Adjusting past hydrology for changes in climate

Figure 1: 120-year trend in growing wet season share of annual flow in California’s Central Valley (DWR data)

by Jay Lund

Segal’s Law: “Someone with one watch knows what time it is. Someone with two watches is never sure.”  Time is certain, but its estimation and measurement are uncertain, yet we are not in total ignorance.

Many water management and regulation decisions require an understanding of current and future hydrology.  These include regulatory decisions on new water rights, plans and design for habitat restoration projects, long-lived water infrastructure (conveyance, storage, and levees, etc.), water demands (orchards and vines), groundwater sustainability plans and policies, negotiating long-term agreements and contracts among water agencies and water users, etc.

Current and future hydrology are busy topics involving uncertainties in precipitation, streamflows, groundwater recharge, and evapotranspiration and evaporation off the landscape and crops.  Many of these are changing with the climate and the evolution and management of watershed lands.  Details of these changes are uncertain.

Representing future hydrologies?

Every year California seems to see a different hydrology.  Much of this variability is natural for California’s Mediterranean climate.  Climate change is making California’s high-variability climate more variable, seasonally and between years.

The future has always been uncertain, and changes over time. There is only one way the climate does not change, but many ways that it can and is changing.

Several approaches can be taken to represent our understanding of uncertain future hydrology in water system reliability analyses, especially given that our understanding of the future is imperfect:

Guessing future hydrology is probably not the best approach, as its results are usually pretty inaccurate.  Intuition is also a generally a poor guide.  People tend to guess or intuit poorly, relative to more systematic approaches to preparing for the future.  But in retrospect, someone will appear to be brilliantly lucky.

Worst case plausible hydrology? For some decisions, the consequences of failure are so bad, relative to the costs of avoiding failure, that we plan and design for an understanding of the worst plausible hydrology.  We do this, for example, with spillway design.  The downstream consequences of a major dam failure are usually so high, relative to the costs of building a bigger spillway that we usually design spillways for a “probable maximum flood,” which assumes a compounding of hydrologic extremes (the “probable maximum precipitation” running quickly off a thoroughly wet watershed into completely full reservoirs).  So spillway failures are usually not from poor hydrologic estimation, but from other failure mechanisms.  Nevertheless, climate change could increase the “probable maximum flood” for different locations, and so worst case estimates might change over time.

Historical hydrology?  Past experience gives scientific, statistical, and practical insights for what could happen in the future.  California’s 100-year record illustrates nicely the high variability of its hydrology for most years, although it has always been naive to think this record alone has been adequate to manage for hydrologic extremes.  Many water and environmental decisions are based on records of historical hydrology.  Sometimes decisions are based on the worst recorded event, an estimated 100-year flood, extrapolations from the historical record, or a probability distribution of performance based on the hydrologic record.  It is long recognized that the historical record alone has limitations for estimating future hydrology, particularly hydrologic extremes.

For planning and operation purposes, the historical hydrologic record often is adjusted for past or projected changes in land and water management, to account for their expected effects on streamflow.  There is no reason not to make similar adjustments for other sources of “non-stationarity,” such as climate change, especially where trends in streamflow changes are well-demonstrated in the recent hydrologic record. Statistically significant shifts of streamflow from spring to winter seasons are occurring in recent decades, roughly shifting an addition 1% of annual runoff per decade (Aguado et al. 1992; Shelton 1998), analysis repeated in Figure 1 below.  Additional adjustments to reservoir evaporation and sea level can be made with similar confidence.

Supplements to historical hydrologic records are sometimes made using paleo-hydrologic records, such as from tree rings and lake sediments.  These can include extreme events and climate changes beyond the historical climate. Their use to augment or adjust historical hydrologic characterizations is clouded by the coarseness of these past flow estimates, which are often only annual or decadal estimates.  Nevertheless, these are insightful cases (Harou et al. 2010).

Climate model results?  Many climate projections exist for California.  Each projection combines a climate model, a global emission scenario, and bias correction, with streamflows derived from these results after various techniques for downscaling, land cover projection, and precipitation-runoff and evaporation modeling.  These model results all show temperature and evaporation increases and shifts of streamflow from spring to winter, with more hydrologic extremes (floods and droughts).  However, the magnitudes of these changes vary considerably across models, modeling scenarios, and post-processing methods and assumptions.

The use of climate model results to replace or adjust hydrologic records often yields useful impact and adaptation insights.  These models are continually updated and improved.  These newer models and modeling approaches tend to confirm changes in temperature and sea level that are well-established over the past three decades, sometimes with incremental additional insights for policy-making and management.

However, climate change model projections of hydrology rarely reflect the broader range of uncertainties involved in their calculations or the broader spatial and temporal hydrologic patterns and uncertainties reflected in the historical record.  The variances could be greater than we think.

Hybrid Approaches. Both the historical record (including recent statistically reliable trends in climate change) and long-term climate models provide important insights for estimating likely and extreme future hydrologies.

