea level by −0.40 mm/year versus IPCC estimates of contributing 0.38 mm/year from 1993–2010 to sea level rise.

5)    Assume the remaining unaccounted contribution to sea level rise is small enough to be attributed to melting in Antarctica. 
Debatably, Antarctic melting is too often used as the catch-all fudge factor to explain the unexplainable. Furthermore, there is no consensus within the Antarctic research community if there have been any human effects on Antarctica’s ice balance. Regions that are losing ice are balanced by regions that are gaining ice. Claims of net ice loss have been countered by claims of net ice gain such as NASA 2015. Additionally, unadjusted GRACE gravity data has suggested no lost ice mass and all estimates of ice gains or loss depend on which Glacial Isostatic Adjustments modelers choose to use. We cannot dismiss the possibility that unaccounted for groundwater discharge has been mistakenly attributed to hypothetical Antarctic melting? 

A better accounting of natural groundwater discharge is needed to constrain the range of contributions to sea level rise suggested by researchers such as Gregory 2012. The greater the contribution from groundwater discharge, the smaller the adjustments used to amplify contributions from meltwater and thermal expansion. Until a more complete accounting is determined, we can only appreciate Munk’s earnest concern. How can we predict future sea level rise if we don’t fully understand the present or the past?