PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, NATURAL HAZARD SCIENCE ...... Stroudsburg, PA: Hutchinson Ross. Jordan, A. C., & Kean ...
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Oxfor d Resear ch Encyclopedia of N at ur al H azar d Science Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s Vijay Gupt a Subject : Risk Assessm ent , Floods Online Publicat ion Dat e: Sep 2017 DOI : 10.1093/acr efor e/9780199389407.013.301
Su m m ar y an d K eyw o r d s Pr edict ion of floods at locat ions w her e no st r eam flow dat a exist is a global issue because m ost of t he count r ies involved don’t have adequat e st r eam flow r ecor ds. The U nit ed St at es Geological Sur vey developed t he r egional flood fr equency (RFF) analysis t o pr edict annual peak flow quant iles, for exam ple, t he 100-year flood, in ungauged basins. RFF equat ions ar e pur e st at ist ical char act er izat ions t hat use hist or ical st r eam flow r ecor ds and t he concept of “ hom ogeneous r egions.” To supplem ent t he accur acy of flood quant ile est im at es due t o lim it ed r ecor d lengt hs, a physical solut ion is r equir ed. I t is fur t her r einfor ced by t he need t o pr edict pot ent ial im pact s of a changing hydr o-clim at e syst em on flood fr equencies. A nonlinear geophysical t heor y of floods, or a scaling t heor y for shor t , focused on r iver basins and abandoned t he “ hom ogeneous r egions” concept in or der t o incor por at e flood pr oducing physical pr ocesses. Self-sim ilar it y in channel net w or k s plays a foundat ional r ole in under st anding t he obser ved scaling, or pow er law r elat ions, bet w een peak flow s and dr ainage ar eas. Scaling t heor y of floods offer s a unified fr am ew or k t o pr edict floods in r ainfall-r unoff (RF-RO) event s and in annual peak flow quant iles in ungauged basins.
Page 1 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s Theor et ical r esear ch in t he cour se of t im e clar ified sever al k ey ideas: (1) t o under st and scaling in annual peak flow quant iles in t er m s of physical pr ocesses, it w as necessar y t o consider scaling in individual RF-RO event s; (2) a unique par t it ioning of a dr ainage basin int o hillslopes and channel link s is necessar y; (3) a cont inuit y equat ion in t er m s of link st or age and dischar ge w as developed for a link-hillslope pair (t o com plet e t he m at hem at ical specificat ion, anot her equat ion for a channel link involving st or age and dischar ge can be w r it t en t hat gives t he cont inuit y equat ion in t er m s of dischar ge); (4) t he self-sim ilar it y in channel net w or k s plays a pivot al r ole in solving t he cont inuit y equat ion, w hich pr oduces scaling in peak flow s as dr ainage ar ea goes t o infinit y (scaling is an em er gent pr oper t y t hat w as show n t o hold for an idealized case st udy); (5) a t heor y of hydr aulic-geom et r y in channel net w or k s is sum m ar ized; and (6) highlight s of a t heor y of biological diver sit y in r ipar ian veget at ion along a net w or k ar e given. The fir st obser vat ional st udy in t he Goodw in Cr eek Exper im ent al Wat er shed, M ississippi, discover ed t hat t he scaling slopes and int er cept s var y fr om one RF-RO event t o t he next . Subsequent ly, diagnost ic st udies of t his var iabilit y show ed t hat it is a r eflect ion of var iabilit y in t he flood-pr oducing m echanism s. I t has led t o developing a m odel t hat link s t he scaling in RF-RO event s w it h t he annual peak flow quant iles feat ur ed her e. Rainfall-r unoff m odels in engineer ing pr act ice use a var iet y of t echniques t o calibr at e t heir par am et er s using obser ved st r eam flow hydr ogr aphs. I n ungagged basins, st r eam flow dat a ar e not available, and in a changing clim at e, t he r eliabilit y of hist or ic dat a becom es quest ionable, so calibr at ion of par am et er s is not a viable opt ion. Recent pr ogr ess on developing a suit able t heor et ical fr am ew or k t o t est RF-RO m odel par am et er izat ions w it hout calibr at ion is br iefly r eview ed. Cont r ibut ions t o gener alizing t he scaling t heor y of floods t o m edium and lar ge r iver basins spanning differ ent clim at es ar e r eview ed. Tw o st udies t hat have focused on under st anding floods at t he scale of t he ent ir e planet Ear t h ar e cit ed. Finally, t w o case st udies on t he innovat ive applicat ions of t he scaling fr am ew or k t o pr act ical hydr ologic engineer ing pr oblem s ar e highlight ed. They include r eal-t im e flood for ecast ing and t he effect of spat ially dist r ibut ed sm all dam s in a r iver net w or k on r ealt im e flood for ecast ing. Keyw or ds: pow er law s, m ass conser vat ion quest ion in a net w or k , self-sim ilar it y of net w or k s, hillslope-link syst em s
I n t r o d u c t i o n t o t h e Sc al i n g T h eo r y o f Fl o o d s Flood is a geophysical phenom enon. Floods ar e also a nat ur al hazar d because t hey can r esult in m ajor loss of life and subst ant ial dam age t o pr oper t ies and civil infr ast r uct ur e. Accur at e est im at es of t he m agnit ude and fr equency of flood flow s ar e needed for t he design of br idges and r oads, w at er -use and w at er -cont r ol pr oject s, and for floodplain Page 2 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s definit ion and m anagem ent (Daw dy, Gr iffis, & Gupt a, 2012 ). The U nit ed St at es Geological Sur vey (U SGS) developed t he r egional flood fr equency (RFF) equat ions t o est im at e quant iles (e.g., t he 100-year flood) for annual peak st r eam dischar ges, or floods, in ungauged basins, w her e no st r eam flow dat a exist . RFF equat ions ar e pur e st at ist ical char act er izat ions t hat use hist or ical st r eam flow r ecor ds t o car r y out r egr essions. They do not include t he physical pr ocesses t hat pr oduce floods in r ainfall-r unoff (RF-RO) event s (Fur ey, Tr out m an, Gupt a, & Kr ajew sk i, 2016 ). A geophysical under st anding of annual flood quant iles is a long-st anding unsolved pr oblem . The scaling t heor y of floods explained her e developed a physical fr am ew or k t o m ak e flood pr edict ions in ungauged basins. To supplem ent t he accur acy of flood quant ile est im at es due t o lim it ed r ecor d lengt hs— t ypically less t han 30 year s—a physical solut ion t o t he est im at ion of flood quant iles at annual t im e scale is r equir ed. The RFF analysis assum es a st at ionar y hydr o-clim at e, w hich m eans t hat t he fut ur e w ould be st at ist ically sim ilar t o t he past . Gr eenhouse gases and ot her fir st -or der hum an influences ar e changing t he hydr o-clim at e syst em (Pielk e et al., 2009 ). Consequent ly, fut ur e flood fr equencies ar e not expect ed t o be st at ist ically sim ilar t o t he past (Daw dy, 2007 ). A solut ion t o t his pr oblem r equir es t hat annual flood quant iles be under st ood in t er m s of physical m echanism s pr oducing floods. M andelbr ot ’s book ( 1982 ) on t he fr act al geom et r y of nat ur e offer ed self-sim ilar it y as a new scient ific par adigm . Scaling ar ises in self-sim ilar syst em s. A lar ge num ber of st udies had obser ved scaling, or pow er law s, in annual flood quant iles ver sus dr ainage ar eas in r iver basins (L insley, Kohler, & Paulhus, 1982 ). Gupt a and Daw dy ( 1995 ) syst em at ically invest igat ed t he pr esence of pow er law s bet w een flood quant iles and dr ainage ar eas using t he RFF analyses of t he U SGS in t hr ee st at es. A geophysical under st anding of how slopes and int er cept s can be pr edict ed fr om physical pr ocesses r equir ed under st anding t he scaling in RF-RO event s in r iver basins and led t o Gupt a, Cast r o, and Over ( 1996 ). I t abandoned t he concept of “ hom ogeneous r egions” and ident ified t he fundam ent al r ole of self-sim ilar it y in channel net w or k t opology and geom et r y. Ogden and Daw dy ( 2003 ) obser ved scaling r elat ions in peak dischar ges for 226 RF-RO event s t hat spanned hour ly t o daily t im e scales in t he 21 k m 2 Goodw in Cr eek Exper im ent al Wat er shed (GCEW) in M ississippi. They found t hat scaling slopes and int er cept s var y fr om one event t o anot her. Fur ey and Gupt a ( 2007 ) conduct ed a diagnost ic analysis t o under st and t he var iabilit y in t er m s of physical pr ocesses gener at ing floods. Tw o over view paper s (Gupt a, Tr out m an, & Daw dy, 2007 ; Gupt a & Waym ir e,
1998 )
and
subsequent r esear ch have ident ified m any ar eas of science and engineer ing t his ar t icle addr esses, for exam ple, hydr ologic and hydr aulic engineer ing, hydr ologic science, nonlinear geophysics, fr act al geom et r y, fluvial geom or phology, applied pr obabilit y, fluid m echanics, biology, and landscape ecology.
Page 3 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
A B r i ef H i st o r y o f t h e Sc al i n g T h eo r y o f Fl o o d s Kir k by ( 1976 ) and L ee and Delleur ( 1976 ) fir st dem onst r at ed t he unique r ole of channel net w or k w idt h funct ion in est ablishing a connect ion bet w een net w or k geom et r y and st r eam flow hydr ogr aphs. For any given dist ance fr om t he out let , t he w idt h funct ion is defined as t he num ber of st r eam s at t hat dist ance. I f r ainfall is deposit ed inst ant aneously and unifor m ly over a net w or k and t r avels at a const ant velocit y t o t he out let , t hen t he flow hydr ogr aph at t he out let has t he sam e shape as t he w idt h funct ion. The w idt h funct ion is const r uct ed fr om t he link-m agnit ude classificat ion of a net w or k t hat Shr eve ( 1966 , 1967 ) int r oduced. I t is called geom or phologic inst ant aneous unit hydr ogr aph (GI U H ). Subsequent ly, sever al paper s cont r ibut ed t o t his line of t hink ing; for exam ple, Tr out m an and Kar linger ( 1984 , 1998 ), and Gupt a and M esa ( 1988 ). Rodr iguez-I t ur be and Valdes ( 1979 ) and Gupt a, Waym ir e, and Wang ( 1980 ) pr oposed anot her ver sion of a GI U H t hat w as const r uct ed fr om H or t on-St r ahler or der ing and H or t on law s. A subst ant ial lit er at ur e developed in t his line of r esear ch (Rodr iguez-I t ur be & Rinaldo, 1997 ). The above body of published r esear ch clear ly dem onst r at ed t hat channel net w or k t opology and geom et r y have a fundam ent al r ole t o play in t he fut ur e developm ent s of r iver basin hydr ology. M andelbr ot ( 1982 ) published a book on t he fr act al geom et r y of nat ur e, and a union session of t he Am er ican Geophysical U nion w as or ganized ar ound t he book in t he fall m eet ing in 1982. Fr act al geom et r y w as based on t he hypot hesis of self-sim ilar it y, a new for m of invar iance under a change of scale. I t s im pact becam e evident t hr ough m any published paper s and book s w it h applicat ions t o physics, geophysics, geology, and hydr ology (Feder, 1988 ;
Rodr iguez-I t ur be & Rinaldo, 1997 ; Tur cot t , 1997 ).
