From @PNWClimate: A New Approach to Tracking Hydrologic Responses to Climate Change - Déjà Vu?

The following paper is featured in the July/August 2015 issue of The Climate CIRCulator:

Vano, J. A., B. Nijssen, and D. P. Lettenmaier (2015), 'Seasonal Hydrologic Responses to Climate Change in the Pacific Northwest', Water Resour. Res., 51, 1959–1976. doi:10.1002/2014WR01590

Abstract

Increased temperatures and changes in precipitation will result in fundamental changes in the seasonal distribution of streamflow in the Pacific Northwest and will have serious implications for water resources
management. To better understand local impacts of regional climate change, we conducted model experiments to determine hydrologic

sensitivities of annual, seasonal, and monthly runoff to imposed annual and seasonal changes in precipitation and temperature. We used the Variable Infiltration Capacity (VIC) land-surface hydrology model applied at 1/16° latitude-longitude spatial resolution over the Pacific Northwest (PNW), a scale sufficient to support analyses at the hydrologic unit code eight (HUC-8) basin level. These experiments resolve the spatial character of the sensitivity of future water supply to precipitation and temperature changes by identifying the seasons and locations where climate change will have the biggest impact on runoff. The PNW exhibited a diversity of responses, where transitional (intermediate elevation) watersheds experience the greatest seasonal shifts in runoff in response to cool season warming. We also developed a methodology that uses these hydrologic sensitivities as basin-specific transfer functions to estimate future changes in long-term mean monthly hydrographs directly from climate model output of precipitation and temperature. When principles of linearity and superposition apply, these transfer functions can provide feasible first-order estimates of the likely nature of future seasonal streamflow change without performing downscaling and detailed model simulations.

Here is a PDF of the paper:  Download Vano_et_al-2015-Water_Resources_Research
Supplemental material:  Download Vano_et_al-2015-Water_Resources_Research.sup-1

Here is what  The Climate CIRCulator said about the Vano et al. approach:

Rising temperatures and changing precipitation patterns are leading to major shifts in when and how streamflows occur. But deciphering how climate change might affect any given watershed can be both time-consuming and expensive. Now a new study focusing on the Pacific Northwest offers a quick, inexpensive alternative to traditional climate downscaling and hydrologic modeling.
  
The method is called the sensitivity approach. The brainchild of CIRC and OCCRI researcher Julie Vano, the sensitivity approach is the subject of a recent paper written by Vano and CIRC researchers Bart Nijssen and Dennis Lettenmaier in Water Resources Research. Essentially, the approach tries to determine how sensitive runoff is to incremental changes in temperature and precipitation during different seasons.
 
Watersheds in the Northwest are typically classified as rain-dominant, snow-dominant, or transitional (a mix of both rain and snow), based on their predominant type of precipitation.
 
Rain-dominant watersheds receive mostly rainfall during the cool season, making streamflow highest in the winter. Snow-dominant watersheds receive mostly snow, which remains stored in mountain snowpack until it melts during the spring and summer when peak streamflow occurs. Transitional watersheds receive both rain and snow so they tend to have two streamflow peaks: one in winter and the other in spring.
 
Using the sensitivity approach, Vano and colleagues identified watersheds in the Pacific Northwest that are most likely to experience streamflow seasonality changes.

The team found transitional, or mixed, watersheds that receive a substantial portion of streamflow from spring snowmelt are most sensitive to warming during the cool season, as they are likely to experience increased cool season runoff and decreased warm season runoff. Rain-dominated watersheds are less sensitive to warming because there is little snow to melt. And snow-dominated watersheds are also less sensitive because they are likely to remain cold enough that warming does not affect the timing of snowmelt. Here’s step-by-step how the researchers used the sensitivity approach to get their results:

1. They ran a hydrologic model using baseline historical temperature and precipitation data (taken from the Columbia Basin Climate Change Scenarios Project). This produced a series of simulated streamflows.
2. Then they ran the model again, but with temperatures 0.1°Celsius (0.2°F) higher every day of the year, producing still more streamflows.
3. Next, they compared streamflows from step 2 with the baseline streamflows from step 1 to determine how much runoff changes at each location and in each month occurred in response to the 0.1°C temperature increase.
4. They repeated steps one through three, but this time changed the temperature only during the cool season or warm season, or fall, winter, spring, or summer.
5. They then repeated the entire process by imposing a precipitation increase of one percent while keeping temperature the same.
6. And finally, they multiplied the resulting temperature and precipitation sensitivities for each location and month by the projected change in temperature or precipitation that came directly from Global Climate Models (no downscaling necessary) to approximate future changes in streamflow.

So just how well does the sensitivity approach measure up to other methods? The resulting projected monthly hydrographs are strikingly similar to those generated using the “full-simulation” approach in which Global Climate Model data is first downscaled and then run through a hydrological model.
 
These results suggest that Vano’s method can capture the projected seasonality shift quickly (not to mention inexpensively) compared to the full-simulation method, but there are a couple caveats. The approach only works so long as the runoff sensitivity to small changes (e.g. 0.1°C or nearly 0.2°F) is the same as the sensitivity to large changes (e.g. 3°C or 5.4°F), and changes in individual months or seasons can add up to equal the projected annual changes. It’s no surprise then that the sensitivity approach works best when considering near-term change (roughly 30 years) and in locations with small sensitivities. What’s more, while the approach quickly provides insights into the nature of average changes in the hydrograph, it does not provide daily flow sequences as in the “full-simulation” approach. (Subsequent work by Vano and other OCCRI and CIRC colleagues yet to be published explored sensitivities to large temperature change.) 
 

The approach described seems somewhat similar to what we hydrologists did in the 'old days' (i.e., pre-climate change, pre-GCMs), when changes in precipitation were input to a rainfall-runoff model to predict incremental changes in runoff due to incremental changes in precipitation. The tricky part was deciding what the increments in precipitation should be in order to produce meaningful (realistic) results. 

This was often done with respect to 'ephemeral watersheds' - those in which there was flow only after a precipitation event. Such watersheds are notorious nonlinear. In 1983, one of my MS students, Tony Truschel, did his thesis using a similar approach (I'm going from memory now - a dangerous thing): A reservoir-routing model calibration method relating storage elements to basin geomorphology for peak runoff prediction from extreme summer storm events in ungaged arid watersheds. 

I am not saying that Vano et al. are reinventing the wheel; they cite papers from over 20 years ago. They are also dealing with far more complex and variable systems - both hydrologically and climatologically - than we could handle or even imagine 30+ years ago. In addition, the models today are nothing like we used in 1983. Tony's model was a simple reservoir storage model that did not account for groundwater, ET, or anything else. He just routed flow from one storage element to the other; I think he used about 20 or 25 elements. The VIC model used by Vano et al. is far more complicated than what we had in the 1980s. I know, I built some of those models. 

So déjà vu? I don't think so; but some 'crude' similarities. Good job, folks. Sure easier than downscaling GCMs, and that's the point.

"What goes around, comes around." - Unknown