Assessing the potential of the dart model to discrete return lidar simulation—application to fuel type mapping

emo e sensing A icle Assessing he Po en ial o he DART Model o Disc e e Re u n LiDAR Simula ion—Applica ion o Fuel Type Mapping Se gio Re illa 1,2 , Ma ía Te esa Lamelas 2,3,*,† , Da ío Domingo 2,4,† , Juan de la Ri a 2, Raquel Mon o io 2, An onio Luis Mon ealeg e 2and Albe o Ga cía-Ma ín2,3   Ci a ion: Re illa, S.; Lamelas, M.T.; Domingo, D.; de la Ri a, J.; Mon o io, R.; Mon ealeg e, A.L.; Ga cía-Ma ín, A. Assessing he Po en ial o he DART Model o Disc e e Re u n LiDAR Simula ion—Applica ion o Fuel Type Mapping. Remo e Sens. 2021,13, 342. h ps://doi.o g/ 10.3390/ s13030342 Academic Edi o : Pe e K zys ek Recei ed: 11 Decembe 2020 Accep ed: 16 Janua y 2021 Published: 20 Janua y 2021 Publishe ’s No e: MDPI s ays neu al wi h ega d o ju isdic ional claims in published maps and ins i u ional a il- ia ions. Copy igh : © 2021 by he au ho s. Licensee MDPI, Basel, Swi ze land. This a icle is an open access a icle dis ibu ed unde he e ms and condi ions o he C ea i e Commons A ibu ion (CC BY) license (h ps:// c ea i ecommons.o g/licenses/by/ 4.0/). 1Ins i u o Geog á ico de A agón, Pº Ma ía Agus ín, 36, Edi icio Pigna elli, 50071 Za agoza, Spain; [email p o ec ed] 2 GEOFOREST-IUCA Resea ch G oup, Depa men o Geog aphy, Uni e si y o Za agoza, Ped o Ce buna 12, 50009 Za agoza, Spain; [email p o ec ed] (D.D.); dela i a@uniza .es (J.d.l.R.); mon o io@uniza .es (R.M.); mon eale@uniza .es (A.L.M.); alga cia@uniza .es (A.G.-M.) 3Cen o Uni e si a io de la De ensa de Za agoza, Academia Gene al Mili a , C a. de Huesca s/n, 50090 Za agoza, Spain 4Land Change Science Resea ch Uni , Swiss Fede al Ins i u e o Fo es , Snow and Landscape Resea ch WSL, Zu che s asse 111, 8930 Bi mensdo , Swi ze land *Co espondence: lamelas@uniza .es; Tel.: +34-976-739-866; Fax: +34-976-739-824 † These au ho s con ibu ed equally o his wo k. Abs ac : Fuel ype is one o he key ac o s o analyzing he po en ial o i e igni ion and p opaga ion in ag icul u al and o es en i onmen s. The inc ease o h ee-dimensional da ase s p o ided by ac i e senso s, such as LiDAR (Ligh De ec ion and Ranging), has imp o ed he classi ica ion o uel ypes h ough empi ical modelling. Empi ical me hods a e si e and senso speci ic while Radia i e T ans e Models (RTM) app oaches p o ide b oade uni e sali y. The aim o his wo k is o analyze he sui abili y o Disc e e Aniso opic Radia i e T ans e (DART) model o eplica e low densi y small- oo p in Ai bo ne Lase Scanning (ALS) measu emen s and subsequen uel ype classi ica ion. Field da a measu ed in 104 plo s a e used as g ound u h o simula e LiDAR esponse based on he senso and ligh cha ac e is ics o low-densi y ALS da a cap u ed by he Spanish Na ional Plan o Ae ial O hopho og aphy (PNOA) in wo di e en da es (2011 and 2016). The accu acy assessmen o he DART simula ions is pe o med using Spea man ank co ela ion coe icien s be ween he simula ed me ics and he ALS-PNOA ones. The esul s show ha 32% o he compu ed me ics o e passed a co ela ion alue o 0.80 be ween simula ed and ALS-PNOA me ics in 2011 and 28% in 2016. The highes co ela ions we e ela ed o high heigh pe cen iles, canopy a iabili y me ics as o example s anda d de ia ion and Rumple di e si y index, eaching co ela ion alues o e 0.94. Two me ic selec ion app oaches and Suppo Vec o Machine classi ica ion me hod wi h a ian s we e compa ed o classi y uel ypes. The bes - i ed classi ica ion model, ained wi h he DART simula ed sample and alida ed wi h ALS-PNOA da a, was ob ained using Suppo Vec o Machine me hod wi h adial ke nel. The o e all accu acy o he classi ica ion a e alida ion was 88% and 91% o he 2011 and 2016 yea s, espec i ely. The use o DART demons a es i s alue o simula ing gene alizable 3D da a o uel ype classi ica ion p o iding ele an in o ma ion o o es manage s in i e p e en ion and ex inc ion. Keywo ds: 3D Radia i e ans e model (RTM); low-densi y ai bo ne lase scanning (ALS) da a; P ome heus uel ypes; Medi e anean o es 1. In oduc ion Fuel ypes a e de ined by Me ill and Alexande [ 1 ] as “an iden i iable associa ion o uel elemen s o dis inc i e species, o m, size, a angemen and con inui y ha will exhibi cha ac e is ic i e beha io unde de ined bu ning condi ions.” Fuel ype mapping is c ucial o o es managemen and i e isk assessmen as he spa ial dis ibu ion o uel a ec s wild i e igni ion and p opaga ion. Remo e Sens. 2021,13, 342. h ps://doi.o g/10.3390/ s13030342 h ps://www.mdpi.com/jou nal/ emo esensing Remo e Sens. 2021,13, 342 2 o 20 His o ically, o es i es ha e had a ele an impac in Medi e anean landscapes, howe e , in he las decades, he ecu ence, magni ude and se e i y o wild i es ha e inc eased [ 2 ]. In addi ion o o he ac o s de i ed om clima e change, o example, he inc ease in empe a u e, one o he main d i ing o ces o his i e occu ence in- ensi ica ion is he inc ease o ege a ion combus ibili y because o land use and co e changes [ 2 , 3 ]. The abandonmen o he ield c ops leads o he p oli e a ion o bushes inc easing he uel load [ 4 ]. Consequen ly, uel ype mapping has been accomplished by se e al au ho s using emo e sensed da a [ 4 , 5 ], mos ly based on mul ispec al medium- esolu ion senso s [5–8] bu also using hype spec al images cap u ed by senso s on boa d o ai c a s [ 9 , 10 ], ai bo ne and sa elli e LiDAR (Ligh De ec ion and Ranging) da a [ 11 – 13 ] and he combina ion o di e en ypes o senso s [14–19]. The de elopmen o LiDAR echnology cons i u es an impo an ad