Snowmelt Runoff Modeling in the Upper Colorado River Basin using MODIS Fractional Snow Cover Christopher J. Crawford Oak Ridge Associated Universities Cryospheric Sciences Laboratory (Code 615) NASA/GSFC Dorothy K. Hall Cryospheric Sciences Laboratory (Code 615) NASA/GSFC Nicolo E. DiGirolamo Science Systems & Applications Inc. George A. Riggs Science Systems & Applications Inc. James L. Foster Emeritus, Hydrological Sciences Laboratory (Code 617) NASA/GSFC C.J. Crawford, MtnClim 2014, Midway, Utah 1. Problem Statement 2. Historical Context / Study Region 3. Study Region Climatology 4. Research Objective #1 5. Research Objective #2 6. MODIS Collection 6 Fractional Snow Cover 7. Snowmelt Runoff Model (SRM) Overview 8. SRM Results 9. Concluding Thoughts Photo: D.K. Hall / NASA C.J. Crawford, MtnClim 2014, Midway, Utah Northern Hemisphere Spring Snow Cover Reductions INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 27: 739–748 (2007) Published online 10 October 2006 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/joc.1426 Trends in snow ablation over North America GEOPHYSICAL RESEARCH LETTERS, VOL. 39, L19504, doi:10.1029/2012GL053387, 2012 Jamie L. Dyera * and Thomas L. Moteb a Department of Geosciences, Mississippi State University, P.O. Box 5448, Mississippi State, MS 39762-5448 b Department of Geography, University of Georgia Spring snow cover extent reductions in the 2008–2012 period exceeding climate model projections Abstract: C. Derksen1 and R. Brown2 A substantial decrease in snow cover extent (SCE) and snow depth over North America has been observed over the creates distinct challenges to the determination of 1960–2000 trends in period. One explanation for the changes in North American snow cover is a change in the frequency and/or intensityofof snow ablation. This study uses a gridded dataset of United States and Canadian surface observations from 1960 Arctic snow depth or snow water equivalent, analysis to examine patterns of snow ablation over North America. An ablation event is defined as an interdiurnal snow multiple datasets has shown that we now havetoa2000 good depth change exceeding a critical value. Results show a significant positive trend in the frequency of ablation events during understanding of the uncertainty in snow cover extent (SCE) (p < 0.05) and a significant negative trend in May (p < 0.05), indicating an earlier onset of ablation. This pattern time series across the Arctic, particularly in springMarch [Brown is consistent for ablation of varying intensity. Surface energy budget components and air mass frequencies are examined et al., 2010]. The longest available satellite-derived time in relation to the observed trends in snow ablation. Changes in March ablation frequency were shown to be dominated series of SCE are5,the weekly NOAA produced The Cryosphere, 219–229, 2011 snow charts by increases in the sensible heat flux. A higher frequency of dry moderate instead of moist polar air masses during high from manual analysis of primarily optical satellite imagery www.the-cryosphere.net/5/219/2011/ ablation years may explain the increase in sensible heat flux and ablation over the study period. Copyright 2006 Royal [Robinson et al., 1993]. This dataset is widely used within Meteorological Society doi:10.5194/tc-5-219-2011 Received 30 July 2012; revised 6 September 2012; accepted 9 September 2012; published 10 October 2012. [1] Analysis of Northern Hemisphere spring terrestrial snow cover extent (SCE) from the NOAA snow chart Climate Data Record (CDR) for the April to June period (when snow cover is mainly located over the Arctic) has revealed statistically significant reductions in May and June SCE. Successive records for the lowest June SCE have been set each year for Eurasia since 2008, and in 3 of the past 5 years for North America. The rate of loss of June snow cover extent between 1979 and 2011 (!17.8% decade!1) is greater than the loss of September sea ice extent the climate community and has been subject to detailed Author(s)in2011. CC Attribution 3.0 Wang License. (!10.6% decade!1) over the same period. Analysis of Cou- ©evaluation the spring period [e.g., et al., 2005; KEY WORDS snow; ablation; North America pled Model Intercomparison Project Phase 5 (CMIP5) model Brown et al., 2007, 2010; Frei and Lee, 2010]. Received series11 January 2006; Revised 14 August 2006; Accepted 20 August 2006 output shows the marked reductions in June SCE observed [4] Here we utilize the NOAA snow chart time since 2005 fall below the zone of model consensus defined by (1967–2012) to (1) report the latest trends in snow cover +/!1 standard deviation from the multi-model ensemble during the spring period when snow cover is confined INTRODUCTION 1999; Brown, 2000). Variations in snow depth over mean. Citation: Derksen, C., and R. Brown (2012), Spring snow largely to the Arctic, and temperature induced albedo feedthe same time period also exhibit decreasing trends, cover extent reductions in the 2008–2012 period exceeding climate backs are strongest across high latitudes [Groisman et al., High-latitude environments within North America have particularly throughout the central Canadian Prairies and model projections, Geophys. Res. Lett., 39, L19504, doi:10.1029/ 1994; Déry and Brown, 2007] and (2) place these trends in been shown to have high seasonal and interannual Great Plains region of the United States (Dyer and 2012GL053387. the context of recent CMIP5 climate model simulations. variability in snow cover during the winter season over Mote, 2006). Possible causes for the decrease in spring 1 2 R. D. Brown and D. A. Robinson the past decades (Groisman et al., 1994; Hughes and snow depth and snow extent in the latter part of the 2. Data Sets 1. Introduction 1 Climate Processes Section, Climate Research Division, Environment Canada @ Ouranos, 550 Sherbrooke West, include decreasing snowfall (i.e. from a Robinson, 1996; Frei and Robinson, 1999; Frei et al., 20th St. Century [5]Floor, We focus on theQC, recently NOAA1999). snow chart [2] Reliable information is needed on ongoing and future 19th Brown (2000) demonstrated that North American general decrease in precipitation, or from an increased Montréal, H3Areleased 1B9, Canada climate data record (CDR) [see Brown and Robinson, 2011] extent (SCE) during the fall and early winter changes in terrestrial snow cover for a wide range of geo- 2 Department snow cover of Geography, Rutgers University, 54 Global Joyce Kilmer Avenue, Piscataway, NJ 08854-8054, fraction USA of precipitation falling as rain, e.g. Akinremi physical applications, to advise policy and decision makers, maintained and housed at the Rutgers University (November–January) has increased since the early part and McGinn, 2000; Zhang et al., 2000; Groisman and Snow Lab. The dataset was acquired (from http://climate. and inform impact and adaptation activities. Of particular Received: 5 November 2010 – Published in The Cryosphere Discuss.: November 2010 from Frei Easterling, 1994), a shorter snow accumulation period of the 20th century. This 24 agrees with results importance is the timing of snow melt in spring, due to the rutgers.edu/snowcover/) following the update ofetcontinental al. (1999), who showed a general pattern of increased (Dye, 2002), or an increase in frequency and/or intensity Revised: 22 February 2011 – Accepted: 27 February 2011 – Published: 16 March 2011 climatological, hydrological, and ecosystem impacts of this monthly SCE through June 2012. The CDR combines the SCE in the fall over the same period; however, Frei et al. of winter thaws. Although there is evidence supporting original 190 km resolution NOAA snow charts (1967–1999) seasonal transition of the land surface [Callaghan et al., also showed a decreasing pattern of spring SCE, the relationship between changes in snowfall and snow Interactive 2011]. Contemporary variability and trends in Arctic snow [Robinson et al., 1993] with the 24 km resolution(1999) indicating a shift of1theIntroduction snow season to begin and end duration with decreasing spring snow depth and snow Abstract. An update is provided of Northern Hemisphere Multi-Sensor (IMS) snow product (1999-present) described cover are occurring in the context of amplified increases in Snow in Ramsay andApril) updated by cover Helfrich etearlier. al.