Scaling, or a pow er law, ar ises in self-sim ilar syst em s, w hich offer ed a new scient ific par adigm t o a lar ge num ber of st udies t hat show ed scaling in annual floods in r iver basins (L insley et al., 1982 , p. 370). Gupt a and Daw dy ( 1995 ) syst em at ically invest igat ed t he pr esence of pow er law s bet w een flood quant iles and dr ainage ar eas using t he RFF analyses of t he U SGS. H ow ever, a physical basis of scaling in RFF r equir ed under st anding scaling r elat ions in individual RF-RO event s, because floods ar e gener at ed at t he scales of event s. I t w as a m ajor shift in focus fr om annual t im e scale t o event scales of hour s and days, and fr om hom ogeneous r egions t o r iver basins. Gupt a et al. ( 1996 ) t ook t he fir st st ep t o develop a physical basis of scaling in peak flow s for RF-RO event s in a Peano basin, an idealized self-sim ilar channel net w or k . I t w as analyt ically show n t hat t he peak flow s exhibit a pow er law r elat ionship w it h r espect t o dr ainage ar ea w it h a flood-scaling exponent , . The par am et er is a fr act al dim ension of t he spat ial r egions of a Peano basin t hat cont r ibut e t o peak flow s at successively lar ger dr ainage ar eas. Gupt a and Waym ir e ( 1998 ) int r oduced an equat ion of cont inuit y for a channel net w or k in t he fir st r eview paper on t he t opic. Tr out m an and Over ( 2001 ) gener alized t he r esult s on scaling exponent s of peak flow s t o a gener al class of net w or k s gener at ed by a det er m inist ic self-sim ilar algor it hm , such as Tok unaga net w or k s. Page 4 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s Ot her t heor et ical paper s w er e subsequent ly published on t he t opic (M enabde & Sivapalan, 2001 ). I n or der t o m ak e fur t her pr ogr ess, it becam e appar ent t hat a car eful obser vat ional st udy on scaling in peak flow s for RF-RO event s w as needed. Ogden and Daw dy ( 2003 ) obser ved scaling r elat ions in peak dischar ges for 226 individual RF-RO event s t hat spanned hour ly t o daily t im e scales in t he 21 k m 2 GCEW, in M ississippi. They discover ed t hat t he scaling slopes and int er cept s var y fr om one event t o anot her. The m ean of 226 event slopes (0.82) w as close t o t he com m on slope (0.77) of m ean annual and 20-year r et ur n peak dischar ge quant iles. Fur ey and Gupt a ( 2007 ) car r ied out a diagnost ic analysis of Ogden and Daw dy ( 2003 ) and found t hat event -t o-event var iabilit y in t he scaling slopes and int er cept s is due t o a com binat ion of t em por al r ainfall var iabilit y, spat ial var iabilit y in infilt r at ion, and dischar ge velocit y on a hillslope and in a channel link . Gupt a et al. ( 2007 ) w r ot e a r eview paper on t he scaling t heor y of floods in w hich k ey developm ent s up t o 2006 w er e r eview ed. Scaling t heor y connect s sever al disconnect ed fields for t he pur pose of under st anding floods and solving r elat ed pr oblem s. The r esear ch is fir m ly r oot ed in t he peer -r eview ed lit er at ur e. I t r epr esent s a new par adigm in hydr ologic science and engineer ing. I n t he r em ainder of t his sect ion, foundat ional and not able advances as w ell as cur r ent per spect ives ar e discussed. I n 1956 David Daw dy w as assigned by t he U SGS as chief assist ant t o M anuel Benson in developing t he U SGS RFF Analysis Pr ogr am . Aft er t hat he w as assigned t o develop t he U SGS physically based RF-RO m odel for ext ending RFF r ecor ds t o lar ger scales t han individual st at es and st at ions. As st at e RFF r epor t s began t o appear he cont inued his int er est in RFF analysis. Daw dy obt ained t he base dat a for each of t hese st at e r epor t s and becam e convinced t her e w as m or e infor m at ion cont ained in t he dat a t han just t hose r epor t ed in t he st at e r epor t s, but he could not figur e out how t o use t he dat a on a lar ger scale. I t w as his feeling t hat infor m at ion w as being over look ed, w hich led him t o consult w it h Vijay Gupt a, in t he lat e 1980s. H e found out t hat Gupt a and Edw ar d Waym ir e w er e collabor at ing on fr act als and scaling in hydr ology. Daw dy felt t hat scaling m ight help in m ak ing pr ogr ess. Gupt a and Daw dy m et on sever al occasions in t he ear ly 1990s and t ook t he fir st st eps. Subsequent ly, a fr am ew or k for solving t he pr oblem slow ly developed (Daw dy, 2016 ). Br ent Tr out m an played a subst ant ive r ole t hr ough published r esear ch on t he t opic and co-aut hor ed t he over view paper (Gupt a et al., 2007 ). H e also collabor at ed w it h Gupt a’s gr aduat e st udent s on scaling in floods. H e w as a m at hem at ical st at ist ician in t he U SGS N at ional Resear ch Pr ogr am fr om 1975 unt il his r et ir em ent 2011. Tr out m an’s r esear ch addr essed t he applicat ion of st at ist ical m et hods, pr obabilit y, and st ochast ic pr ocesses t o t he pr edict ion and m odeling of sur face-w at er flow s, par t icular ly quest ions of scale and spat ial/t em por al var iabilit y of pr ocesses influencing flow s (Tr out m an, 2016 ). Waym ir e and Gupt a collabor at ed in under st anding t he m at hem at ics of fr act als and scaling in t he 1980s and 1990s w it h applicat ions t o hydr ology. Waym ir e co-aut hor ed t he fir st over view paper on scaling in floods (Gupt a & Waym ir e, 1998 ) and w r ot e a k ey paper on channel net w or k s (Bur d, Waym ir e, & Winn, 2000 ; Waym ir e, 2016 ).
Page 5 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
D yn am i c Fo r m u l at i o n s f o r R ai n f al l -R u n o f f Even t s i n R i ver B asi n s L et
,
,
be a space-t im e field r epr esent ing r iver dischar ge, or volum e of flow per
unit t im e, t hr ough t he dow nst r eam end of link e in a channel net w or k . L et denot e t he r unoff int ensit y int o link e fr om t he t w o adjacent hillslopes, and let denot e t he com bined ar eas of t he t w o hillslopes. m ay also include r unoff deplet ion per unit t im e due t o channel infilt r at ion or evapor at ion fr om channel sur face in link e. Finally, let denot e st or age in link e at t im e t . Then t he equat ion of cont inuit y for t he syst em of hillslopes link s can be w r it t en as (Gupt a et al., 2007 ): (1)
The t er m s
and
on t he r ight -hand side r epr esent dischar ges fr om t he t w o
upst r eam link s joining link e.
denot es t he r unoff fr om t he t w o hillslopes dr aining int o
link e. Solut ions of Eq. ( 1 ) depend on t he br anching and geom et r ic st r uct ur e of a channel net w or k t hr ough t hese t w o t er m s. Equat ion ( 1 ) as w r it t en is based on an assum pt ion t hat t he channel net w or k is binar y. I t m ay easily be gener alized, how ever, t o non-binar y net w or k s, or t hose for w hich m or e t han t w o upst r eam link s can flow int o a junct ion, for exam ple, a Peano net w or k . H ow ever, t his sit uat ion does not ar ise in act ual r iver net w or k s. Equat ion ( 1 ) assum es t hat no loops ar e pr esent in t he net w or k . A specificat ion of r equir es a hillslope-scale m odel for r epr esent ing r unoff gener at ion fr om a ver y lar ge num ber of hillslopes in a net w or k . For exam ple, GCEW has 544 hillslopes. I t pr esent s a gr eat scient ific challenge because m ost of t he focus on m odel r unoff gener at ion has been in a single hillslope (Cor r adini, Govindar aju, & M or bidelli, 2002 ;
Duffy, 1996 ; Fr eeze, 1980 ). Such appr oaches, called “ bot t om -up,” use a com binat ion of
w ell-k now n unsat ur at ed-sat ur at ed flow cont inuum equat ions. By cont r ast , Fur ey, Gupt a, and Tr out m an ( 2013 ) developed and t est ed a “ t op-dow n” m odel in GCEW, w hich is needed t o m odel r unoff gener at ion
fr om a lar ge num ber of hillslopes in Eq. ( 1 ).
A funct ional r elat ionship bet w een link st or age
and link dischar ge
is needed t o
expr ess Eq. ( 1 ) in t er m s of one dependent var iable. I t can be obt ained fr om t he definit ion of st or age and dischar ge and a specificat ion of t he link velocit y, fr om a m om ent um balance equat ion gover ning t hr ee-dim ensional velocit y field in an open channel at t he sub-link scale. Kean and Sm it h ( 2005 ) m ade pr ogr ess t ow ar d solving t his pr oblem in t he cont ext of pr edict ing a t heor et ical r elat ionship bet w een flow dept h and dischar ge k now n as a r at ing cur ve. They developed a fluid-m echanical m odel t hat r esolves boundar y r oughness elem ent s fr om field m easur em ent s over a nat ur al channel r each and calculat ed t he cr oss-sect ional aver age velocit y t o pr edict t heor et ical r at ing cur ve. Their r esear ch est ablishes a scient ific fr am ew or k t o specify for a link . Jor dan and Kean ( 2010 ) applied t he m odel t o est im at e r at ing cur ves at t en ungagged st r eam r eaches,
Page 6 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s r anging fr om t hir d t o sixt h or der channels, in t he Whit ew at er basin in Kansas (M ant illa & Gupt a, 2005 ). L et ,
denot e a link lengt h,
t he m ean link w idt h, and
t he m ean link dept h.
Then, (2)
I n view of t he hydr aulic-geom et r ic (H -G) r elat ions for a net w or k (Gupt a & M esa, 2014 ; L eopold & M iller, 1956 ), w idt h, dept h, and velocit y m ay be expr essed as funct ions of dischar ge, or
, w her e f(.) is an ar bit r ar y funct ion. Subst it ut ing t he st or age-
dischar ge r elat ion int o Eq. ( 1 ) gives (3)
w her e
The funct ions
and
in Eq. ( 3 ) var y fr om one hillslope-link pair t o t he ot her. Even
sm all basins lik e GCEW cont ain a lar ge num ber of hillslope-link pair s. A specificat ion of t he physical par am et es in all of t hese pair s is r equir ed for solving Eq. ( 3 ). Gupt a ( 2004A ) st at ed t hat “ st at ist ical scaling is an em er gent pr oper t y of a com plex syst em w hich a-pr ior i is not built int o t he physical equat ions.” Scaling can be used t o t est physical hypot heses (Gupt a, 2004B ). Once a lar ge num ber of par am et er s ar e specified, Eq. ( 3 ) can be solved it er at ively t o obt ain r unoff hydr ogr aph, , , , at t he bot t om of ever y link , and t her efor e in all sub-basins of a r iver basin. I t s solut ions pr oduce st r eam flow hydr ogr aphs at all junct ions in a channel net w or k . They have been used t o obt ain r esult s on spat ial scaling st at ist ics of floods under idealized physical condit ions in idealized channel net w or k s, nam ely M andelbr ot -Vicsek (M enabde & Sivapalan, 2001 ).
Sel f -Si m i l ar R i ver N et w o r k s Gupt a et al. ( 2007 ) r eview ed 20 year s of pr ogr ess on self-sim ilar it y in r eal channel net w or k s t hat included det ails of t he H or t on-St r ahler or der ing and t he H or t on law s. A br ief over view of som e of t he k ey t opics and pr ogr ess since t hen is given her e.
H o r t o n R el at i o n sh i p s f o r R i ver N et w o r k s H or t on-St r ahler or der ing is defined as follow s. A channel w it h no upst r eam inflow s is given or der one. When t w o st r eam segm ent s of t he sam e or der m er ge, t he or der of t he out -flow ing st r eam incr eases by one. I n case t w o m er ging st r eam s have differ ent or der s, t he out -flow ing st r eam has t he higher of t he t w o or der s. The t hr ee m ost w idely st udied H or t on law s ar e for t he num ber of St r ahler st r eam s of or der , , t he aver age lengt h of st r eam s of or der , , and t he aver age upst r eam ar ea of st r eam s of or der , . I t has Page 7 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s been obser ved t hat t he r at ios of t hese quant it ies in successive or der s t end t o be independent of or der, t hat is, , , and , w hich ar e k now n as t he classical H or t on law s. ar e called H or t on r at ios. Self-sim ilar channel net w or k m odels have show n t hat H or t on law s hold in t he lim it as net w or k or der incr eases t o infinit y (M cConnell & Gupt a, 2008 ; Veit zer & Gupt a,
2000 ).
Peck ham and Gupt a ( 1999 ) gener alized t he classical H or t on law s for m ean dr ainage ar eas and m ean channel lengt hs t o full pr obabilit y dist r ibut ions. Specifically, t hey gave obser vat ional and som e t heor et ical ar gum ent s for t he w ell-k now n r andom m odel (Shr eve, 1967 )
t o show t hat pr obabilit y dist r ibut ions of var iables r escaled by t heir m eans
collapse int o a com m on pr obabilit y dist r ibut ion of som e r andom var iable, , t hat is independent of or der. I t is w r it t en as (4)
w her e
is t he H or t on r at io, and
denot es equalit y in pr obabilit y dist r ibut ions of r andom
var iables on bot h sides. The r elat ion in Eq. ( 4 ) is k now n as a gener alized H or t on law . Thus, not only t he m ean value, but any finit e m om ent , if it exist s, obeys H or t on law s. As an exam ple, Figur e 1 show s t he classical and t he gener alized H or t on law s for dr ainage ar eas
in t he Whit ew at er basin in Kansas (M ant illa & Gupt a, 2005 ).
Click t o view lar ger Figur e 1. (L eft ) H or t on law of m ean ar eas. (Right ) Pr obabilit y dist r ibut ions of r escaled ar eas in t he Whit ew at er Basin, KS. (Fr om M ant illa & Gupt a.,
2005 .)