ance in o es managemen h ough emo e sensing echniques [ 20 ] due o he possibili y o cap u ing he ege a ion e ical p o ile; con a ily o mul ispec al image y om op ical passi e senso s, only sensi i e o he uppe canopy [ 21 , 22 ]. Howe e , mos o he abo emen ioned app oaches o classi ying uel ypes, independen ly om he emo e sensing echnology used, ely on in si u da a o ain algo i hms using s a is ical app oaches [ 23 ]. These empi ical me hods a e si e and senso speci ic and hei esul s a e only applicable o uel ypes p esen in he s udy a ea [ 24 ]. An al e na i e app oach in ol es model aining h ough adia i e ans e simula ions o he e ain cha ac e is ics. These simula ions can p o ide a cos -e ec i e al e na i e o ield su eys while imp o ing he con ol in he expe imen s [ 25 ]. Radia i e T ans e Models (RTM) app oaches a e subjec o an app op ia e model pa ame e iza ion ha also equi es in si u da a, howe e hey o e a be e uni e sali y han empi ical app oaches [26]. P e ious o es pa ame e simula ions conduc ed wi h 3D RTM capable o simula ing he LiDAR esponse, such as FLIGHT [ 27 ] and Disc e e Aniso opic Radia i e T ans e (DART) [ 28 ] models, ha e mainly ocused on la ge- oo p in wa e o m LiDAR acquisi ions om sa elli e [ 29 , 30 ] o onboa d ai c a [ 24 , 26 , 31 ]. Howe e , ew in es iga ions ha e ex- amined small- oo p in disc e e- e u n measu emen s due o he compu a ional demands o simula ing mul i-pulse ALS acquisi ions o e complex o es ed landscapes. F om he bes o ou knowledge, he only s udy a ailable was conduc ed by Robe s e al. [ 25 ]. These au ho s examined he accu acy o he DART model o eplica e small- oo p in ALS mea- su emen s collec ed o e I ish coni e plan a ions and how su ey cha ac e is ics in luenced he p ecision o disc e e- e u n me ics. Thei s udy demons a ed ha DART is a obus model o simula ing high poin densi y (a mean poin densi y o 29 ± 10 e u ns/m 2 ) disc e e- e u n measu emen s o e s uc u ally complex o es s, opening a p omising line o esea ch. Today, se e al coun ies, such as Aus ia, Canada, Denma k, Es onia, England, Finland , I eland, Li huania, Luxembu g, Mal a, No way, Spain, Slo akia, Swi ze land, The Ne he lands and he Uni ed S a es o Ame ica, ha e na ionwide ALS da a co e ages [ 32 ] a ailable o ee in he In e ne . Poin clouds de i ed om such na ional campaigns gen- e ally a e cap u ed by small- oo p in (<1 m) pulsed lase sys ems capable o eco ding a ini e numbe o disc e e e u ns ( ypically <5) pe lase pulse and ha e low poin densi ies in o de o educe cos s [ 33 ]. One o he pu poses o hese campaigns is o de i e di e en o es y a iable mapping a egional scale wi h an ope a ional objec i e. The egional o na ional scale o hese p ojec s makes e en mo e imp ac ical o quan i y di ec ly in si u da a h ough con en ional o es mensu a ion echniques. In o de o o e come his handicap, he speci ic objec i es o he p esen s udy a e: 1. To analyze he accu acy o he DART model o eplica e low densi y small- oo p in ALS measu emen s 2. To assess he abili y o simula ions o model aining o classi y uel ypes. Al hough, he abili y o he DART model o classi y uel models was assessed by Lamelas e al. [ 24 ], his app oach ep esen s he i s a emp o simula e he esponse o low densi y small- oo p in senso s o uel classi ica ion. Remo e Sens. 2021,13, 342 3 o 20 The esul s o his s udy a e impo an o ope a ional o es y as could imply a conside able dec ease in he human and economic esou ces in es ed in he ield su eys conduc ed o o es a iable mapping. 2. Ma e ial 2.1. S udy A ea The s udy a ea is loca ed in he cen al pa o Eb o alley (41 ◦ 50’ N, 0 º 57’ W) ( Figu e 1 ), no heas o Spain. The o es unde s udy co esponds o monospeci ic s ands o Pinus halepensis Mill. agmen ed in s ands o a iable sizes and occupies app oxima ely 8000 ha. In some a eas, Aleppo pine o es is in e spe sed wi h e e g een sh ubs, domina ed by Que cus cocci e a L., Junipe us oxyced us L. subsp. mac oca pa (Sib h. & Sm.) Ball and Thymnus ulga is L. Pa o he s udy a ea is loca ed inside he Mili a y T aining Cen e (CENAD) “San G ego io,” in ol ing a di ec isk o i e [34]. Remo e Sens. 2021, 13, x FOR PEER REVIEW 4 o 20 Figu e 1. S udy a ea wi h he loca ion o ield plo s. The high spa ial esolu ion o hopho og aphy is p o ided by he Spanish Na ional Plan o Ae ial O hopho og aphy (PNOA). Table 1. Technical speci ica ions o Ai bo ne Lase Scanning (ALS) da a. RMSEz s ands o Roo Mean Squa e E o in heigh , mJ s ands o millijoule, e s ands o Eule numbe , ns s ands o na- noseconds, kHz s ands o kilohe z. Cha ac e is ics Yea 2011 Yea 2016 Pulse epe i ion equency ~ 70 kHz 176-286 kHz Scanning equency ~ 45 kHz 28-59 Hz Maximum scan angle 29° 25° Nominal poin densi y 0.5 poin s m-2 1 poin s m-2 A e age poin densi y 0.64 poin s m-2 1.25 poin s m-2 Accu acy o he poin cloud (RMSEz) ≤ 0.2 m 0.09 m Beam diame e (1/e and 1/e2, mm) 5.6, 8.0 6.2 Beam di e gence (1/e and 1/e2, mm) 0.15, 0.22 0.23 Pulse wid h (ns) 9 3 Maximum ene gy in a single pulse (mJ) 0.2 0.5 2.2.2. Field Da a In si u da a measu ed om July o Sep embe 2014 in 104 ield plo s we e used o adjus g ound- u h o uel ype simula ions. A s a i ied andom sampling echnique was applied o de ine he ield plo loca ion ensu ing ha i co e s he ange o e ain slopes and ege a ion co e wi hin he s udy a ea, which we e es ima ed h ough ALS da a. The cen oid o he ci cula plo s (15 m adius) was posi ioned in he ield using a Leica VIVA GS15 CS10 GNSS eal- ime kinema ic Global Posi ioning Sys em wi h an a e age accu acy o he planime ic