(SCE) [2007]. spring[1998] (March, snow extent overdepth records, which provide additional extent, little attention has been paid to changes in observed high latitude temperature relative to other regions (NH) information beyond Reliable SCE, indicate decreasing SCE and Production of the CDRincorporating resulted in minor modifications [Serreze et al., 2009] with climate model scenarios showing the 1922–2010 period the new climate data frequency and/orinintensity of winter and spring information on spatial and the temporal variability to the IMS product, primarily by increasing the volume snow extent across much of North America in March and ablation. this warming is expected to continue [Vavrus et al., 2012]. record (CDR) version of the NOAA weekly SCE dataset, continental and hemispheric snow cover extent (SCE) is imin mountainous areas. The NOAA snow chart April, data record particularly in the central Canadian Prairies (Dyer [3] Monitoring Arctic snow cover is complicated by a lack with Snow melt is known annual 95% confidence intervals estimated from reportant for climate monitoring (e.g. Arndt et al., 2010), cli-to be influenced by the radiative SCE 2006). This augments the conclusion that of surface observations, and the high degree of spatial vari- has been found to agree well with other independent and Mote, balance (Groisman et al., 1994), and variations in air gression analysis and intercomparison of multiple datasets. mate model evaluation (e.g. Foster et al., 1996; Frei et al., ability relative to other snow covered regions [Liston, 2004] datasets during the Arctic spring melt period [Brown North et American snow cover is decreasing in extent and temperature (Karl etal., al., 1993; Brown and Goodison, uncertainty analysis indicates a 95% confidence interval 2010]. 2005; which Roesch, 2006; Brown and Frei, 2007) and cryospheredue to pronounced topographic and vegetative controls on The depth earlier in the spring, is of critical importance 1996), as well as surface energy fluxes (Leathers et al., Hemisphere spring snow cover [6] Northern spring SCE of ±5–10% overterrestrial the pre-satellite period wind-induced snow catchment and redistribution. In turn, in NH climate feedback studies (e.g. Hall Qu, 2006; Fernanbecause of the influence of the timing and magnitude of and 2004; Dyer and Mote, 2002; Kuusisto, 1986); therefore, extent (SCE) anomalies (excluding Greenland) for April, this heterogeneity introduces uncertainty into gridded snow and ±3–5% over the satellite era. The multi-dataset analdes et 2009;hydrologic Flanner etsystems. al., 2011).with Previously published spring snow melt runoff to al., regional a shallower winter snow cover, associated changes May, and June (when snow cover is confined largely to the datasets, including satellite-derived time series [i.e., Takala ysis shows larger uncertainties monitoring spring SCE overto satellite According snow cover data over North estimates of Northern Hemisphere (NH) monthly SCE used in surface albedo and thermal properties of the snowpack Arctic) were calculated for the 1967–2012 period to inveset al., 2011], conventional analyses [i.e., Brasnett, 1999], Eurasia (EUR) than North America (NA) due to the more America, SCE has been decreasing through the 1980s for evaluating climate models and monitoring variability and could lead to an increased sensitivity to any or all of and reanalysis products [i.e., Dee et al., 2011]. While this tigate trends and variability during the satellite era. July and The Cryosphere Northern Hemisphere spring snow cover variability and change over 1922–2010 including an assessment of uncertainty C.J. Crawford, MtnClim 2014, Midway, Utah Western North America Mountain Snowpack and Streamflow Declines 476 JOURNAL OF HYDROMETEOROLOGY VOLUME 6 Trends and Variability in Snowmelt Runoff in the Western United States GREGORY J. MCCABE U.S. Geological Survey, Denver, Colorado MARTYN P. CLARK University of Colorado, Boulder, Colorado (Manuscript received 2 September 2004, in final form 5 January 2005) ABSTRACT The timing of snowmelt runoff (SMR) for 84 rivers in the western United States is examined to understand the character of SMR variability and the climate processes that may be driving changes in SMR 1136 indicate that the timing of SMR for many JOU RNA F CLIM A T E States has shifted to timing. Results rivers inLtheOwestern United earlier in the snowmelt season. This shift occurred as a step change during the mid-1980s in conjunction with a step increase in spring and early-summer atmospheric pressures and temperatures over the western United States. The cause of the step change has not yet been determined. ! 