To k u n ag a Sel f -Si m i l ar M o d el Tok unaga ( 1966 ) int r oduced a det er m inist ic m odel of r iver net w or k s t hat w as based on H or t on-St r ahler or der ing r at her t han on link m agnit ude as is t he case for t he r andom t opology m odel. I t w as char act er ized by t he m ean num ber of side t r ibut ar ies. For a st r eam of or der
, let
denot e aver age num ber of side t r ibut ar ies of or der k , k now n as
gener at or s (Tok unaga, 1966 , 1978 ). Gener at or s ar e self-sim ilar if t hey obey t he const r aint , . I t m eans t hat t he gener at or s only depend on t he differ ence independent of . U nder t he addit ional const r aint t hat have t he for m St r ahler or der,
is a const ant , gener at or s
(Tok unaga, 1966 ). The num ber of st r eam s of differ ent H or t onobey a r ecur sion equat ion (Tok unaga,
1978 ),
Page 8 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s (5)
M cConnell and Gupt a ( 2008 ) r igor ously pr oved t he H or t on law of st r eam num ber s as an asym pt ot ic r esult fr om t he solut ion of Eq. ( 5 ) and ext ended it t o a H or t on law for m agnit ude
. Tok unaga also show ed t hat t he gener at or s of t he r andom t opology m odel
(Shr eve, 1967 ) exhibit Tok unaga self-sim ilar it y w it h par am et er s Eq. ( 5 ), w it h t hese values of par am et er s, pr edict s
. The solut ion of
, w hich is t he sam e t hat Shr eve
( 1967 ) obt ained for t he r andom m odel. Gupt a et al. ( 2007 ) r eview ed ot her k ey r esult s for Tok unaga net w or k s.
R an d o m Sel f -Si m i l ar N et w o r k M o d el The geom or phologic and hydr ologic t hink ing w er e gr eat ly influenced by t he r andom m odel of channel net w or k based in a link-m agnit ude classificat ion (Shr eve, 1966 , 1967 ). Jar vis and Woldenber g ( 1984 ) r epr oduced t he k ey paper s fr om 1945 t o 1976 in a book . The paper s w er e divided int o four par t s, and each par t included edit or ial over view, w hich m ak es it an excellent r efer ence. The fluvial geom or phologic r esear ch t ook a new t ur n dur ing t he 1990s. Det ailed em pir ical analysis of lar ge r iver basins w as gr eat ly aided by t he availabilit y of finer esolut ion digit al elevat ion m odels (DEM s). Obser vat ions fr om lar ge basins show ed t hat r andom m odel pr edict ions deviat e subst ant ially fr om em pir ical values (Peck ham , 1995 ). Bur d et al. ( 2000 ) pr oved t hat t he r andom m odel is t he only one am ong finit e, binar y Galt on-Wat son st ochast ic br anching t r ees t hat exhibit s t he m ean self-sim ilar t opology of Tok unaga net w or k s. These findings opened t he door t o develop a new class of st at ist ical channel net w or k m odels called r andom self-sim ilar net w or k s (RSN s), w hich do not belong t o t he class of binar y Galt on-Wat son st ochast ic br anching pr ocesses (Veit zer & Gupt a, 2000 ).
RSN s ar e const r uct ed r ecur sively fr om r andom gener at or s, w hich ar e essent ially
sim ple or der -2 net w or k s w it h a r andom num ber of int er ior nodes. The for m ulat ion allow s for differ ent gener at or pr obabilit y dist r ibut ions for t he r eplacem ent of int er ior and ext er ior link s. Veit zer and Gupt a ( 2000 ) show ed t hat t he gener alized H or t on law, or dist r ibut ional sim ple scaling, holds for t he num ber of link s per st r eam , m agnit ude, and st r eam num ber s as net w or k or der incr eases. The m ean side t r ibut ar y st r uct ur e of a subset of RSN s w er e show n t o obey Tok unaga self-sim ilar it y.
Pr o g r ess o n R an d o m Sel f -Si m i l ar N et w o r k M o d el Tr out m an ( 2005 ) developed an algor it hm t hat allow s a unique ext r act ion of gener at or s fr om act ual net w or k s under cer t ain r est r ict ions. H e gave r esult s show ing t hat a geom et r ic dist r ibut ion for t he num ber of side t r ibut ar ies in bot h int er ior and ext er ior gener at or s fit r easonably w ell t he dat a for t he Flint River basin in Geor gia. M ant illa, Tr out m an, and Gupt a ( 2010 ) t est ed t he RSN m odel in 30 basins fr om diver se clim at ic and geogr aphic set t ings in t he U nit ed St at es. Self-sim ilar it y in a st at ist ical sense w as found t o
Page 9 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s hold in 26 of t he 30 basins, and geom et r ic dist r ibut ion show ed good agr eem ent w it h dat a in all cases. An im por t ant consider at ion in t he analysis of scaling pr oper t ies of r iver basins and t he channel net w or k s t hat dr ain t hem is spat ial em bedding, w hich is also necessar y for sim ulat ing spat ially dist r ibut ed r ainfall int ensit y fields on r iver basins. The idealized basins, Peano and M andelbr ot -Vicsek , ar e spat ially em bedded, w hich enabled t o consider a spat ial dist r ibut ion of r ainfall int ensit ies on t hese basins (Gupt a et al., 1996 ; M enabde & Sivapalan, 2001 ). For net w or k s gener at ed by t he r andom t opology m odel, det er m inist ic self-sim ilar net w or k s lik e Tok unaga, RSN s, t he br anching st r uct ur e is defined but how t he net w or k s fit int o a t w o- or a t hr ee-dim ensional space is not . M ant illa, Gupt a, and Tr out m an ( 2012 ) pr esent ed an algor it hm for em bedding RSN s in a given spat ial r egion.
N u m er i c al St u d i es o n R an d o m Sel f -Si m i l ar Ch an n el N et w o r k M o d el s w i t h A p p l i c at i o n s t o Fl o o d s M ant illa, Gupt a, and M esa ( 2006 ) sim ulat ed peak flow s in t he Walnut Gulch basin, Ar izona, and analyzed t he scaling pr oper t ies of flow hydr ogr aphs by solving t he m ass conser vat ion show n in Eq. ( 3 ) under a const ant flow velocit y assum pt ion. They obser ved t hat t he scaling exponent s for peak flow s ar e lar ger t han t he m axim a of t he w idt h funct ions (Veit zer & Gupt a, 2001 ). This pr oper t y for a r eal net w or k cont r adict s t he pr evious findings for idealized self-sim ilar net w or k s; for exam ple, M andelbr ot -Vicsek , M ant illa, Gupt a, and Tr out m an ( 2011 ) t est ed t his hypot hesis in sim ulat ed RSN s using geom et r ic dist r ibut ions, w it h par am et er s
and
cor r esponding t o int er ior and ext er ior gener at or s, r espect ively.
They com par ed t he num er ical r esult s w it h t he analyt ic expr essions t hat Tr out m an ( 2005 ) had obt ained, w hich ser ved as benchm ar k for t he accur acy of r esult s fr om num er ical sim ulat ions. Their r esult s suppor t ed t he finding of M ant illa et al. ( 2006 ).
A n al yt i c al R esu l t s f o r H yd r au l i c -Geo m et r y i n To k u n ag a Sel f -Si m i l ar Ch an n el N et w o r k s an d I m p l i c at i o n s f o r Sp ec i es R i c h n ess i n R i p ar i an Veg et at i o n I n a classic paper, L eopold and M iller ( 1956 ) discover ed H or t on law s for H -G var iables in dr ainage net w or k s, t her eby ext ending t he H or t on law s fr om t opological and geom et r ic var iables t o H -G var iables (st r eam dischar ge Q, w idt h W, dept h D, velocit y U , slope S, and Page 10 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s M anning’s fr ict ion n’). M ot ivat ed by t he need t o t heor et ically under st and H or t on law s for t he H -G var iables, a long-st anding unsolved pr oblem , Gupt a and M esa ( 2014 ) developed an analyt ical t heor y in self-sim ilar Tok unaga net w or k s. The H -G dat a set s in channel net w or k s fr om t hr ee published st udies and one unpublished st udy w er e used t o t est t heor et ical pr edict ions. Gupt a and M esa ( 2014 ) used dim ensional analysis and t he Buck ingham Pi t heor em t o ident ify six independent dim ensionless r iver -basin num ber s. A m ass conser vat ion equat ion in t er m s of H or t on bifur cat ion and dischar ge r at ios in Tok unaga net w or k s w as der ived. U nder t he assum pt ion t hat t he H -G var iables ar e hom ogeneous and self-sim ilar funct ions of st r eam dischar ge, it w as show n t hat t he funct ions ar e of a w idely assum ed pow er law for m . Asym pt ot ic self-sim ilar it y of t he fir st k ind, or SS-1 (Bar enblat t , 1996 ), w as applied t o pr edict t he H or t on law s for W, D, and U as asym pt ot ic r elat ionships. Pr edict ions of t he exponent s agr eed w it h t hose pr eviously pr edict ed for t he opt im al channel net w or k (OCN ) m odel. Pr edict ed exponent s of w idt h and t he Reynolds num ber w er e t est ed against t hr ee field dat a set s. Ashley basin show ed deviat ions fr om t heor et ical pr edict ions. Test s of ot her pr edict ed exponent s suggest ed t hat H -G in net w or k s does not obey SS-1. I t fails because one of t he dim ensionless r iver -basin num ber s, slope goes t o 0 as net w or k or der incr eases, but it cannot be elim inat ed fr om t he asym pt ot ic lim it . Ther efor e, a gener alizat ion of SS-1, based on t he asym pt ot ic selfsim ilar it y of t he second k ind, or SS-2 (Bar enblat t , 1996 ), w as consider ed. I t int r oduced t w o anom alous scaling exponent s as fr ee par am et er s, w hich enabled t hem t o show t he exist ence of H or t on law s for W, D, U , S, and n’. The t w o anom alous scaling exponent s w er e not pr edict ed. I nst ead, t hey used t he obser ved exponent s of D and S t o pr edict t he exponent of n’ and t o t est it against exponent s fr om t hr ee field st udies m ent ioned above. The Ashley basin show ed som e deviat ion fr om t heor et ical pr edict ions. A physical r eason for t his deviat ion w as given, w hich ident ified an im por t ant t opic for r esear ch. Finally, how t he t w o anom alous scaling exponent s could be est im at ed fr om t he t r anspor t of suspended sedim ent load and t he bed load w as br iefly sk et ched. St at ist ical var iabilit y in t he H or t on law s for t he H -G var iables w as also discussed. Bot h ar e im por t ant open pr oblem s for fut ur e r esear ch.
H yd r o -ec o l o g i c al T h eo r y o f I n t er m i t t en t R i p ar i an D i ver si t y o n St r eam N et w o r k s Dunn, M ilne, M ant illa, and Gupt a ( 2011 ) gener alized t he scaling fr am ew or k t o include ecological var iables coupled t o w at er. Based on digit al land cover m aps, t hey w er e able t o discer n scaling and H or t on law s in t he ar eas occupied by r ipar ian veget at ion in t he Whit ew at er basin. M ilne and Gupt a ( 2017A ) ar e developing a new line of r esear ch t o under st and biological diver sit y and w at er balance in self-sim ilar Tok unaga net w or k s and t est t he developm ent s in t he Whit ew at er basin. The idea or iginat ed t o under st and deviat ion in t he Ashley basin fr om t heor et ical pr edict ion m ent ioned in t he pr evious sect ion. I t r equir es t hat t he Tok unaga gener at or be m odified t o incor por at e int er m it t ency in r unoff gener at ion and t he gr ow t h of r ipar ian veget at ion on hillslopes, w hich is evident fr om t he digit al m aps of r ipar ian veget at ion. Consequent ly, t he r igor ous m at hem at ical Page 11 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s r esult s for Tok unaga net w or k s (M cConnell & Gupt a, 2008 ) ar e used t o der ive H or t on law s for species r ichness. A “ law of hydr o-ecology” is post ulat ed t hat t r eat s w at er balance t hr oughout t he net w or k as a consequence of self-sim ilar non-local int er act ions. I n a com panion paper M ilne and Gupt a ( 2017B ) developed t he ent r opy-H or t on fr am ew or k and dem onst r at ed how it infor m s quest ions of biodiver sit y, r esilience t o per t ur bat ions in w at er supply, changes in pot ent ial evapot r anspir at ion, and land use changes t hat m ove ecosyst em s aw ay fr om opt im al ent r opy w it h concom it ant loss of pr oduct ivit y and biodiver sit y.