coo dina es o 0.33 m. The o al ee heigh (h) and he g een c own heigh we e measu ed in all ees wi h a diame e a b eas heigh (dbh) highe han 7.5 cm using a Ve ex ins umen o p ecise heigh measu emen (Haglö Sweden). T ee di- ame e s we e measu ed a b eas heigh a he s anda d heigh o 1.3 m, using a Man ax P ecision Blue diame e calipe (Haglö Sweden). Addi ionally, he a e age heigh o he di e en sh ub species and hei co e age pe cen age a di e en heigh le els wi h e- spec o he plo su ace we e collec ed. Fuel ype was assigned using he P ome heus classi ica ion [36] ha is based on he ype, heigh and co e age pe cen age o he p opaga ion elemen s [8]. This classi ica ion Figu e 1. S udy a ea wi h he loca ion o ield plo s. The high spa ial esolu ion o hopho og aphy is p o ided by he Spanish Na ional Plan o Ae ial O hopho og aphy (PNOA). Aleppo pine o es s play an impo an ole in he p o ec ion and eco e y o o es in Medi e anean egion cha ac e ized by nu ien -poo , gypsi e ous soils, as i is he case o he si e unde s udy, since his species is p ac ically he only one adap ed o he ad e se clima ic and edaphic condi ions o he a ea. The a ea p esen s a hilly opog aphy, wi h al i udes anging om abou 400 m o 750 m a.s.l. The clima e o he egion is Medi e anean wi h con inen al ea u es, cha ac e ized by i egula annual p ecipi a ion, cold win e s and ho and d y summe s [35]. 2.2. Da ase s 2.2.1. ALS Da a The ALS da a o simula ing and alida ing he model is eely p o ided by he Spanish Na ional Plan o Ae ial O hopho og aphy (PNOA) h ough he Na ional Cen e o Geog aphic In o ma ion (CNIG) (h p://cen odedesca gas.cnig.es). The ALS da a we e acqui ed in 2011 and 2016 wi h wo sligh ly di e en acquisi ion speci ica ions. The i s co e age was cap u ed in se e al su eys conduc ed be ween Janua y and Feb ua y 2011 wi h a Leica ALS60 senso . The second campaign was conduc ed be ween Sep embe Remo e Sens. 2021,13, 342 4 o 20 and No embe 2016 wi h a Leica ALS80 senso . Bo h senso s can eco d up o ou e u ns pe pulse and ope a e a a wa eleng h o 1064 nm. Da a a e deli e ed in 2 km × 2 km iles o classi ied poin s in LAS bina y ile, o ma . 1.2, wi h coo dina e sys em in Uni e sal T ans e sal Me ca o (UTM) uni s, Zone 30, da um Eu opean Te es ial Re e ence Sys em 1989 (ETRS 1989), (EPSG 25830). The lying heigh o he i s and second ALS campaigns we e a ound 3000 and 3150 m abo e g ound le el. De ailed in o ma ion on he espec i e acquisi ion speci ica ions a e shown in Table 1. Table 1. Technical speci ica ions o Ai bo ne Lase Scanning (ALS) da a. RMSEz s ands o Roo Mean Squa e E o in heigh , mJ s ands o millijoule, e s ands o Eule numbe , ns s ands o nanoseconds, kHz s ands o kilohe z. Cha ac e is ics Yea 2011 Yea 2016 Pulse epe i ion equency ~ 70 kHz 176–286 kHz Scanning equency ~ 45 kHz 28–59 Hz Maximum scan angle 29◦25◦ Nominal poin densi y 0.5 poin s m−21 poin s m−2 A e age poin densi y 0.64 poin s m−21.25 poin s m−2 Accu acy o he poin cloud (RMSEz) ≤0.2 m 0.09 m Beam diame e (1/e and 1/e2, mm) 5.6, 8.0 6.2 Beam di e gence (1/e and 1/e2, mm) 0.15, 0.22 0.23 Pulse wid h (ns) 9 3 Maximum ene gy in a single pulse (mJ) 0.2 0.5 2.2.2. Field Da a In si u da a measu ed om July o Sep embe 2014 in 104 ield plo s we e used o adjus g ound- u h o uel ype simula ions. A s a i ied andom sampling echnique was applied o de ine he ield plo loca ion ensu ing ha i co e s he ange o e ain slopes and ege a ion co e wi hin he s udy a ea, which we e es ima ed h ough ALS da a. The cen oid o he ci cula plo s (15 m adius) was posi ioned in he ield using a Leica VIVA GS15 CS10 GNSS eal- ime kinema ic Global Posi ioning Sys em wi h an a e age accu acy o he planime ic coo dina es o 0.33 m. The o al ee heigh (h) and he g een c own heigh we e measu ed in all ees wi h a diame e a b eas heigh (dbh) highe han 7.5 cm using a Ve ex ins umen o p ecise heigh measu emen (Haglö Sweden). T ee diame e s we e measu ed a b eas heigh a he s anda d heigh o 1.3 m, using a Man ax P ecision Blue diame e calipe (Haglö Sweden). Addi ionally, he a e age heigh o he di e en sh ub species and hei co e age pe cen age a di e en heigh le els wi h espec o he plo su ace we e collec ed. Fuel ype was assigned using he P ome heus classi ica ion [ 36 ] ha is based on he ype, heigh and co e age pe cen age o he p opaga ion elemen s [ 8 ]. This classi ica ion comp ises se en ca ego ies (Figu e 2): one g ass co e , h ee sh ub co e s wi h di e en mean heigh s (0 o 0.6 m, 0.6 o 2 m, 2 o 4 m) and h ee di e en ee co e s (wi h no unde s o y, wi h small unde s o y, wi h unde s o y connec ed o he base o he canopy). Top-o -canopy e lec ance measu emen s we e acqui ed du ing he same ield cam- paign using an Analy ical Spec al De ices spec ome e (ASD FieldSpec 4 SR) in he 400 − 2500 nm spec al ange (spec al esolu ion o 3–10 nm a Full Wid h a Hal Maxi- mum (FWHM) and a sampling in e al o 1 nm). Re lec ance was calib a ed using a whi e Spec alon panel (Labsphe e Inc., No h Su on, NH, USA) egis e ed be o e e e y sample measu emen . O icial p ocedu es o ield spec ome y we e applied o gua an ee he quali y o acquisi ions [ 37 , 38 ] acco ding o illumina ion condi ions (clea days and close o he sola noon) and imp o emen o signal- o-noise a io by sub ac ion o he da k cu en signal and spec um a e age (25 measu emen s each). As a esul , we ob ained a o al o 330 absolu e e lec ance spec a (wi h an a e age alue o 5–10 di e en spec a o each species). Remo e Sens. 2021,13, 342 5 o 20 Remo e Sens. 2021, 13, x FOR PEER REVIEW 5 o 20 comp ises