1. Introduction ! ! VOLUME 18 Changes toward Earlier Streamflow Timing across Western NorthDettinger America summer temperatures (Aguado et al. 1992; and Cayan 1995; Regonda et al. 2005; Stewart et al. T. STEWART Snowmelt runoff (SMR) is an important source of 2004)IRIS and possibly global warming (Regonda et al. Scripps Institution Oceanography, Jolla, California water for much of the western United States (McCabe 2005; of Stewart et al. La 2004). and Wolock 1999; Stewart et al. 2004), and for snowAlthough a number of studies have identified a shift DANIEL R. CAYAN melt-dominated basins in the western United States to earlier SMR for many rivers in the western United Scripps Institution and U.S. Geological Survey, La Jolla, California spring/summer runoff can account for from 50% oftoOceanography, States, these studies have depended on trend analyses 80% of total annual runoff (Serreze et al. 1999; Stewart to identify these changes (Stewart et al. 2004). Trend MICHAEL D. DETTINGER et al. 2004). One of the expected hydrologic effects of analyses are unable to determine if a trend is gradual or U.S. Geological Survey, and Scripps Institution of Oceanography, La Jolla, California global warming is a shift in the timing of SMR to earlier a step change. Often the interpretation of trend analysis in the year (Gleick 1987; McCabe and Ayers 1988; is that the change identified is gradual. Since changes in (Manuscript received 2 January 2004, in final form 9 July 2004) Gleick and Adams 2000; Mote et al. 2005; Regonda et SMR timing have been identified by linear trends, there GEOPHYSICAL RESEARCH LETTERS, doi:10.1029/2007GL031022, ABSTRACT al. 2005). Consistent with this hypothesis a numberVOL. of 34, is aL16402, tendency to attribute these changes2007 to global warmClick in spring snowpack Herestudies have identified declines ing because of large correlations linear trends The highly variable timing of streamflow in snowmelt-dominated basins acrossbetween western North America is for consequence, and indicator, of climate Changes in the timing of snowmeltto important earlier SMR for many in SMR timingfluctuations. and the increasing trend in global temFull(Mote et al. 2005) and a shift an derived streamflow from 1948 to 2002 were investigated in a network of 302 western North America gauges Article rivers in the western United States (Aguado et al. of1992; often difficult to determine by examining the center mass forperature. flow, springIt pulse onsetisdates, and seasonal fractional the flowsphysical through trend and principal component Statisticaldriving analysis of the streamflow measures with Pacific Wahl 1992; Pupacko 1993; Dettinger and Cayan 1995;analyses. processes linear trends timing because so many variclimate indicators identified local and key large-scale processes that govern the regionally coherent parts of Rajagopalan and Lall 1995; Cayan et al. 2001; Regonda ables with linear trends will correlate highly with one the changes and their relative importance. Significance of trends toward earlier snowmelt runoff, Columbia and and regionally trends toward of springtime snowmelt and streamflow et al. 2005; Stewart et al. 2004).Widespread The observed shift coherent to another. To earlier betteronsets understand the climate processes have taken place across most of western North America, affecting an area that is much larger than previearlier SMR is especially noticeable for basins inUnited the driving Missouri Basin headwaters, western States the changes to fractions earlier of SMR timing in theearlier westously recognized. These timing changes have resulted in increasing annual flow occurring northwestern United States (Mote et al.year 2005; Regonda in the water by 1–4 weeks. The immediate (or proximal) for the spatially coherent parts of the ern United States itforcings is important to appropriately char1 1 2 year-to-year fluctuations longer-term trends of streamflow timing have been higher winter and spring Johnnie N.2005; Moore, Joel T. 2004). Harper, and Markalso C.and Greenwood