COHYST SUMMARY OF FINDINGS Pilot ProjectWood River WatershedSurface Runoff ModelWood River at Riverdale and AldaPresented toCOHYST SponsorsFebruary 3, 2010
TWO VERSIONS • Two versions of this presentation are available • Sponsor’s version – this one • Tech Committee version – detailed • The Tech Committee version has greater detail with footnotes, and constitutes the final report for this project
WOOD RIVER PILOT STUDYPresentation format • Detailed Presentation to T.C. on 1/19 • Lengthy Discussion w/Numerous Suggestions • Too Much Info to Cover Everything • Same Results as T.C, Just Less Detail • Focus on Key Findings vs. Details • Several “Samples” of Results • 3-Slide “Primer” on CN Method
HANDOUTS • Inventory of Key Spreadsheets* and Contents • Descriptions of Six “Tests” with Commentary • Copies of Slides *All posted on ftp site under “Wood River Pilot Study/New Folder”
July 12, 2007 Proposal to SponsorsSurface Water Model Options • Annual CN Model – Develop a model to estimate land use impacts on surface runoff on an annual basis using annual rainfall and composite CNs for subareas of the test watershed ($5K) . • Monthly CN Model – Develop a model to estimate land use impacts on surface runoff on a month-by-month basis using monthly rainfall and composite CNs for subareas of the test watershed ($10K). • Weekly CN Model – Develop a model to estimate land use impacts on surface runoff on an annual, monthly, and week-by-week basis using weekly rainfall and spatially-varied, GIS-distributed CNs to determine field runoff ($25K). • Daily CN Model – Develop a model to estimate land use impacts on surface runoff on a daily basis using time-varied, GIS-distributed CNs to calculate daily values of field runoff, then route the field runoff to the river using time of travel methods ($100K). No. 3 was approved, but actual effort was ~ 90% of No. 4
PILOT STUDY OBJECTIVES • Develop and perform a pilot test of a spatially-distributed, CN-based weekly daily rainfall-runoff model for the COHYST area that can predict the short and long-term impacts of land use changes on runoff from rainfall. • Evaluate whether the CN method can be used to reveal anything about the validity of the recharge “bump” used in the COHYST models.
PILOT STUDY OBJECTIVES • Compare CN Model estimates of field-level runoff with WHAT estimates of direct surface runoff at the gauges (not a definitive study of WHAT). • Recommend how the CN method can best be applied in future COHYST (and Conjunctive Management) work.
IS “CN” AN ACCEPTED TOOL FOR predicting impacts of land use changes? ASCE Task Force, “Curve Number Hydrology, State of the Practice,” 2009, p. 73: “Applying the CN technology to determine land use influences on hydrology should be encouraged.”
“Primer” SCS CN Method Decreasing CN CN = 90 Smaller CNs give less runoff for the same rain amount CN = 70
CN TABLE Effect on CN (and runoff) of converting from row crops to small grain crops to meadows to pasture Direction of decreasing CN and decreasing runoff for same rain
EXAMPLES – EFFECTS OF CN ON RUNOFF FOR SAME RAIN AMOUNT AMC = “Antecedent Moisture Condition”: I = dry, II = normal, III = wet
PROCEDURE PILOT TEST OF CN-METHOD • Determine total daily edge-of-field runoff from the area above two stream gauges by CN method during two 7-yr test periods for (a) individual runoff events, (b) monthly runoff, (c) growing season runoff, and (d) annual runoff. • Identical to method used in Cropsim • Determine the gauged amount of streamflow from USGS discharges for the same periods. • Compare and interpret differences between Nos. 1 and 2.
PROCEDURE PILOT TEST OF CN-METHOD • Determine total daily edge-of-field runoff from the area above two stream gauges by CN method during two 7-yr test periods for (a) individual runoff events, (b) monthly runoff, (c) growing season runoff, and (d) annual runoff. • Identical to method used in Cropsim • Determine the gauged amount of streamflow from USGS discharges for the same periods. • Compare and interpret differences between Nos. 1 and 2. • What could the differences in Nos.1&2 represent?
Model CellsRiverdale and Alda GaugesRiverdale 58% of Total Area above Alda 163,186 ac 221,737 ac
KEY QUESTIONS ANSWERED TODAY • What were the results of comparing the edge-of-field runoff by the CN method with gauged flows (and WHAT estimates of direct runoff) at Riverdale and Alda? • By individual “Identifiable” Events? • By “Identifiable” Events as a group? • Monthly? • Growing Season? • Annual? • What do the findings imply regarding how COHYST (and Conjunctive Management) should proceed? Key results spreadsheets are posted on ftp site under “Wood River Pilot Study.”
“IDENTIFIABLE” RUNOFF-RAINFALL EVENTS • Definition: • Stand-alone direct runoff hydrograph with 1- to 10-day rain, and no apparent AMC I, III “Problem” * • How Found (21 years of daily flows): • Use WHAT estimates to identify beginning and ending dates of direct runoff hydrographs. • Examine daily precipitation to assess dates and amounts of rainfall that produced each hydrograph. • Determine CN model estimates for runoff on each day of rain that produced the hydrograph. *The “AMC I, III Problem” is described in more detail later.