se en ca ego ies (Figu e 2): one g ass co e , h ee sh ub co e s wi h di e en mean heigh s (0 o 0.6 m, 0.6 o 2 m, 2 o 4 m) and h ee di e en ee co e s (wi h no unde s o y, wi h small unde s o y, wi h unde s o y connec ed o he base o he canopy). Top-o -canopy e lec ance measu emen s we e acqui ed du ing he same ield cam- paign using an Analy ical Spec al De ices spec ome e (ASD FieldSpec 4 SR) in he 400- 2,500 nm spec al ange (spec al esolu ion o 3-10 nm a Full Wid h a Hal Maximum (FWHM) and a sampling in e al o 1 nm). Re lec ance was calib a ed using a whi e Spec- alon panel (Labsphe e Inc., No h Su on, NH, USA) egis e ed be o e e e y sample measu emen . O icial p ocedu es o ield spec ome y we e applied o gua an ee he quali y o acquisi ions [37,38] acco ding o illumina ion condi ions (clea days and close o he sola noon) and imp o emen o signal- o-noise a io by sub ac ion o he da k cu en signal and spec um a e age (25 measu emen s each). As a esul , we ob ained a o al o 330 absolu e e lec ance spec a (wi h an a e age alue o 5-10 di e en spec a o each species). Figu e 2. P ome heus uel ypes. Pho og aphs aken by Col. Esc ibano and A. Mon ealeg e. 3. Me hods 3.1. Simula ion in DART Model The scene gene a ed in DART ep esen s a plo wi h a la su ace o 30 × 30 m in o de o esemble he ield plo a ea (15 m adius). In DART, o cons uc ing he scenes, Figu e 2. P ome heus uel ypes. Pho og aphs aken by Col. Esc ibano and A. Mon ealeg e. 3. Me hods 3.1. Simula ion in DART Model The scene gene a ed in DART ep esen s a plo wi h a la su ace o 30 × 30 m in o de o esemble he ield plo a ea (15 m adius). In DART, o cons uc ing he scenes, di e en elemen s a e p o ided such as plo s wi h di e en cha ac e is ics (soil, soil+ ege a ion and ege a ion) and ees. Fo mo e in o ma ion on he scene componen s see Lamelas e al. [ 24 ] and Robe s e al. [ 25 ]. These elemen s a e c ea ed using oxels ha can be illed using u bid medium o ace s. In ou case a oxel o 0.5 m size was selec ed o ep esen he gene al scene, he plo s we e illed wi h u bid medium and he ace s we e used o he ee s a um. A oxel size o 1 m and he use o oxels ins ead o ace s in ees we e also es ed ob aining wo s esul s. Di e en pa ame e s a e equi ed o gene a e he plo s and ees, ela ed mainly o hei size and s uc u e, lea a ea index (LAI), e lec ance and ansmi ance alues. The simula ion o he g asslands and sh ublands was pe o med using plo s and adjus ed o he co e age pe cen age and heigh measu ed in he ield in he case o hose species wi h co e age pe cen ages g ea e han 4% o he plo a ea, o he wise his su ace was assigned o a species wi h simila cha ac e is ics and heigh . The ees we e pa ame- e ized using he mean and s anda d de ia ion o he plo in o ma ion. The shape o he Remo e Sens. 2021,13, 342 6 o 20 c own was selec ed om uncal o ellipsoidal acco ding o he species o be ep esen ed, ha is, uncal o genus Pinus and ellipsoidal o Que cus. I should be men ioned ha he c own wid h a iable had o be es ima ed om he diame e and heigh o he ees ollowing he equa ions de eloped by Condés and S e ba [39]. The plo s and he ees loca ion ollow a andom spa ial dis ibu ion since in o ma ion on hei loca ion was no a ailable. Howe e , his was no conside ed a handicap since he me hodology ollowed uses he e ical dis ibu ion in heigh o he e u ns and no hei ho izon al dis ibu ion. The simula ion equi es he LAI alues o he di e en species. In absence o ield da a, he LAI was es ima ed om wo Sen inel 2-A op o canopy no malized e lec ance da a scenes (Le el-2A images) using he Biophysical P ocesso ool in eg a ed in he SNAP so wa e [ 40 ]. The i s image was cap u ed on Janua y 12 h, 2016 and he second one on Oc obe 21s , 2016 (Table 2). Pu e pixels, cha ac e is ic o he di e en co e s, we e selec ed o ex ac he LAI alue om he Le el-2B Biophysical p oduc . The ime lapse in yea s om he simula ion (2011) o he i s a ailable scene (2016) was no conside ed a handicap since he LAI alue a ies wi h a seasonal pa e n and his was co e ed. Table 2. Lea a ea index (LAI) alues assigned o he wo LiDAR-PNOA cap u es. Type o Land Co e Simula ion 1s Cap u e Simula ion 2nd Cap u e G assland 0.15 0.26 Low Bush 0.23 0.29 Medium sh ubs 0.35 0.51 High Bush 0.73 1.01 Pine ees 0.85 1.14 In absence o labo a o y e lec ance and ansmi ance in o ma ion, he alues o some land co e s we e p o ided by D . Ma iano Ga cía (Uni e si y o Alcalá) and D . Olga Rose o (Uni e si y o Za agoza) (pe sonal communica ion) (see Table 3). In he case o ege a ion, e lec ance was measu ed in labo a o y condi ions wi h a lea clamp, while ansmi ance alues we e es ima ed using PROPECT and LIBERTY, adjus ing hem o he cu e o he measu ed e lec ance. La e , he alues o hese la ge g oups we e assigned o he di e en species loca ed in he s udy a ea, aking as e e ence he e lec ance measu ed in he ield. In addi ion, in all scenes i is equi ed o include he e lec ance and ansmi - ance o he e ain o ba e soil. In his case, he e lec ance alue p o ided was measu ed wi h a con ac p obe [41]. Table 3. Re lec ance and ansmi ance alues assigned o he majo land co e ypes p esen in he plo s. Type o Land Co e Re lec ance T ansmi ance Holm oak 0.52 0.35 Pine 0.59 0.25 Soil 0.40 0 G assland 0.27 0 Wood 0.28 0 Table 4p esen s he ALS senso pa ame e s en e ed in he DART model, adjus ed o he eal cap u e. As some o he pa ame e s we e no p o ided by he senso de elope (e.g., a ea o LiDAR-PNOA senso ), hey we e sugges ed by D . Tiangang Yin, de elope o DART, based on his expe ise. Remo e Sens. 