Implications of irrigation seepage for fisheries management

I’ve been traveling much of the past two weeks and haven’t had time to post a blog lately.  This week’s post is the presentation I gave at the Idaho Chapter American Fisheries Society meeting in Boise on March 4.

Here is a link to the slides, in pdf format. An explanation of each slide is given below.

  1. The title is a little long and not necessarily descriptive of what I actually ended up talking about. That’s the consequence of having to submit a title and abstract a couple of months ahead of the meeting. I didn’t know at the time I submitted the abstract that Brian Patton would be giving a talk on aquifer management earlier in the day. Brian is the Planning Bureau Chief at Idaho Department of Water Resources and executive officer of the Idaho Water Resource Board. His talk provided background on groundwater management in Idaho, with an emphasis on the Eastern Snake Plain Aquifer.
  2. This presentation contains original material from my field and modeling work over the past ten years, done with funding from numerous sources and in collaboration with a large number of faculty and student colleagues.
  3. Although most of the slides cover hydrology, the main message is contained in the last slide, which presents important implications of groundwater management for fisheries.
  4. The hydrologic effects of large dams on stream hydrologic regimes are well known and include reduction in peak flows, reduction in delivery of sediment, decreased flow variability, and simplification of the stream channel and floodplain. This hydrograph compares natural and regulated flows in the Snake River downstream of Palisades Dam, and it is clear that the regulated peak is much lower than the natural peak, as spring runoff is stored in the reservoir for delivery later in the summer during irrigation season. However, much less is known about the hydrologic effects of irrigation diversion, seepage from irrigation systems into the ground, and return flow through groundwater pathways. The primary research question addressed here is: what are the consequences of irrigation seepage and return on stream hydrologic regimes?
  5. Our research approach uses a mass balance calculation—essentially water in and water out. The hydrologic models I have developed track flow through the coupled ground-surface water system. Water originating in streams and rivers, and perhaps stored temporarily in reservoirs, is diverted into canals, which seep water into aquifers before delivering it to crops. Some of the water that remains in stream channels also seeps into these same aquifers. When water is delivered to fields, some of it is used by crops (this is evapotranspiration, or ET). Water in excess of crop ET either flows back into the river via the surface conveyance system of drains and canals or seeps into the aquifer. Water that enters the aquifer flows through the ground and eventually back into streams and rivers. This basic model was used to analyze changes in irrigation practices through time and was applied at three different spatial scales.
  6. The study area is the upper Snake River basin, defined here as the watershed (both groundwater and surface water) of the Snake River upstream of King Hill. The USGS gage station at King Hill is extremely useful for analyzing hydrology of the upper Snake River basin because it is located immediately downstream of Thousand Springs, where the majority of water flowing through the Eastern Snake Plain Aquifer (ESPA) emerges from the ground and rejoins the surface-water system. The other major location of discharge from the ESPA back into the surface system is in the vicinity of American Falls Reservoir. The Henry’s Fork watershed is shown at the upper right corner of the map and serves as an intermediate-scale basin. The smallest watershed used in this analysis is that of the upper Teton River, itself a sub-basin of the Henry’s Fork watershed.
  7. The three nested study watersheds represent three different orders of magnitude in size. What is most interesting about the numbers in this table is that the water supply and amount diverted for irrigation are roughly proportional to basin size. In other words, within the upper Snake River basin, as the size of a watershed unit increases, the supply of water (precipitation) and the amount diverted for irrigation increase proportionally. If the basin size increases by a factor of 10, so do water supply and irrigation withdrawal.
  8. This graph shows the water budget for water withdrawn from streams and rivers for irrigation. Although there are slight differences among the three study watersheds, the overall pattern is constant. Return flows via surface canals are minimal, and evaporative losses from reservoirs, irrigation canals, and sprinklers are even lower—just a few percent of total withdrawal. You may have read about larger evaporative losses in places like California and Arizona, where the climate is warmer and evaporation rates are much higher. However, here in Idaho, evaporative losses are very minimal. On the other hand, the largest component of the water budget is seepage to aquifers. This water all returns to the surface system eventually. Between 50% and 70% of the water withdrawn into the irrigation system from streams and rivers eventually returns to streams and rivers. Timing and location of these return flows varies substantially with geology and geography. In some locations, such as the lower Henry’s Fork, water traveling through the aquifer returns to the river only a few miles downstream from where it was diverted, and peak groundwater returns lag peak irrigation withdrawals by only a few weeks. On the other hand, some water diverted for irrigation travels over one hundred miles through the ESPA before discharging back to the river at Thousand Springs, near Hagerman. Only 25-45% of the water withdrawn for irrigation is actually used by the crops.