“IDENTIFIABLE”* RUNOFF-RAINFALL EVENTS • Riverdale Gauge, 1964 – 1970 35 Events • Alda Gauge, 1964 – 1970 6 Events • Alda Gauge, 1999 – 2005 15 Events *Isolated runoff event with 1- to 10-day rain, and no apparent “AMC I, III Problem.”
ALGORITHM DEFININTIONS Two algorithms were used for CN modeling, and assessments of each are described later: • AMC I, II, III Algorithm • This is a built-in part of the SCS CN Method, also used in CROPSIM • Irrigation Algorithm for Irrigated Crops • This was added by the Technical Committee
AMC I, II, III algorithm*Applied Daily 5-day Antecedent P: P < 1.4” 1.4” < P < 2.1” 2.1” < P Then: Switch CN to AMC I Switch CN to AMC II Switch CN to AMC III *Growing season algorithm used in both Cropsim and COHYST.
PARTIAL AMC I, II, III CONVERSION TABLE These are discrete steps, not uniformly distributed.
LITERATURE ON AMC I, II, III METHOD • Data for published tables “cannot be found.” • AMC classes I, II, III “discouraged” by NRCS in 1993 but admit that “application is widespread.” • 2008 NRCS recommended ARC I, II, III Curves based on frequencies of runoff events. • Can also use soil moisture accounting such as in SWAT where ARC I and III are the soil’s wilting point and field capacity, respectively.
“IRRIGATION ALGORITHM” The edge-of-field runoff for both the Riverdale and Alda watersheds was modeled for both 7-year periods with and without the “irrigation algorithm” running. Because the CN method doesn’t distinguish between irrigated crops and dryland, this algorithm was created by the Technical Committee in order to evaluate runoff for irrigated crops. The algorithm adds 1” of water in 3 days whenever a 4-day rainless period precedes “today.” This caused the CN method to recognize that the soil would be “wetter” if the crop was irrigated, increasing the CN from I to II, or II to III depending on the 5-day amounts. (Note: Tests with this showed that about 10 applications of irrigation water would occur during the growing season)
TESTS CONDUCTED • Test #1: Run the Riverdale runoff model for the 1964-1970 growing seasons with the 3-day “irrigation algorithm” applied to all lands classified as irrigated • Test #2: Run the Riverdale runoff model for the 1964-1970 growing seasons without the irrigation algorithm • Test #3: Run the Alda runoff model for the 1964 – 1970 period with the irrigation algorithm on all lands classified as irrigated • Test #4: Run the Alda runoff model for the 1964 – 1970 period without the irrigation algorithm • Test #5: Run the Alda runoff model for the 1999 – 2005 period with the irrigation algorithm on all lands classified as irrigated • Test #6: Run the Alda runoff model for the 1999 – 2005 period without the irrigation algorithm • Test #6a: Run the Alda model for the 1999 – 2005 period with all land classified as irrigated converted to “native” conditions (pasture & rangeland) • Test #6b: Run the Alda model for the 1999 – 2005 period with all crops (irrigated and dryland) converted to “native” conditions (pasture & rangeland)
SUMMARY OF FINDINGSWith and Without Irrigation Algorithm • RIVERDALE SUBWATERSHED, 1964-1970 (Tests 1 & 2) • ALDA WATERSHED, 1964-1970 (Tests 3 & 4) • ALDA WATERSHED, 1999-2005 (Tests 5 & 6, and 6a and 6b)
Riverdale Subwatershed RESULTS • Compare CN-Based Annual Estimates of Direct Runoff to Gauged Flows and WHAT D.R. Values at Riverdale • By Individual “Identifiable” Events • By “Identifiable” Events as a Group • Monthly • Growing Season • Annual • Is there base flow at Riverdale?
IS THERE BASE FLOW AT RIVERDALE? Distance, miles above Alda Gauge
RIVERDALE RESULTS: “IDENTIFIABLE” EVENTS35 Events, 1964-1970 Growing Season
Riverdale ResultsMonthly, Tests 1 and 2 “BEST” Years 1965, 1967
Riverdale ResultsMonthly, Tests 1 and 2 “WORST” Years 1964, 1969
Riverdale ResultsMonthly, Tests 1 and 2 “WORST” Years 1964, 1969
Growing season Results of Tests 1 and 2Comparison of Growing Season CN Model Results with and without Irrigation Algorithm Running at Riverdale Observations: Gauged Flow was 42,700 af (0.193”), Total Precip was 99.46” Moderately variable by year; both models had EOF runoff for most months < gauged, suggesting possible base flow in gauged amounts. WHAT direct runoff at gauge = 30,407 af (71 percent of gauged flow), suggesting “loss” of 10,504 af (0.05”)
RIVERDALE ANNUAL RESULTSCN Runoff vs. Gauged Discharge • Pilot Study only tested growing season edge-of-field CN runoff against growing season gauged discharge. • But, SCS annual runoff charts* are available, which are also CN-based. *SCS Engineering Field Manual, Notice from the SCS Nebraska State Office as a Supplement to the Standard Field Manual.