2021,13, 342 7 o 20 Table 4. Pa ame e s assigned in Disc e e Aniso opic Radia i e T ans e (DART) o simula e each ALS-PNOA. Senso Pa ame e s Simula ion 1s Cap u e Simula ion 2nd Cap u e LiDAR mode Image (mul iple click) Image (mul iple click) LiDAR Type Disc e e Re u n Disc e e Re u n Minimum Ta ge Re lec ance o de ec ion 0.1 0.1 Numbe o Poin s pe pulse 4 4 A ea o LIDAR senso (m2)0.001 0.001 Diame e o lase beam gene a ed (mm) 5.6 6.2 Lase scanning mode ALS ALS De ini ion o oo p in ange op ion Hal angles Hal angles LIDAR pla o m al i ude (km) 3 3 Pla o m azimu h (◦) 0 0 Swa h wid h (m) 29 29 Look angle (◦) 0 0 G id pa ame e s azimu hal esolu ion (m) 2 1.5 G id pa ame e s - Range esolu ion (m) 2 1.5 Foo p in ( ad) 0.000075 0.000085 Fai h ul o iew ( ad) 0.00009 0.000095 Ene gy o each pulse (mj) 0.2 0.5 Hal pulse du a ion (e ec i e) 3 3 Pulse ela i e powe 0.5 0.5 Hal pulse du a ion a ela i e powe (ns) 8 2 Pho ons numbe (kHz) 1000 1000 F ac ion o pho ons a LiDAR adius 0.368 0.368 LiDAR acquisi ion a e (pe iod) 2 2 3.2. P ocessing o ALS-PNOA-Da a and Simula ed DART Poin Clouds The i s p ocessing s ep o ALS–PNOA da a was noise emo al. Then, g ound poin s we e classi ied using he mul iscale cu a u e classi ica ion algo i hm [ 42 ], implemen ed in he MCC 2.1 command-line ool, acco ding o Mon ealeg e e al. [ 43 ]. The Poin -TIN-Ras e in e pola ion me hod [ 44 ], implemen ed in A cGIS 10.5 so wa e (ESRI, Redlands, CA, USA), was applied o he g ound poin s o p oduce a digi al ele a ion model (DEM) wi h a 1-m g id size, ollowing Mon ealeg e e al. [ 45 ]. The g ound ele a ion alue o he DEM was sub ac ed om he ALS poin heigh in o de o ob ain he no malized heigh s using FUSION LDV 3.30 open sou ce so wa e [46]. DART poin clouds a e s o ed in *. x ile o ma . The simula ed poin clouds we e con e ed in o LAS o ma using he “ x 2las” ool a ailable in LAS ools so wa e (h ps: // apidlasso.com/las ools/). No maliza ion o he da a was no equi ed since he heigh s we e e e ed o he g ound le el. Bo h poin clouds, simula ed by DART and acqui ed by PNOA, we e clipped o i he 15 m adius o he ield plo s. A se ies o s a is ical me ics, commonly used as independen a iables in o es y, we e compu ed using he poin clouds o he simula ed da a and he ALS-PNOA, o de- sc ibe he canopy heigh , canopy heigh a iabili y and canopy densi y. Fu he mo e, h ee di e si y indexes we e also compu ed. Canopy heigh me ics (CHM) include pe cen iles a di e en in e als (P01, P05, P10, P20, P30, P40, P50, P60, P70, P75, P80, P90, P95, P99), minimum, maximum, median, mode (Ele . min, Ele . max, Ele . mean, Ele . mode) ele a ion, quad a ic and cubic ele a ion (Ele . SQRT mean SQ, Ele . CUR mean CUBE) and L momen s (Ele . L1, Ele . L2, Ele . L3, Ele . L4). Canopy heigh a iabili y compu ed me ics (CHVM) include s anda d de ia ion (Ele . SD), a iance (Ele . a iance), coe icien o a ia ion (Ele .CV), in e qua ile dis ance (Ele .IQ), skewness (Ele . skewness) and ku osis (Ele . ku osis). Canopy densi y me ics (CDM) include canopy elie a io (CRR), pe cen age o i s o all e u ns abo e g ound, he mean o he mode (e.g.: % i s e . Abo e mean), he a io o all e u ns espec o he numbe o o al e u ns (e.g.: (All e . Abo e g ound)/( o al i s e .) by 100). The pe cen age o all e u ns using P ome heus anges: 0–0.6 m; 0.6–2 m; 2–4 m and abo e 4 m (e.g.: P op. 2_4) we e compu ed. Fu he mo e, di e en s a is ics ela ed o s a a heigh s (0.5 m, 0.6 m, 1 m, 1.5 m, 2 m, 2.5 m, 3 m, 3.5 m, 4 m, 4.5 m, 5 m and abo e 5 m) we e de i ed (i.e.: e u n p opo ion, min, max, mean, s anda d de ia ion) using he “s a a” swi ch in FUSION “Cloud Me ics” ool. Remo e Sens. 2021,13, 342 8 o 20 Addi ionally, h ee s uc u al di e si y indices (DI) we e compu ed. The Foliage Heigh Di e si y Index o also called LiDAR heigh di e si y index (LHDI) [ 47 ], Equa ion (1), which is an adap a ion o he Shannon (H ´ ) index, he LiDAR heigh e enness index (LHEI) p oposed by Lis opad e al. [ 47 ], Equa ion (2), ha adap s he Pielou (J ´ ) index and Rumple index [48] as a measu e o oughness o s uc u al he e ogenei y, Equa ion (3). LHDI =−∑[(ph)×ln(ph)] (1) LHEI =LHDI ln(ph)(2) Rumple =3D canopy su ace model a ea g ound a ea , (3) whe e pis he p opo ion o e u ns a egula in e als o 0.5 m o a de ined P ome heus classi ica ion heigh in e als (h). The i s s ep o compu e LHDI and LHEI was he calcula ion o e u n p opo ion a di e en heigh in e als using he “s a a” swi ch wi hin he “Cloud Me ics” command o FUSION. Thus, egula in e als o 0.5 m we e selec ed acco ding o Lis opad e al. [47]. Rumple index is he a io o h ee-dimensional canopy su ace model (CSM) o g ound a ea. Rumple was compu ed as he a io be ween he sum o he h ee-dimensional a ea o iangles om CSM g id poin s o he wo-dimensional a ea o he g id cell su ace. The CSMs we e c ea ed o each plo using a 1.5 m pixel and a 3 × 3 smoo hing algo i hm, conside ing poin cloud densi y o ALS-PNOA. The su ace a ea o each CSM 1.5 m pixel is compu ed by c ea ing iangles ha i he cen oid o he pixel and hose o he neighbo ing ones. CSMs we e c ea ed using he highes e u ns o each heigh ange o accoun o canopy oughness. Rumple was compu ed o P ome heus classi ica ion heigh anges (0.6, 2 and 4) o cha ac e ize he e ogenei y wi hin each s a um and o he o e all o es canopy. The h ee di e si y indexes we e gene a ed in R en i onmen o bo h, simula ed and ALS-PNOA da a. 3.3. Accu acy Assessmen o DART Simula ions The accu acy assessmen o DART simula ions was pe o med by compa ing he simula ed da a wi h ALS-PNOA da a in he 104 ield plo s. A se ies o s a is ics p e i- ously calcula ed, ha commonly used in o es y ha desc ibes he canopy heigh (CHM), canopy heigh a iabili y (CHVM), canopy densi y (CDM) and h ee di e si y indexes we e compa ed using he Spea man co ela ion coe icien . The Spea man co ela ion coe icien anges om − 1 o 1, alues closes o 1 indica es highes posi i e co ela ion, alues closes o − 1 indica e highes nega i e co ela ions and alues closes o 0 indica es null co ela ion [49]. 