Ph ysi c al B asi s o f A n n u al Fl o o d Qu an t i l es i n R i ver B asi n s Building on Ogden and Daw dy (O-D), Fur ey et al. ( 2016 ) conduct ed t he fir st r igor ous analysis t ow ar d a physical under st anding of annual flood quant iles in GCEW. They st at ed t w o hypot heses based on t he r esult s in O-D: (1) scaling slopes of annual peak (AP) quant iles ar e t he sam e for all r et ur n per iods and (2) t he m ean scaling slope of st r eam dischar ge peak s fr om RF-RO event s equals t he m ean slope of AP quant iles. H er e, a “ m ean” r efer s t o an aver age over exceedance pr obabilit ies. H ypot hesis 1 is a for m al st at em ent of sim ple scaling. I t st ands in cont r ast t o m ult iscaling w her e scaling slopes depend on r et ur n per iods (Gupt a, M esa, & Daw dy, 1994 ). H ypot hesis 2 per t ains t o only t he m ean slopes bot h for event s and AP quant iles. Ther efor e, a r eject ion of hypot hesis 1 does not im ply a r eject ion of hypot hesis 2. Result s in O-D suppor t hypot hesis (1), w hile suppor t for hypot hesis (2) is unclear w it hout fur t her analysis. Given t hat hypot hesis (1) holds, t he m ean slope of AP quant iles in hypot hesis (2) w ill coincide w it h t he com m on AP quant ile slope in hypot hesis (1). L et denot e t he com m on slope of AP quant iles and denot e t he m ean scaling slope for RF-RO event s. Result s in O-D pr ovide t he est im at es
and
given in Table 1 . To t est
hypot hesis (2), Fur ey et al. ( 2016 ) evaluat ed t he st at ist ical significance of t he differ ence, . They found t hat t he est im at ed st andar d er r or (SE) of t he differ ence obeys SE > 0.039, yielding a p-value > 0.15. I t indicat es t hat
is not significant at t he 5%
level. Thus, O-D r esult s do not r eject H ypot hesis (2). Result s in O-D r epr esent only one dat aset and r equir ed som e appr oxim at ions in t he above calculat ions due t o lack of infor m at ion. Thus, t o fur t her t est t he t w o hypot heses, Fur ey et al. ( 2016 ) assem bled a second dat aset for 148 RF-RO event s in GCEW. Exam ining dat a fr om a single basin is necessar y t o obt ain a physical under st anding of scaling in dischar ge quant iles. As one goes t o lar ger basins, r ainfall est im at ion r equir es r adar -r ainfall dat a t hat int r oduce r em ot e-sensing er r or s not found in t he r ain gauge dat a in GCEW. A st andar d linear r egr ession m odel w as used t o evaluat e scaling slopes for RF-RO event s
Page 12 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s and for AP quant iles. Figur e 1 show s t he scaling slopes for 148 event s. The m ean and st andar d deviat ion of slopes, 0.78 and 0.16, is sim ilar t o t he cor r esponding values in O-D (Figur e 3 ). A Weibull plot t ing posit ion for m ula w as used t o assign pr obabilit ies t o AP values. For each gauge, t he quant iles w it h em pir ical pr obabilit ies closest t o sever al t ar get pr obabilit ies w er e ident ified. The sm allest pr obabilit y evaluat ed, p = 0.071, r epr esent s a r et ur n per iod of 14 year s, one year beyond t he r ecor d lengt h. The ot her value w as p = 0.5. The r esult s in Figur e 2 show t w o differ ent scaling slopes, suggest ing t he possibilit y of m ult iscaling (Gupt a et al., 1994 ). I t m eans t hat slopes of AP quant ile scaling r elat ions ar e differ ent for differ ent r et ur n per iods. I f such a pr oper t y holds, t he m ean of t he slopes for a given set of r et ur n per iods can be denot ed by . Then hypot hesis 2 is an equalit y bet w een
and μb. Table 1 show s t hat
and
. These values denot e t he m ean
slope of t he quant ile r elat ionships pr esent ed in Figur e 2 and t he m ean slopes of event s given in Figur e 1 . Assum ing t hat hypot hesis 1 holds, t he significance of t he differ ence, , can be assessed as done ear lier for t he O-D r esult s. A p-value of ≥ 0.83 w as obt ained, indicat ing t hat t he differ ence is not significant at t he 5% level, again suppor t ing hypot hesis 2. The sam e r esult s ar e found if quant ile r elat ionships for ot her r et ur n per iods ar e consider ed. I n sum m ar y, r esult s fr om t he second dat a set suppor t hypot hesis 2. To t est for sim ple ver sus m ult iscaling in AP quant iles, indicat ed in Figur e 2 , dat a fr om GCEW t hat include addit ional year s beyond 1994 ar e needed.
Click t o view lar ger Figur e 2. Var iabilit y in scaling slopes for RF-RO event s (Fur ey et al., 2016 ). (Repr int ed w it h per m ission fr om ASCE.)
Page 13 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Click t o view lar ger Figur e 3. Evidence of m ult iscaling in annual peak quant iles (Fur ey et al., 2016 ). (Repr int ed w it h per m ission fr om ASCE.)
Table 1. M ean Scaling Slopes of RF-RO Event s and Annual Peak Quant iles
Sour ce: Fur ey et al. ( 2016 ). The above r esult s clear ly dem onst r at e t hat a connect ion exist s bet w een scaling in event s and in AP quant iles. Fur ey et al. ( 2016 ) char act er ized such a connect ion w it hin a single physical-st at ist ical m odel called a nest ed m ixed-effect s linear (N M EL ) m odel. The N M EL m odel char act er izes event -t o-event var iabilit y in scaling r elat ionships bet w een st r eam dischar ge peak s and dr ainage ar eas. The N M EL m odel leads t o scaling r elat ionships for dischar ge peak quant iles and annual peak (AP) quant iles t hat ar e t he basis for RFF equat ions. Since event -based scaling r elat ionships can be connect ed t o physical pr ocesses, it im plies t hat annual peak quant iles can be connect ed as w ell. To sum m ar ize, t est r esult s of m odel assum pt ions w er e suppor t ive of N M EL m odel st r uct ur e, t hough som e discr epancies w er e found. While t he m odel link s event s t o quant iles, r esult s show t hat t her e ar e im por t ant differ ences bet w een quant ile-based scaling st at ist ics using APs and r elat ed event -based st at ist ics. I n par t icular, quant ile scaling allow s for event m ixing, using dat a fr om differ ent event s t o det er m ine a scaling r elat ionship and pr ovides slope values having a significant ly nar r ow er r ange t han t hat found w it h event s. The lat t er r esult m eans t hat AP-quant ile scaling r elat ionships and RFF r elat ions could under est im at e flooding pot ent ial. Given t hat flood char act er izat ion acr oss a r iver basin should be t ied t o physical pr ocesses and condit ions t hat include pr e-event Page 14 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s soil m oist ur e, infilt r at ion gover ning r unoff gener at ion fr om each hillslope in a basin, and space-t im e var iable hydr aulic-geom et r y gover ning r unoff dynam ics in a net w or k . The findings in Fur ey et al. ( 2016 ) show t hat m et hods for flood char act er izat ion based on event scaling ar e m or e infor m at ive and r obust t han t r adit ional RFF m et hods based on analysis of AP quant iles. A lar ger sam ple st udy fr om GCEW and ot her w at er sheds is needed t o fur t her t est , and possibly r efine, t hese ideas. Test of m ult iscaling is an im por t ant issue t hat needs t o be addr essed in fut ur e r esear ch.
H o r t o n L aw s f o r D i ag n o si n g a R ai n f al l -R u n o f f M o d el W i t h o u t Cal i b r at i o n o f I t s Par am et er s The m easur em ent and collect ion of hydr ologic dat a in channel net w or k s is t edious and expensive. Gupt a and M esa ( 2014 ) illust r at ed t w o field st udies as exam ples (I bbit t , M cKer cher, & Duncan, 1998 ; M cKer cher, I bbit t , Br ow n, & Duncan, 1998 ). As a r esult , t he discover y of t he H or t on law s for hydr ologic var iables has gr eat ly lagged behind geom or phology. The hydr ologic dat a in Gupt a, Ayalew, M ant illa, and Kr ajew sk i ( 2015 ) cam e fr om t he 32,400 k m 2 I ow a River basin locat ed in east er n I ow a befor e it joins t he M ississippi River (Figur e 4 ). The basin is cont inuously m onit or ed by 34 U SGS gauging st at ions. The t op four annual m axim um peak dischar ges obser ved at t he basin out let over t he past 112 year s occur r ed in t he pr evious 7 year s (Sm it h, Baeck , Villar ini, Wr ight , & Kr ajew sk i, 2013 ). Figur e 4 show s t he I ow a River basin and t he locat ion of t he U SGS gauging sit es. A St r ahler or der w as assigned t o each channel link w her e t he U SGS st r eam gauges ar e locat ed, and t he dr ainage ar eas dr aining int o t he gauges w er e est im at ed. The obser ved peak flow s w er e also assigned t he sam e or der. This infor m at ion w as used t o t est t he H or t on law s for dr ainage ar eas and for peak flow s. Ayalew, Kr ajew sk i, and M ant illa ( 2015 ) select ed RF-RO event s t hat occur r ed in t he 12-year per iod fr om 2002 t o 2013. They obser ved a scale-invar iant peak dischar ge w hen t he ent ir e basin got a r unoff-gener at ing r ainfall event at som e point dur ing a t im e w indow equivalent t o t he basin’s concent r at ion t im e (defined as t he t im e r equir ed for a par cel of w at er t o t r avel fr om t he m ost r em ot e hillslope in t he basin t o t he out let ). The concent r at ion t im e is about 15 days for t he I ow a River basin (see Ayalew et al., 2015 , for a det ailed discussion of t he RF-RO event select ion and for t he cr it er ia used t o define peak dischar ge event s). Based on t hese st r ict cr it er ia, 51 RF-RO event s w er e ident ified w hose r esult ing peak dischar ges exhibit scaling w it h r espect t o dr ainage ar ea. Figur e 5 show s a RF-RO event dur ing w hich t he ent ir e basin did not r eceive r ainfall at som e point dur ing t he 15-day t r avel t im e w indow. Figur e 6 illust r at es a RF-RO event dur ing w hich t he ent ir e basin r eceived r ainfall at som e point dur ing t he 15-day t r avel t im e w indow.
Page 15 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Click t o view lar ger Figur e 4. The I ow a River basin and t he locat ion of t he U SGS gauging sit es. St r eam s below or der four ar e not show n for t he sak e of clar it y. (Repr oduced fr om Gupt a et al., of AI P Publishing.)
2015 ,
w it h per m ission
Click t o view lar ger Figur e 5. Exam ple st r eam flow t im e ser ies fr om r epr esent at ive U SGS gauging sit es in t he basin (t op panels) and t he associat ed peak-dischar ge scaling plot (bot t om panel) for t he case w her e only a por t ion of t he basin got r ainfall at som e point dur ing t he 15day t r avel t im e w indow. The st r eam flow t im e ser ies is nor m alized by t he annual m axim um flow for each gauging sit e. (Repr oduced w it h per m ission fr om Ayalew et al., 2015 .)
Page 16 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Click t o view lar ger Figur e 6. Exam ple st r eam flow t im e ser ies fr om r epr esent at ive U SGS gauging sit es in t he basin (t op panels) and t he associat ed peak-dischar ge scaling plot (bot t om panel) for t he case w her e t he ent ir e basin exper ienced r ainfall at som e point dur ing t he 15-day t r avel t im e w indow. The st r eam flow t im e ser ies is nor m alized by t he annual m axim um flow for each gauging sit e. (Repr oduced fr om Gupt a et al., per m ission of AI P Publishing.)
2015 ,
w it h t he
A Co n si st en t T h eo r et i c al Fr am ew o r k t o Est i m at e H o r t o n R at i o s Fur ey and Tr out m an ( 2008 ) developed a consist ent st at ist ical fr am ew or k t hat r esolved im por t ant pr oblem s w it h t he int er pr et at ion and use of t r adit ional H or t on r egr ession st at ist ics. Their appr oach agr ees w it h dist r ibut ional sim ple scaling or t he gener alized H or t on law illust r at ed in Figur e 1 . I t is used in t he dat a analysis of peak flow s. To under st and t he k ey concept s, let be a r andom var iable t hat r epr esent s a geom or phologic pr oper t y such as dr ainage ar ea for basins of or der ω. Assum e t hat (6)
H er e, a and b ar e const ant s and
is a zer o-m ean r andom var iable. Equat ion ( 6 ) is a linear
r egr ession m odel. I t uses individual sam ple values r at her t han t he sam ple m ean t hat is t he st andar d pr act ice in t he classical H or t on analysis. Fur ey and Tr out m an ( 2008 ) t ook t he expect at ion on bot h sides in Eq. ( 6 ) t o get t he expr ession
, and
(7)
Subst it ut ing
w her e
is t he geom et r ic m ean, int o Eq. ( 7 ) gives
Page 17 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s (8)
Equat ion ( 8 ) gives t he expr ession for t he H or t on r at io,
, as
(9)
Equat ion ( 9 ) is based on geom et r ic r at her t han ar it hm et ic m ean and ser ves as an alt er nat ive expr ession t o est im at e H or t on r at io. Gupt a et al. ( 2015 ) est im at ed t he H or t on ar ea r at io,
, using dr ainage ar ea infor m at ion for
471,101 com plet e or der st r eam s using CU EN CAS (M ant illa & Gupt a, 2005 ). Tw o independent m et hods w er e used t o est im at e
. The fir st m et hod uses t he r egr ession Eq.