  9. Having seen the previous graph, it should not be a surprise that irrigation accounts for a large portion, if not the majority, of recharge to local and regional aquifers. At all spatial scales, irrigation seepage accounted for 40-70% of all aquifer recharge. Precipitation and stream-channel seepage make up the remainder, with variability across the study watersheds due primarily to differences in local geology.
  10. Over 130 years of seepage incidental to irrigation in the Snake River basin created a landscape that is much different than the arid, sagebrush steppe ecosystem it replaced. Irrigation created new wetland and riparian habitats and even new streams. Of the two photos on the right side of the slide, the upper one is a stream channel (even though much of the flow in it results from irrigation seepage) and the lower one is an irrigation canal. In eastern and southern Idaho, there is a fine line between canals and streams.
  11. Historically, all irrigation was done by surface application, through flood, furrow and other direct-application methods. Beginning in the 1970s, surface irrigation was replaced by more “efficient” sprinkler application.
  12. The period of most rapid conversion of surface irrigation systems to sprinklers occurred in the 1980s. By 2010, around 90% of the irrigated land in the upper Snake River basin was irrigated with sprinklers. An additional change to irrigation practices was increase in groundwater pumping, which began in the 1950s and peaked shortly thereafter. In the 1970s, the Idaho Department of Water Resources placed a moratorium on new groundwater pumping on the ESPA, so after the initial 20 years or so of groundwater development, the amount of water pumped from the ESPA has remained roughly constant.
  13. This graph shows modeled hydrographs of the upper Teton River under different irrigation scenarios. I used water years 1979-2008 as the modeling period, so that each scenario differed only in the simulated irrigation practices and not in water supply and climatic conditions. The solid blue curve shows the unregulated hydrology, that is, the mean annual hydrograph in the river if no irrigation or other uses of water had occurred. The black dotted curve shows the actual hydrology over that time period. The difference between the blue and black curves during the summer illustrates the effect of irrigation diversion during the period of peak streamflow. As a result, streamflow is lower from May through July than it would be under unregulated conditions. However, notice that around the first of August, the two curves intersect, and from then until around the first of May, the regulated flow is actually higher than it would have been under natural conditions. This is because water that seeped into the ground early in the irrigation season when withdrawals were high has returned to the river from the aquifer, increasing late-summer and winter flows. The aquifer acts as a reservoir, storing water during May, June and July, and releasing it back to the river during late summer, fall, and winter. The red dashed curve shows the hydrograph under simulated flood irrigation conditions, as they were practiced in the first half of the 20th century. Irrigators diverted as much water as they could early in the spring and summer, up to water availability and canal capacity. That is why you see an abrupt decrease in streamflow beginning on April 15, the first day of the legal irrigation season in the Teton watershed. The purpose was to put as much water in the ground as possible and raise the local water table. That increase in groundwater levels provided water to crops later in the summer. The hydrograph also shows that diversion throughout the early summer was generally greater under this practice than under current practices. However, after the middle of July, streamflow was higher under flood irrigation practices than under current conditions, because groundwater returns were higher under flood irrigation practices. During late summer through fall and winter, streamflow under flood irrigation conditions was much higher than under natural conditions or any other type of irrigation practice, again due to large amounts of groundwater recharge and return flows. The dashed green curve shows the stream hydrograph under a hypothetical scenario in which all canals are put into pipelines, and irrigation seepage is eliminated completely. This is the most “efficient” of all of the irrigation scenarios. Diversion is limited to only what crops can use, or the supply, whichever is smaller. All of the diverted water, less a small amount of evaporation from sprinklers, is used by the crops. There is no groundwater seepage under this scenario. There is little difference between current conditions and the pipeline scenario early in the irrigation season, but from about mid-July on, streamflow is lower under the pipeline scenario. During late summer, fall, and winter, streamflow is lower under the pipeline scenario than any of the other scenarios, including the unregulated condition. This occurs because none of the diverted water seeps back to the aquifer, and hence groundwater returns to the river are lower. The difference in total stream discharge between current conditions and the pipeline scenario represents increased consumptive use under the pipeline scenario. Because this scenario is close to perfectly “efficient,” nearly all of the diverted water is used by crops. Under all other irrigation scenarios, a large fraction of diverted water flows through the system unused and returns to the river. The most “efficient” irrigation practice increases crop yields through more uniform irrigation throughout a longer period of the year, but increased crop yield uses more water, resulting in lower total streamflow across the year. Keep in mind that there are no dams and reservoirs in the upper Teton watershed. These striking effects on hydrologic regimes are caused solely by diversion of surface water, consumptive use by crops, and return flows through groundwater pathways. At the scale of the entire upper Snake River basin, the storage capacity of the reservoir system is a little over 4 million acre-feet, less than half of the mean annual water supply. Although these reservoirs have a large effect on streamflow immediately downstream, they actually have relatively little effect at larger spatial scales, where the combined effects of irrigation diversion, crop use, and return flows have the dominant effect on streamflows.   