3.4. Fuel Type Classi ica ion As men ioned in Sec ion 2.2.2. Field da a, uel ype was assigned o he ield plo s using he P ome heus classi ica ion ha is based on he ype, heigh and co e age pe cen age o he i e p opaga ion elemen s. These uel ypes we e assigned o he poin clouds om bo h PNOA cap u es clipped o he ield plo ex ension and he co esponding simula ions. The mos explana o y LiDAR simula ed me ics o uel model disc imina ion, we e selec ed using wo selec ion me hods acco ding o Domingo e al. [ 50 ]: (i) Spea man ank co ela ion selec ion me hod; (ii) all subse selec ion, conside ing ou di e en app oaches: comp ehensi e, o wa d, backwa d and sequen ial eplacemen . Me ic selec ion was pe o med independen ly o 2011 and 2016. All subse selec ion de e mines he bes a iables o a g oup, wi hou conside ing he es o he a iables [ 51 ]. Fou sea ching echniques we e es ed: exhaus i e, backwa d, o wa d and sequen ial eplacemen (seq ep). The maximum size o subse s was se o 6, while es s we e pe o med be ween 4 o 6 subse s. Spea man’s co ela ion and all subse Remo e Sens. 2021,13, 342 9 o 20 selec ion we e compu ed wi hin R en i onmen . R package “leaps” and speci ically he “ egsubse s” unc ion was applied o all subse selec ion. Fuel ype classi ica ion was pe o med o 2011 and 2016 using he Suppo Vec o Machine (SVM) a i icial in elligence me hod [52] and including he mos sui able LiDAR simula ed me ics de e mined by he selec ion me hods. SVM me hod allows mul iclass classi ica ion assigning each class o he one wi h highe p obabili y. In his sense, he “C- classi ica ion” was selec ed using he “e1071” package in R en i onmen . The classi ica ion was ained and pa ame e ized using he simula ed DART me ics, being alida ed wi h he me ics ob ained om ALS-PNOA. An SVM is a supe ised lea ning algo i hm ha allows pa e n ecogni ion and is based on he hypo hesis ha da a a e sepa able in o classes in space, ying o ind he op imal sepa a ion be ween classes h ough mul idimensional hype planes. The da a loca ed in he hype planes a e called suppo ec o s, being hese he mos complex o classi y, since he e is less sepa abili y be ween classes. The SVM models wi h adial ke nel and linea ke nel we e gene a ed. The cos and gamma pa ame e s we e pa ame e ized using he in e als 1–1000 and 0.01–1, espec i ely, in acco dance wi h Domingo e al. [ 50 ]. The classi ica ion o e all accu acy, con usion ma ices, use ’s and p oduce ’s accu acy o he i ing and alida ion phases we e e alua ed o compa e and, subsequen ly, de e mine he bes classi ica ion model [53]. 4. Resul s 4.1. Accu acy Assessmen o DART Simula ions. The Spea man ´ s co ela ion coe icien s be ween he simula ed poin cloud me ics and ALS- PNOA o bo h yea s, 2011 and 2016, shows an a e age alue o 0.55 in 2011 and 0.50 in 2016. The co ela ion exceeded an absolu e alue o 0.80 in 32% o he me ics o he yea 2011 and 28% o he yea 2016, while 26% o me ics eached absolu e alues lowe han 0.3 in 2011 and 32% in 2016, espec i ely. The me ics wi h co ela ion coe icien s highe han 0.80 o bo h, 2011 and 2016 yea s, a e included in Table 6(see Table 1in Appendix A o all co ela ion alues). The highes co ela ions a e associa ed o high heigh pe cen iles and o CHVM me ics such as s anda d de ia ion and a iance o bo h 2011 and 2016. Fu he mo e, se e al me ics om CDM and DI each alues o e 0.90, as o example mean abo e_4 o Rumple. CHM ela ed wi h lowe heigh s, as o example low pe cen iles o minimum heigh , p esen lowe co ela ion han CHM high heigh me ics o bo h 2011 and 2016 ALS-PNOA cap u es. A simila end is obse ed o CDM. Me ics ela ed o lowe s a a show lowe co ela ion han hose om highe s a a. Di e si y indices (DI) show high co ela ion alues, while LHEI p esen lowe alues close o 0.75 o bo h yea s. Table 5. Spea man’s co ela ion coe icien s be ween simula ed in DART and ALS-PNOA poin clouds. All he a iables a e signi ican a he 0.05 le el. Me ic Co ela ion Coe icien s 2011 Co ela ion Coe icien s 2016 Canopy heigh me ics (CHM) P60 0.89 0.84 P70 0.93 0.88 P75 0.93 0.92 P80 0.93 0.91 P95 0.94 0.96 P99 0.97 0.97 * Ele . mean 0.93 0.92 Ele . maximum 0.93 0.97 Ele . SQRT mean SQ 0.93 0.95 Ele . CURT mean CUBE 0.94 0.96 Ele .L1 0.93 0.92 Ele .L2 0.95 0.96 Remo e Sens. 2021,13, 342 16 o 20 Table 1. Spea man’s co ela ion coe icien s be ween simula ed DART and ALS-PNOA poin clouds. Me ic Co ela ion Coe icien s 2011 Co ela ion Coe icien s 2016 Canopy heigh me ics (CHM) P01 −0.35 −0.11 P05 −0.36 −0.13 P10 −0.31 −0.13 P20 −0.15 0.15 P25 0.08 0.26 P30 0.21 0.38 P40 0.56 0.53 P50 0.82 0.72 P60 0.89 0.84 P70 0.93 0.88 P75 0.93 0.92 P80 0.93 0.91 P90 0.93 0.94 P95 0.94 0.96 P99 0.93 0.97 To al. e .coun 0.54 0.14 Ele .min −0.26 0.03 Ele .max 0.93 0.97 Ele .mean 0.93 0.92 Ele .mode 0.12 0.29 Ele .SQRT.mean.SQ 0.93 0.95 Ele .CURT.mean.CUBE 0.93 0.96 Fi s . e .abo e.mean 0.41 0.33 Fi s . e .abo e.mode 0.47 0.14 All. e s.abo e.mean 0.41 0.34 All. e .abo e.mode 0.54 0.27 To al. i s . e . 0.46 −0.03 To al.all. e . 