et al. Stewart et al. These studies have acterize the changes as gradual trends or as step temperatures. Although these temperature changes are partly controlled by the decadal-scale Pacific cliindicated that revised the primary driving force the shiftpublished in changes mate mode [Pacific decadal oscillation (PDO)], a separate significant part2002). of the variance is associated (McCabe and Wolock Received 15 June 2007; 9 July 2007; accepted 17of July 2007; 21 August 2007. and with ainspringtime trend that spans the PDO phases. SMR timing has been increases spring warming and earlyThe objectives of this study are to 1) characterize the [1] We assess changes in runoff timing over the last 55 years [2005] found negative linear trends in several measures of changes in the timing of SMR, 2) identify when changes at 21 gages unaffected by human influences, in the spring runoff timing for the period 1948 – 2003. For examin SMR timing have occurred, and 3) determine what headwaters of the Columbia-Missouri Rivers. Linear ple, spring runoff in streams located in the headwaters of the climate processes are concurrent likely driving the observed little change in the annual 1. address: Introduction Corresponding author Dr. Gregory McCabe, USGS, Columbia regression models and tests for significance that control andbeen Missouri Rivers shifted earlier bytotal about 6 – dischanges in thecharge timing of SMR the western United (Wahl 1992; in Aguado et al. Stewart 1992; Dettinger Denverdiscoveries’’ Federal Center,ofMSmany 412, Denver, CO 80225. with a for ‘‘false tests, combined 19 days for the 55 year period of record. et al. and The loss of mountain snow accumulation and reducCayan 1995) and only spotty evidence for decreasing E-mail: gmccabe@usgs.gov States. and McCabe conceptual runoff response model, were used to examine the [2005] and Clark [2005] attributed their tions in snowmelt-derived water supply are among the April–June streamflow volumes (Wahl 1992). Climatic detailed structure of spring runoffconsequences timing. Weexpected concludefrom thatclimate calculated toward runoff primary warming.trends factors affectearlier the timing andmostly amounttooftemperadischarge in Thus it is with somesignificant concern that a shift toward earlier river only about one third of the gages exhibit trends ture increase ratherbasins, than introducing precipitation trends.spatial The and recent significant temposnowmelt-derived starting in the late Working © 2005 Meteorological Society with time butAmerican over half ofspring the gages tested showstreamflow, significant report from Group of as theregional International Panel ral variability as Iwell coherence (seeon Cayan 1940s, Therefore, has been observed several across et al.(IPCC) 2001). The regional coherence in the observed relationships with discharge. runoff for timing is regions Climate Change summarized these findings by western North America (Roos 1987, 1991; Wahl 1992; changes for snowmelt-dominated basins strongly sugmore significantly correlated with annual discharge than stating that in western North America earlier stream flows Aguado et al. 1992; Pupacko 1993; Dettinger and gests that climate changes and, in particular, temperawith time. This result differs from previous studies of Stewart runoff et imply peak water accumulation has shifted forward by Cayan 1995; Cayan et al. 2001; al. 2004; Re-snow ture, are playing a dominant role, as moderate temperain the JHM428 western USA that equate trends to a during aboutthetwo 1950 [Lemke al., 2007]. snowmeltSuch gonda et al.linear 2005).time Spring streamflow lastweeks ture since changes mostly affectetmidelevation five decades shifted so that the April–July response to global warming. Ourhasresults imply thatmajor changes in snowpack snowmelt runoff could have(i.e., dominated and basins most susceptible to melting streamflow peak arrives Rockies one or more weeksimplications ear- Mote 2003a). predicting future snowmelt runoff in thenow northern major for water resource management and lier, resulting in decliningcontrolling fractions of spring and early of Concurrent with the For observed snowmelt and streamwill require linking climate mechanisms sustainability river ecosystems. example, predictions river