Some climatic change directions are certain and solidly observed (such as rising temperatures and sea levels), but the exact extent and timing of climate changes are very uncertain.  This makes adjustment (or replacement) of the hydrologic record to represent our understanding of current and future hydrology uncertain and potentially controversial (both scientifically and for policy).  This conundrum seems unavoidable. 

Some form of hybrid approach to estimating the range of likely future hydrologies seems appropriate, involving both historical and recent hydrologic records, and climate change modeling.

Making adjustments

There are precedents and examples for how to adjust the historical hydrology for hydrologic non-stationarities in water planning and management.  Hydrologists routinely adjust the historical hydrologic record for changes in land and water use in the past and expected in the future.  Such adjustment are long-standing in the development of hydrologic records for water planning and operations, adjusting past flow measurements and estimates for current and anticipated future conditions.

Although there is consensus on the need for such adjustments, these adjustments are not without differences in scientific and technical methods and judgments.  The resulting flow differences will probably be large enough to attract policy controversies since many, sometimes opposite, policies are advocated as adaptations or mitigations to climate change.

Nevertheless, there should be broad support for technically-driven adjustments to historical flow records based on observed firm changes in climate, such as seasonal runoff shifts and some amount of sea level rise.  Statistical adjustments for increasing extremes also seem desirable, but are more scientifically difficult.  Greater amounts of climate change might be explored with interest, perhaps insightfully, for planning and policy decisions.

Over time it will be important to make further adjustments, perhaps based on new scientific and technical work.  The climate is changing.  Some climate changes are happening with as much certainty as routine hydrologic adjustments we make for future land use and water demands.  There seems firm ground for making such climate change adjustments.

Further Reading

Aguado E, Cayan DR, Riddle LG, Roos M (1992) Climatic fluctuations and the timing of west coast streamflow. J Clim 5:1468–1483

Harou, J.J., J. Medellin-Azuara, T. Zhu, S.K. Tanaka, J.R. Lund, S. Stine, M.A. Olivares, and M.W. Jenkins (2010), “Economic consequences of optimized water management for a prolonged, severe drought in California,” Water Resources Research, doi:10.1029/2008WR007681, Vol. 46, 2010

Mote PW, Hamlet AF, Clark MP, Lettenmaier DP (2005) Declining mountain snowpack in Western North America. Bull Am Meteorol Soc 86(1):39–49

Roos M (1987) Possible changes in California snowmelt patterns. In: Proceedings fourth annual pacific climate (PACLIM) workshop, Pacific Grove, CA, pp 22–31

Shelton ML (1998) Seasonal hydroclimate change in the Sacramento River Basin, California. Physical Geography 19(3):239–255

Jay Lund is a Professor of Civil and Environmental Engineering at the University of California – Davis, where is also Co-Director of the Center for Watershed Sciences.

About jaylund

Professor of Civil and Environmental Engineering Director, Center for Watershed Sciences University of California - Davis
This entry was posted in Uncategorized. Bookmark the permalink.

2 Responses to Adjusting past hydrology for changes in climate

  1. Michael Mierzwa says:

    Thanks for the thoughts Jay! Another thing that I’d like to point out (an area that as a society we do poorly) is in making investments that actually not only address future flood risk, but do so in a way that does not intensify the consequences when there is a failure. Specifically I’m speaking about the tendency for us to build taller *and* deeper levees, which ironically only increases the hydrostatic pressure on these (crude) flood defense systems and at the same time decreases the awareness of the overall, long-term risk. There are many other structural and non-structural solutions that in non-urban areas are much more cost effective and resilient to uncertain hydrologic hazards (floods) — such as setbacks, detention basins, bypasses, elevation, and dare I say it … managed retreat.

    With that last subject in mind, I think there was an interesting article about an entire Welsh village that is going to be below (actually already there) sea level, and the United Kingdom has determined that by mid-century that it is cheaper to relocate the entire community. The driver in their decision wasn’t the absolute water surface elevation due to sea level rise, but rather the *trend*.

    So while there is uncertainty in our best hydrologic estimates (your two watch example), there is generally great value in the trends to the two watches provide. Perhaps a better value would be calendars. Two different calendar systems might disagree which exact day we are in, but they are useful in letting us know the season, which in turn is key for planting and storing food.

    • jaylund says:

      Thanks Mike for these useful thoughts. The flood planning and preparation problem is even harder since we are usually concerned with even more extreme events, which are harder to adjust for climate change.

      We have done some Bayesian decision-analysis for this problem to bring in a range of climate projection results:
      Hui, R., J. Herman, J. Lund, and K. Madani, “Adaptive Water Infrastructure Planning for Nonstationary Hydrology,” Advances in Water Resources, Vol 118, pp 83-94, August 2018.
      and a nearly-recent dissertation applied this to future expansions of the Yolo Bypass:
      Alessia Siclari Melchor, PhD, “Hydro-Economic Modeling of Flood Bypasses,” 2018

Leave a Reply to jaylund Cancel reply