0.54 0.14 Ele .L1 0.93 0.92 Ele .L2 0.95 0.96 Ele .L3 −0.25 0.30 Ele .L4 0.13 0.52 Canopy heigh a iabili y me ics (CHVM) Ele s .de . 0.95 0.97 Ele . a iance 0.95 0.97 Ele .CV −0.25 0.35 Ele .IQ 0.92 0.92 Ele .skewness −0.16 0.31 Ele .ku osis 0.28 0.54 Ele .AAD 0.94 0.96 Ele .MAD.median 0.85 0.81 Ele .MAD.mode 0.89 0.85 Ele .L.CV −0.17 0.39 Ele .L.skewness −0.16 0.33 Ele .L.ku osis 0.26 0.52 CRR −0.25 0.02 % all e . Abo e 0 −0.23 −0.19 X.All. e .abo e.0/To al. i s . e .100 −0.25 −0.13 Fi s . e .abo e.0 0.48 −0.02 All. e .abo e.0 0.56 0.15 %. i s . e .abo e.mean −0.07 0.35 %. i s . e .abo e.mode 0.31 0.19 %.all. e .abo e.mean −0.09 0.35 Remo e Sens. 2021,13, 342 17 o 20 Table 1. Con . Me ic Co ela ion Coe icien s 2011 Co ela ion Coe icien s 2016 Canopy densi y me ics (CDM) % i s e . Abo e 0 −0.24 −0.19 %.all. e .abo e.mode 0.27 0.19 X.All. e .abo e.mean/To al. i s . e .100 −0.07 0.35 X.All. e .abo e.mode/To al. i s . e .100 0.41 0.24 o al. e .coun 0_0.6 −0.16 0.47 P op. 0_0.6 0.85 0.80 Mean 0_0.6 −0.23 −0.03 Max 0_0.6 0.32 0.29 Mean 0_0.6 0.30 0.36 Mode 0_0.6 0.05 −0.04 Median 0_0.6 0.38 0.21 s .de 0_0.6 0.41 0.48 CV 0_0.6 0.48 0.15 Skewness 0_0.6 0.21 0.09 Ku osis 0_0.6 0.03 −0.02 o al. e .coun 0.6_2 0.44 0.49 P op. 0.6_2 0.47 0.56 Min 0.6_2 0.39 0.27 Max 0.6_2 0.54 0.54 Mean 0.6_2 0.57 0.64 Mode 0.6_2 0.59 0.37 Median 0.6_2 0.57 0.59 S .de . 0.6_2 0.32 0.54 CV 0.6_2 0.14 0.38 Skewness 0.6_2 0.07 0.13 Ku osis 0.6_2 0.37 0.14 To al. e .coun 2_4 0.85 0.78 P op. 2_4 0.86 0.80 Min 2_4 0.56 0.43 Max 2_4 0.79 0.80 Mean 2_4 0.69 0.84 Mode 2_4 0.75 0.76 Median 2_4 0.70 0.78 S .de . 2_4 0.60 0.69 CV 2_4 0.61 0.65 Skewness 2_4 0.71 0.47 Ku osis 2_4 0.75 0.67 o al. e .coun abo e_4 0.93 0.93 P op abo e_4 0.93 0.94 Min abo e_4 0.73 0.76 Max abo e_4 0.90 0.94 Mean abo e_4 0.89 0.94 Mode abo e_4 0.88 0.90 Median abo e_4 0.89 0.93 S .de . abo e_4 0.88 0.94 CV abo e_4 0.87 0.92 Skewness abo e_4 0.72 0.68 Ku osis abo e_4 0.82 0.76 P op. 0_0.5 0.85 0.79 P op.0.5_1.00 0.11 0.33 P op.1.00_1.50 0.34 0.60 P op.1.50_2.00 0.64 0.56 Remo e Sens. 2021,13, 342 18 o 20 Table 1. Con . Me ic Co ela ion Coe icien s 2011 Co ela ion Coe icien s 2016 P op.2.00_2.50 0.67 0.72 P op.2.50_3.00 0.83 0.70 P op.3.00_3.50 0.85 0.80 P op.3.50_4.00 0.85 0.80 P op.4.00_4.50 0.87 0.81 P op.4.50_5.00 0.84 0.88 P op. Abo e_5 0.89 0.92 Di e si y indices (DI) D0 NA NA D1 0.17 0.50 D2 −0.17 0.33 D3 −0.38 0.22 D4 −0.44 0.11 D5 −0.44 0.06 D6 −0.37 −0.05 D7 −0.34 −0.12 D8 −0.29 −0.14 D9 −0.30 −0.16 Lhdi 0.85 0.83 Lhei 0.76 0.75 Rumple 0.94 0.95 Rumple.0_0.6 0.40 0.41 Rumple.0.6_2 0.21 0.10 Rumple.2_4 0.58 0.41 Rumple.4_40 0.75 0.72 Table 2. O e all accu acy o all subsec selec ion selec ed me ics using SVMl classi ica ion me hod. seq ep s ands o sequen ial eplacemen . Yea Me ics App oach Fi ing phase Valida ion 2011 P30+ Ele .CV + P op. 2.5_3 + P op. abo e_4 seq ep and Exhaus i e 0.68 0.54 P30+ Ele . L.CV+ % i s e . Abo e mean Fo wa d 0.62 0.48 P30+ Ele . L.CV+ P op 0.5_1+ Mean abo e_4 Backwa d 0.65 0.59 2016 P95+ P99+ Mean 0_0.6+ P op. 2_4 seq ep 0.71 0.65 P60+ Ele . L4+ Median 0_0.6+ P op. 2_4 Exhaus i e 0.71 0.65 Ele . SQRT mean SQ + Ele . CUR mean CUBE + Mean 0_0.6 + Max. abo e_4 Fo wa d 0.68 0.68 Ele . max + Mean 0_0.6 + Mode 0_0.6 + P op. 2_4 Backwa d 0.73 0.65 Re e ences 1. Me ill, D.F.; Alexande , M.E. Glossa y o Fo es Fi e Managemen Te ms; Na ional Resea ch Council o Canada, Commi ee o Fo es Fi e Managemen : O awa, ON, Canada, 1987. 2. González-De Vega, S.; De las He as, J.; Moya, D. Resilience o Medi e anean e es ial ecosys ems and i e se e i y in semi- a id a eas: Responses o Aleppo pine o es s in he sho , mid and long e m. Sci. To al En i on. 2016 ,573, 1171–1177. [C ossRe ] [PubMed] 3. Chu ieco, E. Ea h Obse a ion o Wildland Fi es in Medi e anean Ecosys ems; Sp inge : Alcaláde Hena es, Spain, 2009; p. 257. 4. Vicen e-Se ano, S.M.; Lasan a Ma ínez, T.; Cuad a , J.M. T ans o maciones en el paisaje del Pi ineo como consecuencia del abandono de las ac i idades económicas adicionales. Pi ineos 2000,155, 111–133. [C ossRe ] 5. A oyo, L.A.; Pascual, C.; Manzane a, J.A. Fi e models and me hods o map uel ypes: The ole o emo e sensing. Fo es Ecol. Manag. 2008,256, 1239–1252. [C ossRe ] 6. Gaja do, J.; Ga cía, M.; Riaño, D. Applica ions o Ai bo ne Lase Scanning in Fo es Fuel Assessmen and Fi e P e en ion. In Fo es y Applica ions o Ai bo ne Lase Scanning Concep s and Case S udies; Mal amo, M., Naesse , E., Vauhkonen, J., Eds.; Sp inge : Do d ech , The Ne he lands, 2014; pp. 439–462. 7. Bu gan, R.E.; Kla e , R.W.; Kla e , J.M. Fuel models and i e po en ial om sa elli e and su ace obse a ions. In . J. Wildland Fi e 1998,8, 159–170. [C ossRe ] 8. Riaño, D.; Chu ieco, E.; Salas, J.; Palacios-O ue a, A.; Bas a ica, A. Gene a ion o Fuel Type Maps om Landsa TM Images and Ancilla y Da a in Medi e anean Ecosys ems. Can. J. Fo . Res. 2002,32, 1301–1315. Remo e Sens. 2021,13, 342 19 o 20 9. Jia, G.J.; Bu ke, I.C.; Goe z, A.F.H.; Kau mann, M.R.; Kindel, B.C. Assessing Spa ial Pa e ns o Fo es Fuel Using AVIRIS Da a. Remo e Sens. En i on. 2006,102, 318–327. [C ossRe ] 10. Lasapona a, R.; Lano e, A.; Pigna i, S. Cha ac e iza ion and Mapping o Fuel Types o he Medi e anean Ecosys ems o Pollino Na ional Pa k in Sou he n I aly by Using Hype spec al MIVIS Da a. Ea h In e ac 2006,10, 1–10. [C ossRe ] 11. Ga cía, M.; Popescu, S.; Riaño, D.; Zhao, K.; Neuenschwande , A.; Agca, M.; Chu ieco, E. Cha ac e iza ion o Canopy Fuels Using ICESa /GLAS Da a. Remo e Sens. En i on. 2012,123, 81–89. 