( 6 ) and est im at es t he slope, , w hich gives
. The second m et hod uses Eq. ( 7 )
and calculat es geom et r ic m eans for differ ent or der st r eam s. for or der s 7, 8, and 9 w as ignor ed due t o sm all sam ple sizes. The r em aining est im at es of show ed st at ist ical var iabilit y. To com par e t hese est im at es w it h t he value fr om m et hod 1, t he m ean of est im at es for or der s 2, 3, 4, 5, and 6 w as com put ed, w hich gave 4.66. The r esult s show ed t hat t he infor m at ion fr om t he 34 st r eam flow -gauging st at ions can r easonably pr edict t he value t hat is t he sam e as t hat est im at ed using det ailed dr ainage net w or k infor m at ion. I t pr ovided confidence in t he est im at ion of H or t on r at ios fr om t he 34 gauging st at ions and also gave a t em plat e t o est im at e H or t on r at io for peak flow s, w her e sam ple size is a m or e ser ious issue t han for dr ainage ar eas.
Cl assi c al H o r t o n L aw f o r Peak Fl o w s i n R ai n f al l -R u n o f f Even t s Consider a r iver basin w it h st r eam gauges. L et denot e a gauge and it s subbasin, and let i= 1,2, … denot e a RF-RO event . Dat a show t hat peak dischar ges at t he event t im e scale obey a scaling r elat ionship fir st obser ved in GCEW (Ogden & Daw dy, 2003 )
and lat er in t he m edium -sized I ow a River basin (Ayalew et al., 2015 ; Gupt a, M ant illa,
Tr out m an, Daw dy, & Kr ajew sk i, 2010 ). An or dinar y linear r egr ession (OL R) m odel can char act er ize t his r elat ionship (Fur ey et al., 2016 ): (10)
H er e, Q i,j denot es peak dischar ge at gauge j for event i, a(i) and b(i) ar e OL R coefficient s, A j is t he dr ainage ar ea dr aining gauge j, and is a m ean-zer o r andom var iable t hat r epr esent s t he deviat ion of peak dischar ge at gauge j fr om t he aver age linear r elat ion, a(i) + b(i)A j . Random ness in peak flow s is ent ir ely due t o t he er r or t er m . Figur e 7 show s peak dischar ge for four RF-RO event s as exam ples fr om t he 51 event s t hat Ayalew et al. ( 2015 ) analyzed. All of t he t er m s in Eq. ( 10 ) change fr om one RF-RO event t o anot her due t o changes in t he physical pr ocesses t hat pr oduce peak flow s (Ayalew, Kr ajew sk i, & M ant illa, 2014A ; Ayalew et al., 2015 ; Ayalew, Kr ajew sk i, M ant illa, & Sm all, 2014B ; Fur ey & Gupt a, 2005 , 2007 ). I t can be seen t hat t he R 2 values ar e high, w hich suppor t s t he use of OL R t o det ect scaling in peak flow dat a. I t is also evident t hat t he scat t er of peak dischar ges ar ound t he r egr ession line changes fr om event t o anot her event , w hich can be
Page 18 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s quant ified by t he st andar d deviat ion of t he er r or t er m in Eq. ( 10 ). Physically, it is a m anifest at ion of t he nat ur al var iabilit y of r ainfall and ot her cat chm ent physical var iables t hat cont r ol t he gener at ion of peak dischar ges in space and t im e. To t est t he validit y of t he classical H or t on law for peak flow s, four differ ent event s show n in Figur e 7 w er e select ed. Thr ee differ ent m et hods w er e used t o t est t he st at ist ical sensit ivit y of t he est im at es of t he H or t on r at io for peak flow s,
, i= 1,2,3 … Since t he
U SGS gauges ar e not locat ed at t he end of com plet e St r ahler st r eam s, w hich is necessar y for t he H or t on analysis, it w as assum ed t hat each U SGS st r eam gauge of a given or der is locat ed at t he end of a com plet e St r ahler st r eam of t hat or der. The obser ved peak flow s at t hese gauges for a given RF-RO event const it ut e a r andom sam ple of peak flow s for t hat or der. The r esult of t his assum pt ion pr oduces peak flow s for differ ent St r ahler st r eam s as show n in Figur e 8 w it h gr ey cir cles. The H or t on r at io, RQ (i), for i = 1,2,3,4 w as calculat ed for t he sam e four event s show n in Figur e 7 . I n m et hod 1, Eq. ( 9 ) w as applied t o com put e RQ as t he r at io of t he geom et r ic m eans of t he peak flow s at consecut ive or der s. The r esult s ar e show n in Figur e 8 . I n m et hod 2, peak flow dat a w as ar r anged accor ding t o t he St r ahler or der, and OL R, given in Eq. ( 7 ), w as used t o com put e r elat ion
. I n m et hod 3, t he
der ived in Gupt a et al. ( 2015 ) w as used. The appear ance of t he classical
H or t on law s for peak flow s in RF-RO event s show n in Figur e 8 w as a new hydr ologic discover y. Thr ee est im at ion m et hods t o addr ess t he sm all sam ple size issue for t he peak flow dat a need fur t her invest igat ion and const it ut es an im por t ant ar ea of fut ur e r esear ch.
Click t o view lar ger Figur e 7. Obser ved scaling r elat ions of peak dischar ges w it h dr ainage ar eas in t he I ow a River basin, I ow a. The influence of var iabilit y in t he est im at es of slope b(i) and t he int er cept r eflect s changes in t he physical pr ocesses gener at ing peak dischar ges. (Repr oduced fr om Gupt a et al., per m ission of AI P Publishing.)
2015 ,
w it h t he
Page 19 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Click t o view lar ger Figur e 8. The classical H or t on law s for t he four peak-dischar ge event s show n in Figur e 7 . The gr ey cir cles show t he obser ved peak dischar ges, w her eas t he black line show s t he est im at ed E[ ln(Q ω )] using Eq. ( 10 ). (Repr oduced fr om Gupt a et al., of AI P Publishing.)
2015 ,
w it h per m ission
Gen er al i zed H o r t o n L aw f o r Peak Fl o w s i n R ai n f al l -R u n o f f Even t s Do peak dischar ges at t he event t im e scale obey a gener alized H or t on law ? Gupt a et al. ( 2015 ) invest igat ed t his issue and der ived (11)
Equat ion ( 11 ) denot es t hat peak flow s divided by it s m ean for a RF-RO event , i = 1,2,3, …, and St r ahler or der, . I t look s lik e a gener alized H or t on law in so far as t he r escaled r andom var iable on t he r ight is independent of t he St r ahler or der, . I t w as explained in t he cont ext of geom or phology (Eq. ( 4 )), but t he pr obabilit y dist r ibut ion of changes w it h each event , . This issue does not ar ise for geom or phologic r andom var iables lik e dr ainage ar eas. I n t his sense, t he feat ur e in Eq. ( 11 ) is new and is pur ely hydr ologic in nat ur e. Gupt a et al. ( 2015 ) hypot hesized t hat t he pr obabilit y dist r ibut ion of
is
com m on for all t he event s and t est ed t heir hypot hesis using dat a. The cum ulat ive dist r ibut ion funct ions (CDF) for t he nor m alized peak dischar ges for each of t he 51 event s w as plot t ed, as show n by t he t hin lines in Figur e 8 . Tw o aut hor s have descr ibed k-sam ple t est s: t he Kolm ogor ov–Sm ir nov (KS) (Conover, 1999 ) and a based on lik elihood st at ist ics (Zhang & Wu,
2007 ).
t est
The condit ion t o per for m a k-sam ple KS
t est applies, since each point in a CDF cor r esponds t o a st r eam gauge, w hich doesn’t change w it h event s. H ow ever, t he t est is m or e pow er ful t han t he KS, and t her efor e it w as adopt ed. The r esult s indicat e t hat t her e is insufficient evidence t o r eject t he hypot hesis t hat
, w hich is t he CDFs show n in Figur e 8 by t hin lines, com e fr om a
com m on dist r ibut ion t hat is independent of event s, , as w ell as of t he St r ahler st r eam or der, . This is a new hydr ologic discover y w it h im por t ant fut ur e applicat ions w it h r espect t o flood pr edict ion in gauged and ungauged basins (Gupt a et al., 2015 ).
One applicat ion is br iefly explained next .
Page 20 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Click t o view lar ger Figur e 9. CDF plot for t he nor m alized peak dischar ges fr om each of t he 51 RF-RO event s. The bold line r epr esent s t he com m on CDF aft er com bining t he individual CDFs show n in t hin gr ey lines. (Repr oduced fr om Gupt a et al., of AI P Publishing.)
2015 ,
w it h per m ission
Page 21 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
A p p l i c at i o n s o f H o r t o n L aw s f o r Test i n g a R ai n f al l -R u n o f f M o d el W i t h o u t Cal i b r at i o n o f Par am et er s Consider a st at ist ical-m echanical syst em for dr aw ing a for m al analogy w it h physical pr ocesses in a r iver net w or k . I n a st at ist ical-m echanical syst em , t her e ar e t w o scales: m icr oscopic, w it h a ver y lar ge num ber of degr ees of fr eedom (N DOF) t hat cannot be m easur ed, and m acr osopic, w it h a few degr ees of fr eedom t hat can be m easur ed. A w ellk now n exam ple is ideal gas law for a syst em in equilibr ium , w hich has only four par am et er s: pr essur e, num ber of m olecules per unit volum e, t em per at ur e, and Bolt zm an const ant (Reif, 1965 ). The br anch of st at ist ical-m echanics der ives m acr oscopic law s fr om assum pt ions about m olecular dynam ics at t he m icr oscopic scale. The hydr ology pr oblem under discussion r epr esent s a sim ilar sit uat ion, but it is far m or e com plex t han an ideal gas. I n or der t o pr esent t he for m al analogy bet w een RF-RO pr ocesses in a r iver basin and t he afor em ent ioned st at ist ical-m echanical syst em , Gupt a ( 2004A , 2004B ) pr ovides a cont ext . Solut ions of t he coupled or dinar y differ ent ial equat ions r epr esent ing m ass and m om ent um conser vat ion given in Eq. ( 3 ) r equir e specificat ion of dynam ic par am et er s descr ibing r unoff gener at ion in hillslopes and w at er flow dynam ics in channel link s. These par am et er s var y spat ially due t o differ ences in geom et r ic, hydr aulic, and biophysical pr oper t ies am ong individual channel link s and hillslopes, w hich incr ease w it h t he ar ea (size) of a dr ainage basin. Ther efor e, t he num ber of differ ent values t hat t he physical par am et er s can t ak e also incr eases as a basin becom es lar ger. Gupt a ( 2004A ) called it N DOF in or der t o under scor e t he idea t hat t he hydr ological com plexit y in a r iver basin can be com par ed t o t he com plexit y of a st at ist ical-m echanical syst em t hat has a ver y lar ge N DOF. To illust r at e t hat t he N DOF in t he RF-RO syst em is ver y lar ge, Gupt a ( 2004A , 2004B ) consider ed an idealized buck et -t ype r epr esent at ion of r unoff gener at ion and t r anspor t pr ocesses for a single hillslope-link syst em . Four physical par am et er s ar e r equir ed t o specify t he r unoff gener at ion for each hillslope and t hr ee par am et er s gover ning t r anspor t dynam ics in a link . These dynam ic par am et er s var y fr om one hillslope-link pair t o anot her. Typical hillslope has an ar ea of t he or der of 0.05 k m 2 . Ther efor e, a 1 k m 2 basin can be par t it ioned int o 1/0.05 = 20 hillslopes and 10 link s, because each link is dr ained by t w o adjacent hillslopes. The num ber of differ ent spat ial values of t he seven dynam ic par am et er s is 20 · 4 + 10 · 3 = 110. This sim ple calculat ion leads t o t he for m ula N DOF = 110·A. I t show s t hat t he N DOF incr eases linear ly w it h t he ar ea of a basin, A. For t he I ow a River basin, N DOF ~ 3.5 m illion. The flood pr oblem is far m or e com plex t han a st at ist ical-m echanical syst em , w hich is t ypically in equilibr ium . But t he RF-RO syst em is an open syst em t hat is not even in a st eady st at e. Appear ance of a lar ge num ber of physical par am et er s at t he hillslope-link scale r aises t w o issues. Fir st is t he need t o develop a t heor et ical appr oach for specifying a lar ge num ber of dynam ic par am et er s (see Fur ey et al.,
2013 ),
but m or e w or k is needed on t his t opic. The
issue of diagnosis of a physical RF-RO m odel is consider ed her e, because it is fundam ent al t o flood pr edict ion in ungauged basins. A for m al analogy bet w een Page 22 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s m icr oscopic and m acr oscopic scales in a hydr ologic syst em allow s us t o illust r at e t he st eps r equir ed t o diagnose dynam ic par am et er izat ions in a hydr ologic m odel. Fir st , for a given RF-RO event , est im at e r ainfall int ensit y field in space and t im e. Select a set of dynam ic par am et er s and r un t he hydr ologic m odel t hat is based on t he decom posit ion of a r iver basin int o hillslope and channel-link unit s (M ant illa, Gupt a, & Tr out m an, 2011 ). Sim ulat e a dischar ge hydr ogr aph at each channel junct ion in a net w or k , w hich gives sim ulat ed peak flow s in t he com plet e St r ahler st r eam s of or der ω = 1, 2, 3 … . Then est im at e
using t he t hr ee m et hods as explained in Gupt a et al. ( 2015 ). The
est im at e of scaling slope for t his event is given as
. Sim ilar ly, t he sim ulat ed
peak flow dat a at each com plet e or der St r ahler st r eam can be used t o com put e t he m ean peak flow. The sam e dat a set can be used t o com put e t he r escaled dist r ibut ion r epr esent ing t he gener alized H or t on law as show n in Figur e 9 for t he 51 event s. These t w o H or t on st at ist ics pr ovide t he m acr oscopic law s for diagnosing par am et er izat ions in a RF-RO m odel. The final st ep can be r epeat ed for differ ent event s as needed. The H or t on st at ist ics can also ser ve as a diagnost ic t ool t o under st and changes in peak flow s over annual and longer t im e scales in differ ent clim at es.