  14. These graphs show trends in annual discharge in three different groundwater-dominated systems from around the upper Snake River basin. Of course, Silver Creek supports a popular wild trout fishery, but its flow is largely dependent on irrigation seepage, which has been declining for decades. Box Canyon Spring is one of the points of groundwater emergence from the ESPA in the Thousand Springs area. The bottom graph shows estimated groundwater returns to the lower Henrys Fork and Teton rivers. Annual flow at all three of these locations shows the same declining trend over the past three to four decades.
  15. Remember that the King Hill gage measures total flow in the Snake River at the bottom of the upper Snake River/ESPA system. This flow includes both surface flow and groundwater returns. Because the water budget can be closed at King Hill, annual flow at King Hill represents the net amount of water that is not used in the extensive irrigation system upstream. Although total water supply varies from year to year as climatic conditions vary, it has remained roughly constant over long periods of time. The fact that total flow at King Hill has declined indicates that consumptive use of water has increased. This is due to a combination of increased groundwater pumping and increased irrigation efficiency. Even though total diversions from the surface-water system have generally declined over the past three decades, the water that is diverted is applied more efficiently by agricultural users, increasing overall water use. This increase in consumptive use has contributed to a rapidly expanding agricultural economy in southern and eastern Idaho, centered primarily around dairy production in the Magic Valley.
  16. Over the past 130 years, irrigation has been the primary driver of groundwater-surface water interactions and stream hydrologic regimes in the upper Snake River basin. Streamflow regimes have been transformed from snowmelt-dominated to groundwater-dominated. Increase in the groundwater component of hydrologic regimes has created, enhanced, and maintained fish and wildlife and their habitats throughout the basin, albeit at the expense of many native species and habitats that evolved around snowmelt-dominated hydrology. Despite increases in irrigation efficiency and resulting decreases in groundwater recharge, irrigation seepage remains the single largest source of aquifer recharge basin-wide.
  17. The take-home message for fisheries managers is that “things ain’t what they seem.” Traditional “one-size-fits-all” approaches to problems involving streamflow and fisheries do not work well in the current administrative and hydrologic framework that exists in the Snake River basin. Increases in irrigation efficiency almost always lead to increased consumptive use of water, due to physical, legal, and economic realities. Although increased irrigation efficiency can result in a little more water in some stream reaches at certain times, the overall effect is that more water is used, which means less water in the river at some point downstream. Non-traditional management actions such as diverting surface water to recharge the aquifer can actually benefit fish and wildlife, if done at appropriate locations and times. Fish and wildlife interests often coincide with the legal interest of a water-rights holder somewhere in the system, and I encourage fisheries managers to seek the collaboration of those water-rights holders. Fisheries managers need to participate fully in administrative and legal processes surrounding water-rights applications and managed recharge to give fish a voice at the table. However, this voice must be based on detailed, case-specific hydrologic analysis that accounts for the interconnectedness of surface and groundwater flow.
  18. Aerial photograph of the irrigated landscape surrounding the South Fork Snake River near Lorenzo. View is looking southwest down Highway 20; the Lorenzo bridge is in the lower left portion of the photo.