SCS Runoff Charts for Annual Runoff Estimates Provide annual runoff estimates in inches everywhere across Nebraska as function of watershed CN and chance (%) of annual rainfall amount (50% chance = normal year)
Isogram 2.5 Isogram 3.0 Annual Runoff, ac-in/ac 10% 50% Chance Of Occurrence 80% 70 CN SCS Engineering Field Manual Wood River ~ Isogram 2.5
Check 1964-70 Runoff by SCS Runoff Charts - Riverdale Normal Year (1968) Data (50 % Chance Curve) 23.22” (NOAA) 24.39” (105 % of Avg.) 0.26 inches * 0.148” (57 % of Gauged) 0.257” (99 % of Gauged) 0.460” ( 177% of Gauged) “on the nose” if CN = 70 • Average Annual precip: • 1968 Kear/Ocon precip: • Gauged Flow CY 1968: • Chart Runoff if CN = 65: • Chart Runoff if CN = 70: • Chart Runoff if CN = 75: • Conclusion: SCS chart is *Note: 1968 Growing Season Discharge was 0.18” (of 0.26” total gauged)
GROWING SEASON RUNOFF COMPARED TO ANNUAL Reasonable?
FINDINGS: BASEFLOW QUESTION RIVERDALE GAUGE, TESTS 1&2 • Total gauged flow for entire 1964-1970 growing seasons = 42,700 af • CN model with irrigation estimated edge-of-field runoff = 40,911 af (96% of Gauged) • WHAT method direct runoff at gauge = 30,407 af (71% of Gauged) Observations: Even with the data on elevations, the fact that EOF < Gauge suggests there may be “base flow” at Riverdale; saw earlier that for single runoff-rainfall events, WHAT method runoff at gauge reasonably agrees with CN method EOF runoff.
CONCLUSIONSRiverdale Watershed (Tests 1 & 2) • SCS annual charts are surprisingly accurate. • Gauged, WHAT, & CN runoff are < 2.3% of 99” rain. • CN model with irrigation algorithm estimates growing season runoff as 96% of gauged flow. • Because CN model edge-of-field runoff < gauged flow, suggests there is some base flow in the gauged amounts. • WHAT predicts that 71% of the gauged is direct runoff; CN predicts that 96% of gauged is available at EOF, so these are not out of line.
ALDA WATERSHED RESULTS Paired Runs 3-4; 5-6; 6a-6b 1964-1970 Growing Seasons 1999-2005 Growing Seasons By individual “Identifiable” Events By “Identifiable” Events as a group Monthly Growing Season Annual
Cn MODEL Results for 15 “Identifiable” runoff-rainfall Events at alda for 1999 to 2005 EVENTS AS A GROUP Model Runoff w/irrig, Percent of Gauged: 50% Model Runoff w0/irrig, Percent of Gauged: 37% WHAT Runoff, Percent of Gauged: 48%
Cn MODEL Results for 6 “Identifiable” runoff-rainfall Events at alda for 1964 to 1970
COMPARISON* OF 4 OVERLAPPING 1964-1970 EVENTS AT RIVERDALE AND ALDA • 4 “Identifiable Events” logged at both gauges. • Ending dates of 07/09/1965, 07/11/1967, 06/30/1968, 06/28/1969 . • Average rain during four events ranged from 1.5” to 4.9”. • Average rain over Alda matched Riverdale within ± 8 percent. • Gauged (and modeled) runoff at Alda always > Riverdale. • If CN runoff > gauged at Riverdale, also > at Alda. • If CN runoff < gauged at Riverdale, also < at Alda. • Modeled EOF in./ac values are within 0.001” for both gauges. *Results are detailed in “Runoff Event Analysis for Alda 1964 to 1970.xlsx”
Compare Tests 3 and 4 – 1964 to 1970Comparison of Growing Season CN Model Results with and without the Irrigation Algorithm Running at Alda Observations: [Total 7-season Gauged Flow was 62,384 af (1.94”) , Rainfall = 103.2”] WHAT 7-yr direct runoff at gauge = 40,529 af (1.26”), which is 65% of Gauged. *Significant number of AMC III problems occur in both CN models. Both CN models had EOF runoff for all 7 seasons >> gauged; intuitive?
Comparisons of TESTS 1-4 Average Growing Season RUNOFF 1964-1970 Riverdale and Alda, ac-ft Avg G.S. Gauged Flow at Riverdale: Avg G.S. Gauged Flow at Alda: Test 1. Riverdale w/irrig: 2. Riverdale w.o./irrig: 3. Alda w/irrig: 4. Alda w.o./irrig: • 6,100 ac-ft • 8,912 ac-ft • 5,844 96% • 5,054 83% • 16,961 190% • 16,027 180% CN Model Results a.f.% Gauged (Edge-of-field) Observations: CN model w or w.o./irrig closely predicts average seasonal outflows at Riverdale, suggests EOF runoff at Alda is ~ 185% of gauged (many AMC III probs)