Page 23 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Cl i m at e Var i ab i l i t y an d t h e Sc al i n g T h eo r y o f Fl o o d s i n M ed i u m t o L ar g e B asi n s o f t h e Wo r l d H ow can t he scaling t heor y of floods be gener alized t o global basins spanning differ ent clim at es at t he scale of t he cont inent s and t he ent ir e planet Ear t h? Poveda et al. ( 2007 ) t ook an im por t ant fir st st ep in t his dir ect ion. Their st udy involved conduct ing annual w at er balances over sever al dr ainage basins for t he ent ir e count r y of Colom bia, w hich has w idely var ying clim at es r anging fr om ext r em ely hum id t o sem i-ar id and ar id. Annual st r eam flow s w er e pr edict ed fr om int er polat ed fields of annual pr ecipit at ion and evapot r anspir at ion using gr ound-based and r em ot ely sensed dat a, finding r easonable agr eem ent w it h obser ved st r eam flow s. Am ong m any alt er nat e m odels t o est im at e evapot r anspir at ion, t he best est im at es w er e given by t he w ell-k now n Budyk o equat ion (Budyk o, 1974 ). Poveda et al. ( 2007 ) show ed t hat pow er law s descr ibe t he r elat ionship bet w een annual flood quant iles and dr ainage ar eas, and t he flood scaling par am et er s can be expr essed as funct ions of annual r unoff obt ained fr om t he w at er balance. This line of invest igat ion pr ovides a new r esear ch dir ect ion for m ak ing flood pr edict ions in a changing global hydr o-clim at e syst em due t o hum an influences (Pielk e et al., 2009 ). Gupt a et al. ( 2007 ) suggest ed specific ideas for fut ur e r esear ch. L im a and L all ( 2010 ) invest igat ed t he r ole of clim at e var iabilit y and change using scaling fr am ew or k for floods. They developed and applied hier ar chical Bayesian m odels t o assess bot h r egional and at -sit e t r ends in t im e in a spat ial scaling fr am ew or k and sim ult aneously pr ovided a r igor ous fr am ew or k for assessing and r educing par am et er and m odel uncer t aint ies. The m odels w er e t est ed w it h r econst r uct ed nat ur al inflow ser ies fr om m or e t han 40 hydr opow er sit es in Br azil w it h cat chm ent s ar eas var ying fr om 2,588 k m 2 t o 823,555 k m 2 . Bot h annual m axim um flood ser ies and m ont hly st r eam flow s w er e consider ed. Cr oss-validat ed r esult s show ed t hat t he hier ar chical Bayesian m odels ar e able t o sk illfully est im at e m ont hly and flood flow pr obabilit y dist r ibut ion par am et er s for sit es not used in m odel fit t ing. The m odels developed can be used t o pr ovide r ecor d augm ent at ion at sit es t hat have shor t r ecor ds or t o est im at e flow s at ungauged sit es, even in t he absence of an assum pt ion of t im e st at ionar it y. Since m odel uncer t aint ies ar e account ed for, t he pr ecision of t he est im at es can be quant ified and hypot heses t est s for r egional and at -sit e t r ends can be for m ally m ade. Viglione, M er z, Viet Dung, Par ajk a, N est er, and Bloschl ( 2016 ) developed a new fr am ew or k for at t r ibut ing flood changes due t o at m ospher ic pr ocesses (e.g., incr easing pr ecipit at ion), cat chm ent pr ocesses (e.g., soil com pact ion associat ed w it h land use change), and r iver syst em pr ocesses (e.g., loss of r et ent ion volum e in t he floodplains) based on r egional analysis. Spat ial scaling of flood changes enabled t hem t o invest igat e at t r ibut ion of m ult iple dr iver s, and Bayesian m et hod allow ed est im at ion of t he at t r ibut ion uncer t aint ies. Flood peak dat a for 97 r iver gauges in upper Aust r ia, w it h ar eas r anging
Page 24 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s fr om 10 k m 2 t o 79,500 k m 2 and r ecor ds for at least 40 year s aft er 1950, w er e used in a r eal case st udy t o illust r at e t he fr am ew or k . Tw o st udies t hat focus on a physical under st anding of floods on t he planet ar y scale ar e O’Conner, Gr ant , and Cost a ( 2002 ) and Devineni, L all, Xi, and War d ( 2015 ). O’Conner et al. ( 2002 ) dealt w it h paleo flood hydr ology, and scaling in floods is im plicit in t heir ar t icle (see Figur e 7 ). Devineni et al. ( 2015 ) invest igat ed t he scaling of ext r em e r ainfall ar eas at t he planet ar y scale.
En g i n eer i n g A p p l i c at i o n s o f t h e Sc al i n g T h eo r y o f Fl o o d s An im por t ant need is t o apply t he scient ific fr am ew or k t hat t he scaling t heor y has developed for innovat ive applicat ions t o pr act ical hydr ologic engineer ing pr oblem s. Tw o exam ples fr om t he paper s t hat t he I ow a Flood Cent er (I FC), U niver sit y of I ow a, has published ar e discussed her e. The I FC w as est ablished follow ing t he 2008 r ecor d floods and is t he only facilit y of it s k ind in t he U nit ed St at es and t he w or ld conduct ing com pr ehensive scient ific and applied r esear ch on floods r oot ed in t he scaling fr am ew or k .
R eal -T i m e Fl o o d Fo r ec ast i n g Kr ajew sk i et al. ( 2017 ) developed a r eal-t im e flood for ecast ing and infor m at ion dissem inat ion syst em for use by all I ow ans. The syst em com plem ent s t he oper at ional for ecast ing issued by t he N at ional Weat her Ser vice. At it s cor e t he I FC for ecast ing m odel is a cont inuous RF-RO m odel based on landscape decom posit ion int o hillslopes and channel link s. Rainfall conver sion t o r unoff is m odeled t hr ough soil m oist ur e account ing at hillslopes. Channel r out ing is based on a non-linear r epr esent at ion of w at er velocit y t hat consider s t he dischar ge am ount as w ell as t he upst r eam dr ainage ar ea. M at hem at ically, t he m odel r epr esent s a lar ge syst em of or dinar y differ ent ial equat ions or ganized t o follow r iver net w or k t opology. N one of physical par am et er s ar e calibr at ed. I t illust r at es t he im pact of t he scient ific ideas developed in t he scaling t heor y of floods on engineer ing applicat ions. The I FC also developed an efficient num er ical solver suit able for high-per for m ance com put ing ar chit ect ur e. The solver allow s t he I FC t o updat e for ecast s ever y 15 m inut es for m or e t han 1,000 I ow a com m unit ies. The input t o t he syst em com es fr om a r adar r ainfall algor it hm , developed in-house, t hat m aps r ainfall ever y five m inut es w it h high spat ial r esolut ion. The algor it hm uses L evel I I r adar r eflect ivit y and ot her polar im et r ic dat a fr om t he WSR-88DP r adar net w or k . A lar ge libr ar y of flood inundat ion m aps and r eal-t im e r iver st age dat a fr om over 200 I FC “ st r eam -st age sensor s” com plem ent t he I FC
Page 25 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s infor m at ion syst em . The syst em com m unicat es all t his infor m at ion t o t he gener al public t hr ough a com pr ehensive br ow ser -based and int er act ive plat for m .
I m p ac t o f D am s o n R eal -T i m e Fl o o d Fo r ec ast i n g Dam s ar e ubiquit ous in t he U nit ed St at es, and m or e t han 87,000 of t hem acr oss t he nat ion influence st r eam flow s. The significant m ajor it y of t hese dam s ar e sm all and ar e oft en ignor ed in r eal-t im e flood for ecast ing oper at ions and at -sit e and r egional flood fr equency est im at ions. Even t hough t he im pact of individual sm all dam s on floods is oft en lim it ed, t he com bined flood at t enuat ion effect s of a syst em of such dam s can be significant . Ayalew, Kr ajew sk i, M ant illa, Wr ight , and Sm all ( 2017 ) invest igat ed how a syst em of spat ially dist r ibut ed sm all dam s affect flood fr equency acr oss a r ange of dr ainage basin scales using t he 660 k m 2 Soap Cr eek w at er shed in sout heast er n I ow a, w hich cont ains m or e t han 133 sm all dam s. Result s fr om cont inuous sim ulat ion of t he syst em of sm all dam s indicat e t hat t he peak dischar ges w er e r educed bet w een 20 and 70% but t hat t his effect decr eases as t he dr ainage ar ea incr eases. Consider ing t hat m or e sm all dam s ar e being built acr oss w at er sheds in I ow a and elsew her e in t he count r y, t heir r esult s highlight how t he peak dischar ge at t enuat ion effect s of t hese dam s is an addit ional fact or t hat invalidat es t he st at ionar it y assum pt ion t hat is used in at -sit e and RFF analysis. I n par t icular, t hey show ed t hat neglect ing t he effect s of a syst em of sm all dam s can lead t o an over est im at ion of flood r isk . The r esult s show t hat r eal-t im e flood for ecast ing t hat does not account for t he flood at t enuat ion effect s of t hese dam s m ay suffer fr om t he over est im at ion of flood t hr eat s acr oss a r ange of spat ial scales.