discharge. Importantly, as yet, there has flow timing trends, the average annual temperature precipitation, rather thansummer projecting response to simple of major changes in snowpack timing in the Sierra Nevada have increased 1°–2°C since the 1940s over the (north) linear increases in temperature. Citation: Moore, J. N., J. T. have led the California Department of Water Resources to western part of North America, especially during the C.J. Crawford, MtnClim 2014, Midway, Utah Western Wyoming Study Region / Wind River Range MODIS Terra January 18, 2014 Photo: A.E. Putnam Wind River Range Glaciers C.J. Crawford, MtnClim 2014, Midway, Utah Krimmel (2002) Historical Context for Snowmelt Runoff Modeling C.J. Crawford, MtnClim 2014, Midway, Utah C.J. Crawford, MtnClim 2014, Midway, Utah ~1216 sqkm2 ~259 sqkm2 C.J. Crawford, MtnClim 2014, Midway, Utah Research Objective #1: Evaluate historical trends in discharge timing and amount, and decompose discharge variability in the time domain for adjacent watersheds with and without diversions C.J. Crawford, MtnClim 2014, Midway, Utah C.J. Crawford, MtnClim 2014, Midway, Utah Pine Creek Watersheds Moore et al. (2007), GRL C.J. Crawford, MtnClim 2014, Midway, Utah Green River at Warren Bridge Watersheds Moore et al. (2007), GRL Interannual and Decadal Variability Pine Creek C.J. Crawford, MtnClim 2014, Midway, Utah Green River Ghil et al. 2002, Reviews of Geophysics Ault and St. George 2010, J. Climate C.J. Crawford, MtnClim 2014, Midway, Utah Research Objective #2: Model snowmelt runoff using meteorological observations, snow-water-equivalent measurements, and MODIS fractional snow cover during 2000-2012 MODIS Collection 6 Fractional Snow Cover (MOD10A1F) CGF Daily FSC Map Example http://modis-snow-ice.gsfc.nasa.gov/ Crawford (2014), Hydrological Processes C.J. Crawford, MtnClim 2014, Midway, Utah C.J. Crawford, MtnClim 2014, Midway, Utah Snowmelt Runoff Model (SRM) Overview / Inputs “conceptual semi-distributed hydrological model with physical basis” Martinec and Rango 1979, NASA Pub. 2216 Rango and Martinec 1995, WRB Martinec et al. 2007, SRM Users Man. C.J. Crawford, MtnClim 2014, Midway, Utah SRM Structure Q = average daily discharge [m3s-1] c = runoff coefficient expressing the losses as a ratio (runoff precipitation), with cS referring to snowmelt and cR to rain a = degree-day factor [cm oC-1d-1] indicating the snowmelt depth resulting from 1 degree-day T = number of degree-days [oC d] ΔT = lapse rate adjustment when extrapolating temperature to the average hypsometric elevation of the basin or zone [oC d] S = ratio of the snow covered area to the total area P = precipitation contributing to runoff [cm]. A temperature threshold, TCRIT, determines whether this contribution is rainfall and immediate. If precipitation is determined by TCRIT to be new snow, it is kept on storage A = area of the basin or zone [km2] k = recession coefficient indicating the decline of discharge in a period without snowmelt or rainfall: k = Qm+1/Qm (m, m + 1 are the sequence of days during a true recession flow period). n = sequence of days in discharge computation period Qn+1 = [CSn * an (Tn + ΔTn) Sn + CRn * Pn] A * 10000/86400 * (1-kn+1) + Qn kn+1 C.J. Crawford, MtnClim 2014, Midway, Utah Green River Pine Creek SRM Simulations C.J. Crawford, MtnClim 2014, Midway, Utah Green River Pine Creek SRM Simulations C.J. Crawford, MtnClim 2014, Midway, Utah Green River Pine Creek SRM Simulations Green River Pine Creek SRM Simulations C.J. Crawford, MtnClim 2014, Midway, Utah Interannual SRM Performance C.J. Crawford, MtnClim 2014, Midway, Utah C.J. Crawford, MtnClim 2014, Midway, Utah Concluding Thoughts 1. The trend towards earlier snowmelt runoff is modest along northern periphery watersheds of the Upper Colorado River Basin 2. Below average discharge is evident across percentiles and watersheds since the late 1980s 3. SRM can skillfully simulate snowmelt runoff at daily and seasonal timescales, although diversions and spatial aggregation underpin model accuracy The Way Forward C.J. Crawford, MtnClim 2014, Midway, Utah 1. Transform SRM into an operational environment and move towards an ensemble approach 2. Identify ways propagate uncertainty through model inputs and parameterization