Page 26 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Co n c l u si o n s Pr edict ion of floods in ungauged basins (i.e., at locat ions w her e no st r eam flow dat a exist ) is a global issue because m ost of t he count r ies involved do not have adequat e st r eam flow r ecor ds. Tw o appr oaches have been developed t o solve t he pr oblem . Fir st is t he RFF analysis by t he U SGS. RFF equat ions ar e pur e st at ist ical char act er izat ions t hat use hist or ical st r eam flow r ecor ds in “ hom ogeneous r egions” t o pr edict annual peak flow quant iles in ungauged basins. The second appr oach is t he scaling t heor y of floods t hat abandoned t he hom ogeneous r egions concept and adopt ed r iver basins inst ead. I t has t he explicit goal of incor por at ing flood-pr oducing physical pr ocesses in under st anding t he obser ved scaling, or pow er law r elat ions, bet w een peak flow s and dr ainage ar eas in r iver basins. The scaling t heor y of floods offer s a unified fr am ew or k t o pr edict floods in RF-RO event s and in annual peak flow quant iles in ungauged basins. The t opics cover ed in t he fir st par t of t he ar t icle include: (1) a br ief hist or y is pr esent ed of t he scaling t heor y of floods; (2) dynam ic for m ulat ions gover ning RF-RO event s in r iver basins ar e descr ibed; (3) self-sim ilar r iver net w or k s (t he Tok unaga m odel and t he RSN m odel) ar e br iefly r eview ed; (4) applicat ions of dynam ic for m ulat ion in RSN ar e sum m ar ized; (5) an analyt ical t heor y of H -G in r iver net w or k s is highlight ed and open pr oblem s m ent ioned; and (6) a hydr o-ecological t heor y of int er m it t ent r ipar ian diver sit y on Tok unaga st r eam net w or k s in pr ogr ess is br iefly sk et ched. The t opics in t he second par t include: (1) highlight s of t he fir st paper on a physical basis of annual peak flow quant iles ar e pr esent ed; (2) highlight s of t he discover y of t he classical and t he gener alized H or t on law s for peak flow s in t he 32,400 k m 2 I ow a r iver basin ar e descr ibed. An applicat ion of t he H or t on law s for diagnosing a RF-RO m odel w it hout calibr at ing it s par am et er s is sum m ar ized; (3) gener alizat ion of t he scaling t heor y of floods in m edium t o lar ge basins and t o longer t im e scales t han annual ar e highlight ed (t w o k ey r efer ences on scaling in r ainfall and paleofloods t o under st and scaling in floods on t he planet ar y scale ar e included); and (4) t w o engineer ing applicat ions of scaling t heor y of floods in t he I ow a River basin ar e sum m ar ized. Sever al significant ar eas of r esear ch r em ain: 1 . Com par at ive st udies ar e needed t o t est differ ent m odel par am et er izat ions w it hout calibr at ion using t he H or t on law s for peak flow s. 2 . St udies ar e needed t o t est analyt ical solut ions of Eq. ( 3 ) under physically r ealist ic assum pt ions in r ealist ic self-sim ilar r iver net w or k s. Ram ir ez (N at ional U niver sit y of Colom bia, per sonal com m unicat ion, 2017) has obt ained analyt ical r esult s on floods using t he physical fr am ew or k pr esent ed her e. I t is t he fir st effor t of it s k ind, and paper s ar e being pr epar ed for publicat ion. 3 . The scaling t heor y of floods needs t o be ext ended t o global basins r epr esent ing differ ent hydr o-clim at ic r egions r anging fr om ver y hum id t o sem i-ar id and ar id.
Page 27 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s 4 . The scaling fr am ew or k has im por t ant r elevance t o flood pr edict ion in a changing clim at e, because t he self-sim ilar it y of r iver net w or k s is not expect ed t o change. As a r esult , t he pow er law s in peak dischar ges at event and annual t im e scales w ould r em ain int act , but scaling slopes and int er cept s for RF-RO event s and annual flood quant iles w ould change, w hich can be pr edict ed fr om physical pr ocesses. 5 . M any physical pr ocesses gover n floods in upst r eam basins and in flood plains, but all of t hem cannot be included in a m odel. A dynam ic under st anding of scaling can be used t o elim inat e t hose physical pr ocesses t hat play a “ secondar y r ole” in t he over all cont ext .
A c k n o w l ed g m en t s Sever al out st anding gr aduat e st udent s helped m e sust ain t he line of r esear ch descr ibed her e over sever al decades. M y jour ney st ar t ed in 1980 w it h t he fir st gr aduat e st udent , Oscar M esa, and ended in 2007 w it h m y last gr aduat e st udent , Ricar do M ant illa. I n bet w een t his t im e fr am e, sever al ot her s joined t he gr oup, Tom over, Sandr a Cast r o, Scot t Peck ham , Set h Veit zer, Pet er Fur ey, and cont r ibut ed t o t he line of r esear ch. I expr ess m y deep gr at it ude for t heir dedicat ed cont r ibut ions. The N at ional Science Foundat ion, t he Ar m y Resear ch Office and t he N at ional Aer onaut ics and Space Adm inist r at ion funding over t his t im e per iod m ade t he r esear ch possible.
R ef er en c es Ayalew, T. B., Kr ajew sk i, W. F., & M ant illa, R. (2014a). Co n n ec t i n g t h e p o w er -l aw sc al i n g st r u c t u r e o f p eak -d i sc h ar g es t o sp at i al l y var i ab l e r ai n f al l an d c at c h m en t p h ysi c al p r o p er t i es. Advances in Wat er Resour ces, 71, 32–43. Ayalew, T. B., Kr ajew sk i, W. F., & M ant illa, R. (2015). A n al yzi n g t h e ef f ec t s o f exc ess r ai n f al l p r o p er t i es o n t h e sc al i n g st r u c t u r e o f p eak -d i sc h ar g es: I n si g h t s f r o m a m eso sc al e r i ver b asi n . Wat er Resour ces Resear ch, 51(6), 3900–3921. Ayalew, T. B., Kr ajew sk i, W. F., M ant illa, R., & Sm all, S. J. (2014b). Exp l o r i n g t h e ef f ec t s o f h i l l sl o p e-c h an n el l i n k d yn am i c s an d exc ess r ai n f al l p r o p er t i es o n t h e sc al i n g st r u c t u r e o f p eak -d i sc h ar g e. Advances in Wat er Resour ces, 64, 9–20. Ayalew, T. B., Kr ajew sk i, W. F., M ant illa, R., Wr ight , D. B., & Sm all, S. J. (2017). T h e ef f ec t o f sp at i al l y d i st r i b u t ed sm al l d am s o n f l o o d f r eq u en c y: I n si g h t s f r o m t h e So ap Cr eek w at er sh ed . Jour nal of H ydr ologic Engineer ing, 22(7), 1–10. Bar enblat t , G. I . (1996). Scaling, Self-sim ilar it y and int er m ediat e asym pt ot ics. L ondon: Cam br idge U niver sit y Pr ess.
Page 28 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s Budyk o, M . I . (1974). Clim at e and life (D. H . M iller, Ed., English ed.). N ew Yor k : Academ ic. Bur d, G. A., Waym ir e, E. C., & Winn, R. D. (2000). A self-sim ilar invar iance of cr it ical binar y Galt on-Wat son t r ees. Ber noulli, 6, 1–21. Conover, W. J. (1999). Pr act ical nonpar am et r ic st at ist ics. N ew Yor k : Wiley. Cor r adini, C., Govindar aju, R. S., & M or bidelli, R. (2002). Sim plified m odeling of ar eal infilt r at ion at hillslope scale. H ydr ological Pr occesses, 16, 1757–1770. Daw dy, D. (2007). Pr edict ion ver sus under st anding (The 2006 Ven Te Chow lect ur e) (For um ). Jour nal of H ydr ologic Engineer ing, 12(1), 1–3. Daw dy, D. (2016). Daw dy’s biogr aphical sk et ch and publicat ions ar e given in: h yd r o l o g y.ag u .o r g /vi r t u al -h yd r o l o g i st s-p r o j ec t -d avi d -d aw d y/. Daw dy, D., Gr iffis, V., & Gupt a, V. (2012). R eg i o n al f l o o d -f r eq u en c y an al ysi s: H o w w e g o t h er e an d w h er e w e ar e g o i n g . Jour nal of H ydr ologic Engineer ing, 17, 953–959. Devineni, N., L all, U., Xi, C., & War d, P. (2015). Sc al i n g o f ext r em e r ai n f al l ar eas at a p l an et ar y sc al e. Chaos, 25, 07547. Duffy, C. (1996). A t w o st at e int egr al balance m odel for soil m oist ur e and gr oundw at er dynam ics in com plex t er r ain. Wat er Resour ces Resear ch, 32(8), 2421–2434. Dunn, W. C., M ilne, B. T., M ant illa, R., & Gupt a, V. K. (2011). Sc al i n g r el at i o n s b et w een r i p ar i an veg et at i o n an d st r eam o r d er i n t h e, W h i t ew at er R i ver n et w o r k , K an sas, U SA . L andscape Ecology, 26, 983–996. Feder, J. (1988). Fr act als. N ew Yor k : Plenum . Fr eeze, R. A. (1980). A st ochast ic concept ual m odel of r ainfall-r unoff pr ocesses on a hillslope. Wat er Resour ces Resear ch, 16(2), 391–408. Fur ey, P., & Gupt a, V. K. (2005). Ef f ec t s o f exc ess r ai n f al l o n t h e t em p o r al var i ab i l i t y o f o b ser ved p eak d i sc h ar g e p o w er l aw s. Advances in Wat er Resour ces, 28, 1240–1253. Fur ey, P., & Gupt a, V. K. (2007). D i ag n o si n g p eak -d i sc h ar g e p o w er l aw s o b ser ved i n r ai n f al l -r u n o f f even t s i n Go o d w i n Cr eek Exp er i m en t al Wat er sh ed . Advances in Wat er Resour ces, 30, 2387–2399. Fur ey, P., Gupt a, V. K., & Tr out m an, B. (2013). A t op-dow n m odel t o gener at e ensem bles of r unoff fr om a lar ge num ber of hillslopes. N onlinear Pr ocesses in Geophysics, 20(5), 683.
Page 29 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s Fur ey, P., Tr out m an, B., Gupt a, V. K., & Kr ajew sk i, W. F. (2016). Co n n ec t i n g even t -b ased sc al i n g o f f l o o d p eak s t o r eg i o n al f l o o d f r eq u en c y r el at i o n sh i p s. Jour nal of H ydr ologic Engineer ing, 21(10), 1–11. Fur ey, P. R., & Tr out m an, B. M . (2008). A consist ent fr am ew or k for H or t on r egr ession st at ist ics t hat leads t o a m odified H ack ’s law. Geom or phology, 102(3–4), 603–614. Govindar aju, R. S., Cor r adini, C., & M or bidelli, R. (2012). L o c al - an d f i el d -sc al e i n f i l t r at i o n i n t o ver t i c al l y n o n -u n i f o r m so i l s w i t h sp at i al l y var i ab l e su r f ac e h yd r au l i c c o n d u c t i vi t i es. H ydr ologic Pr ocesses, 26, 3293–3301. Gupt a, V. K. (2004a). Em er gence of st at ist ical scaling in floods fr om com plex r unoff dynam ics on channel net w or k s. Fr act als, Chaos and Solit ons, 19, 357–365. Gupt a, V. K. (2004b). Pr edict ion of st at ist ical scaling in peak flow s for r ainfall–r unoff event s: A new fr am ew or k for t est ing physical hypot heses. I n Scales in hydr ology and w at er m anagem ent , Kovacs Colloquium I nt er nat ional Associat ion of H ydr ologic Sciences. Pub. 287, 97–110. Gupt a, V. K., Ayalew, T. B., M ant illa, R., & Kr ajew sk i, W. F. (2015). Cl assi c al an d g en er al i zed H o r t o n l aw s f o r p eak f l o w s i n r ai n f al l -r u n o f f even t s. Chaos, 25, 075408. Gupt a, V. K., Cast r o, S., & Over, T. M . (1996). On scaling exponent s of spat ial peak flow s fr om r ainfall and r iver net w or k geom et r y. I n P. Bur lando, G. M enduni, & R. Rosso (Eds.), Fr act als, Scaling and N onlinear Var iabilit y in H ydr ology (Special issue). Jour nal of H ydr ology, 187(1–2), 81–104. Gupt a, V. K., & Daw dy, D. (1995). Physical int er pr et at ion of r egional var iat ions in t he scaling exponent s in flood quant iles. H ydr ologic Pr ocesses, 9, 347–361. Gupt a, V. K., M ant illa, R., Tr out m an, B., Daw dy, D., & Kr ajew sk i, W. F. (2010). Gen er al i zi n g a n o n l i n ear g eo p h ysi c al f l o o d t h eo r y t o m ed i u m -si zed r i ver n et w o r k s. Geophysical. Resear ch L et t er s, 37(11), L 11402. Gupt a, V. K., & M esa, O. J. (1988). Runoff gener at ion and hydr ologic r esponse via channel net w or k geom or phology: Recent pr ogr ess and open pr oblem s. Jour nal of H ydr ology, 102, 3–28. Gupt a, V. K., & M esa, O. J. (2014). H o r t o n l aw s f o r H yd r au l i c -Geo m et r i c var i ab l es an d t h ei r sc al i n g exp o n en t s i n sel f -si m i l ar To k u n ag a r i ver n et w o r k s. N onlinear Pr ocesses in Geophysics, 21, 1007–1025. Gupt a, V. K., M esa, O. J., & Daw dy, D. R. (1994). M ult iscaling t heor y of flood peak s: Regional quant ile analysis. Wat er Resour ces Resear ch, 30, 3405–3421.
Page 30 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s Gupt a, V. K., Tr out m an, B., & Daw dy, D. (2007). Tow ar ds a nonlinear geophysical t heor y of floods in r iver net w or k s. An over view of 20 year s of pr ogr ess. I n A. A. Tsonis & J. B. Elsner (Eds.), Tw ent y year s of nonlinear dynam ics in geosciences (pp. 121–152). N ew Yor k : Elsevier. Gupt a, V. K., & Waym ir e, E. (1998). Spat ial var iabilit y and scale invar iance in hydr ologic r egionalizat ion. I n G. Sposit o (Ed.), Scale dependence and scale invar iance in hydr ology (pp. 88–135). Cam br idge, U.K.: Cam br idge U niver sit y Pr ess. Gupt a, V. K., Waym ir e, E., & Wang, C. T. (1980). A r epr esent at ion of an inst ant aneous unit hydr ogr aph fr om geom or phology. Wat er Resour ces Resear ch, 16(5), 855–862. I bbit t , R. P., M cKer char, A. I ., & Duncan, M . J. (1998). Taier i r iver dat a t o t est channel net w or k and r iver basin het er ogeneit y concept s. Wat er Resour ces Resear ch, 34, 2085– 2088. Jar vis, R. S., & Woldenber g, M . J. (Eds.). (1984). Benchm ar k paper s in geology (Vol. 80). St r oudsbur g, PA: H ut chinson Ross. Jor dan, A. C., & Kean, J. W. (2010). Est ab l i sh i n g a m u l t i -sc al e st r eam g ag i n g n et w o r k i n t h e W h i t ew at er r i ver b asi n , K an sas. Wat er Resour ces M anagem ent , 24, 3641–3664. Kean, J. W., & Sm it h, J. D. (2005). Gen er at i o n an d ver i f i c at i o n o f t h eo r et i c al r at i n g c u r ves i n t h e W h i t ew at er r i ver b asi n , K an sas. Jour nal of Geophysical. Resear ch, 110, F04012, 1–17. Kir k by, M . J. (1976). Test s of t he r andom net w or k m odel, and it s applicat ion t o basin hydr ology. Ear t h Sur face Pr ocesses, 1, 197–212. Kr ajew sk i, W. et al. (2017). R eal t i m e f l o o d f o r ec ast i n g an d i n f o r m at i o n syst em f o r t h e st at e o f I o w a. Bullet in Am er ican M et eor ological Societ y, 98(3), 1–16. L ee, M . T., & Delleur, J. W. (1976). A var iable sour ce ar ea m odel of t he r ainfall-r unoff pr ocess based on t he w at er shed st r eam net w or k . Wat er Resour ces Resear ch, 12(5), 1029– 1035. L eopold, L . B., & M iller, J. P. (1956). Epher m al st r eam s—H ydr aulic fact or s and t heir r elat ion t o t he dr ainage net . U S Geological Sur vey Pr of. Paper 282-A. Washingt on, DC: U.S. Govt . Pr int ing Office. L eopold, L . B., Wolm an, M ., & M iller, G. (1964). Fluvial pr ocesses in geom or phology. San Fr ancisco: Fr eem an. L im a, C. H . R., & L all, U. (2010). Spat ial scaling in a changing clim at e: A hier ar chical bayesian m odel for nonst at ionar y m ult i-sit e annual m axim um and m ont hly st r eam fl ow. Jour nal of H ydr ology, 383(3–4), 307–318.
Page 31 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s L insley, R. K., Jr., Kohler, M . A., & Paulhus, J. H . L . (1982). H ydr ology for engineer s (3d ed.). N ew Yor k : M cGr aw -H ill. M andapak a, P. V., Kr ajew sk i, W. F., M ant illa, R., & Gupt a, V. K. (2009). Dissect ing t he effect of r ainfall var iabilit y on t he st at ist ical st r uct ur e of peak flow s. Advances in Wat er Resour ces, 32, 1508–1525. M andelbr ot , B. (1982). The fr act al geom et r y of nat ur e. N ew Yor k : Fr eem an. M ant illa, R., & Gupt a, V. K. (2005). A GI S num er ical fr am ew or k t o st udy t he pr ocess basis of scaling st at ist ics on r iver net w or k s. I EEE Geoscience and Rem ot e Sensing L et t er s, 2(4), 404–408. M ant illa, R., Gupt a, V. K., & M esa, O. J. (2006). Role of coupled flow dynam ics and r eal net w or k st r uct ur es on H or t onian scaling of peak flow s. Jour nal of H ydr ology, 322, 155– 167. M ant illa, R., Gupt a, V. K., & Tr out m an, B. M . (2012). Ext ending gener alized H or t on law s t o t est em bedding algor it hm s for t opological r iver net w or k s. Geom or phology, 151–152, 13–26. M ant illa, R., Tr out m an, B. M ., & Gupt a, V. K. (2011). Scaling of peak flow s w it h const ant flow velocit y in r andom self-sim ilar r iver net w or k s. N onlinear Pr ocesses in Geophysics, 18, 489–502. M ant illa, R., Tr out m an, B. M ., & Gupt a, V. K. (2010). Test i n g st at i st i c al sel f -si m i l ar i t y i n t h e t o p o l o g y o f r i ver n et w o r k s. Jour nal of Geophysical Resear ch, 115, 1–12. M cConnell, M ., & Gupt a, V. K. (2008). A pr oof of t he H or t on law of st r eam num ber s for t he Tok unaga m odel of r iver net w or k s. Fr act als, 16(3), 227–233. M cKer char, A. I ., I bbit t , R. P., Br ow n, S. L . R., & Duncan, M . J. (1998). Dat a for Ashley r iver t o t est channel net w or k and r iver basin het er ogeneit y concept s. Wat er Resour ces Resear ch, 34, 139–142. M enabde, M ., & Sivapalan, M . (2001). L ink ing space-t im e var iabilit y of r ainfall and r unoff fields: A dynam ic appr oach. Advances in Wat er Resour ces, 24, 1001–1014. M ilne, B., & Gupt a, V. K. (2017a). H ydr o-ecological t heor y of int er m it t ent r ipar ian diver sit y on st r eam net w or k s. M ilne, B., & Gupt a, V. K. (2017b). H o r t o n R at i o s L i n k Sel f -Si m i l ar i t y w i t h M axi m u m En t r o p y o f Ec o -Geo m o r p h o l o g i c al Pr o p er t i es i n St r eam N et w o r k s. Ent r opy, 19, 1– 15. O’Conner, J. E., Gr ant , G. E., & Cost a, J. E. (2002). The geology and geogr aphy of floods. I n K. P. H ouse, R. H . Webb, V. P. Bak er, & D. R. L evish (Eds.), Ancient floods, m oder n
Page 32 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s hazar ds: Pr inciples and applicat ions of paleoflood hydr ology (pp. 359–385). Washingt on, DC: Am er ican Geophysical U nion. Ogden, F. L ., & Daw dy, D. R. (2003). Peak dischar ge scaling in sm all H or t onian w at er shed. Jour nal of H ydr ologic Engineer ing, 8(2), 64–73. Peck ham , S. (1995). N ew r esult s for self-sim ilar t r ees w it h applicat ions t o r iver net w or k s. Wat er Resour ces Resear ch, 31(4), 1023–1029. Peck ham , S., & Gupt a, V. (1999). A r efor m ulat ion of H or t on’s law s for lar ge r iver net w or k s in t er m s of st at ist ical self-sim ilar it y. Wat er Resour ces Resear ch, 35(9), 2763– 2777. Pielk e, R., Sr., et al. (2009). Clim at e change: t he need t o consider hum an for cings besides gr eenhouse gases. Eos, 90(45), 413. Poveda, G., et al. (2007). L ink ing long-t er m w at er balances and st at ist ical scaling t o est im at e r iver fl ow s along t he dr ainage net w or k of Colom bia. Jour nal of H ydr ologic Engineer ing, 12(1), 4–13. Reif, F. (1965). Fundam ent als of st at ist ical and t her m al physics. N ew Yor k : M cGr aw -H ill. Rodr iguez-I t ur be, I ., & Rinaldo, A. (1997). Fr act al r iver basins. L ondon: Cam br idge U niver sit y Pr ess. Rodr iguez-I t ur be, I ., & Valdez, J. B. (1979). The geom or phologic st r uct ur e of hydr ologic r esponse. Wat er Resour ces Resear ch, 15(6), 1409–1420. Shr eve, R. L . (1966). St at ist ical law of st r eam num ber s. Jour nal of Geology, 74, 17–37. Shr eve, R. L . (1967). I nfinit e t opologically r andom channel net w or k s. Jour nal of Geology, 75, 178–186. Sm it h, J. A., Baeck , M . L ., Villar ini, G., Wr ight , D. B., & Kr ajew sk i, W. (2013). Ext r em e flood r esponse: The June 2008 flooding in I ow a. Jour nal of H ydr om et eor ology, 14(6), 1810–1825. Tok unaga, E. (1966). The com posit ion of dr ainage net w or k s in Toyohir a r iver basin and valuat ion of H or t on’s fir st law [ in Japanese w it h English sum m ar y] . Geophysics Bullet in H ok k aido U niver sit y, 15, 1–19. Tok unaga, E. (1978). Consider at ion on t he com posit ion of dr ainage net w or k s and t heir evolut ion. Geogr aphy Repor t 13, Tok yo M et r opolit an U niver sit y, 1–27. Tr out m an, B. M . (2016). h t t p s://w at er .u sg s.g o v/n r p /p u b l i c at i o n s.p h p ? p I D = 5 0 4 2 1 6 b 7 e4 b 0 4 b 5 0 8 b f d 3 3 4 f & sc i N am e= B r en t %2 0 M .%2 0 Tr o u t m an .
Page 33 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s Tr out m an, B., & Kar linger, M . (1984). On t he expect ed w idt h funct ion of t opologically r andom channel net w or k s. Jour nal of Applied Pr obabilit y, 22, 836–849. Tr out m an, B., & Kar linger, M . (1988). Asym pt ot ic Rayleigh inst ant aneous unit hydr ogr aph. St ochast ic H ydr ology and H ydr aulaulics, 2, 73–78. Tr out m an, B., & Kar linger, M . (1989). Pr edict or s of peak w idt hs for net w or k s w it h exponent ial link s. St ochast ic H ydr ology and H ydr aulics, 3, 1–16. Tr out m an, B., & Kar linger, M . (1998). Spat ial channel net w or k m odels in hydr ology. I n O. E. Bar ndor ff-N ielsen, V. K. Gupt a, V. Per ez-Abr eu, & E. C. Waym ir e (Eds.), St ochast ic m et hods in hydr ology: Rain, landfor m s and floods. Singapor e: Wor ld Scient ific. Tr out m an, B., & Over, T. (2001). River flow m ass exponent s w it h fr act al channel net w or k s and r ainfall. Advances in Wat er Resour ces, 24, 967–989. Tr out m an, B. M . (2005). Scaling of flow dist ance in r andom self-sim ilar channel net w or k s. Fr act als, 13(4), 265–282. Tur cot t , D. (1997). Fr act als and chaos in geology and geophysics (2d ed.). N ew Yor k : Cam br idge U niver sit y Pr ess. Veit zer, S. A., & Gupt a, V. K. (2000). Random self-sim ilar r iver net w or k s, sim ple scaling and gener alized H or t on law s. Wat er Resour ces Resear ch, 36(4), 1033–1048. Veit zer, S. A., & Gupt a, V. K. (2001). St at ist ical self-sim ilar it y of w idt h funct ion m axim a w it h im plicat ions t o floods. Advances in Wat er Resour ces, 24, 955–965. Viglione, A., M er z, B., Viet Dung, N., Par ajk a, J., N est er T., & Bloschl, G. (2016). At t r i b u t i o n o f r eg i o n al f l o o d c h an g es b ased o n sc al i n g f r am ew o r k . Wat er Resour ces Resear ch, 52, 5322–5340. Waym ir e, E. (2016). h t t p ://m at h .o r eg o n st at e.ed u /p eo p l e/vi ew /w aym i r ee. Zhang, J., & Wu, Y. (2007). K-Sam ple t est s based on t he lik elihood r at io. Com put at ional St at ist ics & Dat a Analysis, 51(9), 4682–4691.
Vi j ay Gu p t a
Civil, Envir onm ent al and Ar chit ect ur al Engineer ing, U niver sit y of Color ado
Page 34 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;
Sc al i n g T h eo r y o f Fl o o d s f o r D evel o p i n g a Ph ysi c al B asi s o f St at i st i c al Fl o o d Fr eq u en c y R el at i o n s
Page 35 of 35
PRI N TED FROM t he OXFORD RESEARCH EN CYCL OPEDI A, N ATU RAL H AZARD SCI EN CE (nat ur alhazar dscience.oxfor dr e.com ). (c) Oxfor d U niver sit y Pr ess U SA, 2016. All Right s Reser ved. Per sonal use only;