12. He mosilla, T.; Ruiz, L.A.; Kazako a, A.N.; Coops, N.C.; Moskal, L.M. Es ima ion o Fo es S uc u e and Canopy Fuel Pa ame e s om Small-Foo p in Full-Wa e o m LiDAR Da a. In . J. Wildland Fi e 2014,23, 224–233. [C ossRe ] 13. Huesca, M.; Riaño, D.; Us in, S.L. Spec al Mapping Me hods Applied o LiDAR Da a: Applica ion o Fuel Type Mapping. In . J. Appl. Ea h Obs. Geoin . 2019,74, 159–168. [C ossRe ] 14. Ga cía, M.; Riaño, D.; Chu ieco, E.; Salas, F.J.; Danson, M. Mul ispec al and LiDAR Da a Fusion o Fuel Type Mapping Using Suppo Vec o Machine and Decision Rules. Remo e Sens. En i on. 2011,115, 1369–1379. [C ossRe ] 15. E dody, T.L.; Moskal, L.M. Fusion o LiDAR and Image y o Es ima ing Fo es Canopy Fuels. Remo e Sens. En i on. 2010 ,114, 725–737. [C ossRe ] 16. Jakubowksi, M.K.; Guo, Q.; Collins, B.; S ephens, S.; Kelly, M. P edic ing Su ace Fuel Models and Fuel Me ics Using LiDAR and CIR Image y in a Dense, Moun ainous Fo es . Pho og amm. Eng. Remo e Sens. 2013,79, 37–49. [C ossRe ] 17. Ma ino, E.; Ranz, P.; Tomé, J.L.; No iega, M.A.; Es eban, J.; Mad igal, J. Gene a ion o High-Resolu ion Fuel Model Maps om Disc e e Ai bo ne Lase Scanne and Landsa -8 OLI: A Low-Cos and Highly Upda ed Me hodology o La ge A eas. Remo e Sens. En i on. 2016,187, 267–280. [C ossRe ] 18. Mu lu, M.; Popescu, S.; S ipling, C.; Spence , T. Mapping Su ace Fuel Models Using Lida and Mul ispec al Da a Fusion o Fi e Beha io . Remo e Sens. En i on. 2008,112, 274–285. [C ossRe ] 19. Domingo, D.; de la Ri a, J.; Lamelas, M.T.; Ga cía-Ma ín, A.; Iba a, P.; Eche e ía, M.; Ho én, R. Fuel ype classi ica ion using ai bo ne lase scanning and Sen inel 2 da a in medi e anean o es a ec ed by wild i es. Remo e Sens. 2020 ,12, 3660. [C ossRe ] 20. Mal amo, M.; Naesse , E.; Vauhkonen, J. Fo es y Applica ions o Ai bo ne Lase Scanning Concep s and Case S udies; Sp inge : Do d ech , The Ne he lands, 2014. 21. Le sky, M.A.; Cohen, W.B.; Pa ke , G.G.; Ha ding, D.J. Lida Remo e Sensing o Ecosys em S udies. BioScience 2002 ,52, 19–30. [C ossRe ] 22. Vosselman, G.; Maas, H.-G. Ai bo ne and Te es ial Lase Scanning; Whi les Publishing: Dunbea h, UK, 2010; p. 320. 23. Gómez, C.; Alejand o, P.; He mosilla, T.; Mon es, T.; Pascual, C.; Ruiz, L.A.; Ál a ez-Taboada, F.; Tanase, M.A.; Valbuena, R. Remo e sensing o he Spanish o es s in he 21s cen u y: A e iew o ad ances, needs, and oppo uni ies. Fo es Sys . 2019 ,28, e00R1. [C ossRe ] 24. Lamelas, M.T.; Riaño, D.; Us in, S.L. A LiDAR signa u e lib a y simula ed om 3-dimensional Disc e e Aniso opic Ra- dia i e T ans e (DART) model o classi y uel ypes using spec al ma ching algo i hms. GIScience Remo e Sens. 2019 ,56, 988–1023. [C ossRe ] 25. Robe s, O.; Bun ing, P.; Ha dy, A.; McIne ney, D. Sensi i i y Analysis o he DART Model o Fo es Mensu a ion wi h Ai bo ne Lase Scanning. Remo e Sens. 2020,12, 247. [C ossRe ] 26. Ga cía, M.; No h, P.; Viana-So o, A.; S a os, N.E.; Rose e, J.; Ma ín, M.P.; F anquesa, M.; González-Cascón, R.; Riaño, D.; Bece a, J.; e al. E alua ing he po en ial o LiDAR da a o i e damage assessmen : A adia i e ans e model app oach. Remo e Sens. En i on. 2020,247, 111893. [C ossRe ] 27. No h, P.R.J. Th ee-dimensional o es ligh in e ac ion model using a Mon e Ca lo me hod. IEEE T ans. Geosci. Remo e Sens. 1996 , 34, 946–956. [C ossRe ] 28. Gas ellu-E chego y, J.P.; Yin, T.; Lau e , N.; G au, E.; Rubio, J.; Cook, B.D.; Mo on, D.C.; Sun, G. Simula ion o sa elli e, ai bo ne and e es ial LiDAR wi h DART (I): Wa e o m simula ion wi h quasi-Mon e Ca lo ay acing. Remo e Sens. En i on. 2016 ,184, 418–435. [C ossRe ] 29. Rose e, J.; No h, P.R.; Rubio-Gil, J.; Cook, B.; Los, S.; Sua ez, J.; Sun, G.; Ranson, J.; Blai , J.B. E alua ing p ospec s o imp o ed o es pa ame e e ie al om sa elli e LiDAR using a physically-based adia i e ans e model. IEEE J. Sel. Top. Appl. Ea h Obs. Remo e Sens. 2013,6, 45–53. [C ossRe ] 30. Mon esano, P.; Rose e, J.; Sun, G.; No h, P.; Nelson, R.; Dubayah, R.; Ranson, K.; Kha uk, V. The unce ain y o biomass es ima es om modeled ICESa -2 e u ns ac oss a bo eal o es g adien . Remo e Sens. En i on. 2015,158, 95–109. [C ossRe ] 31. Yin, T.; Gas ellu-E chego y, J.P.; No o d, L.K. Recen Ad ances o Modeling Lida Da a using Da and Radiome ic Calib a ion Coe icien om LVIS Wa e o ms Compa ison. In P oceedings o he 2017 IEEE In e na ional Geoscience and Remo e Sensing Symposium (IGARSS), Fo Wo h, TX, USA, 23–28 July 2017; pp. 1461–1464. 32. Rose e, J.; Suá ez, J.; Nelson, R.; Los, S.; Cook, B.; No h, P. Lida Remo e Sensing o Biomass Assessmen . In Remo e Sensing o Biomass: P inciples and Applica ions; Fa oyinbo, T., Ed.; InTech: Rijeka, C oa ia, 2012; pp. 3–26. 33. Mon ealeg e, A.L.; Lamelas, M.T.; Ga cía-Ma ín, A.; de la Ri a, J.; Esc ibano, F. Using low-densi y disc e e Ai bo ne Lase Scanning da a o assess he po en ial ca bon dioxide emission in case o a i e e en in a Medi e anean pine o es . GIScience Remo e Sens. 2017,54, 721–740. [C ossRe ] Remo e Sens. 2021,13, 342 20 o 20 34. MFE50. Mapa Fo es al de España a Escala 1:50.000. Monis e io de la T ansición ecológica y el Re o Demog á ico. A ailable online: h ps://www.mi eco.gob.es/es/biodi e sidad/se icios/banco-da os-na u aleza/in o macion-disponible/m e50.aspx (accessed on 15 July 2020). 35. Vicen e-Se ano, S.M.; Lasan a, T.; G acia, C. A idi ica ion de e mines changes in o es g ow h in Pinus halepensis o es s unde semia id Medi e anean clima e condi ions. Ag ic. Fo . Me eo ol. 2010,150, 614–628. [C ossRe ] 36. P ome heus. Managemen Techniques o Op imiza ion o Supp ession and Minimiza ion o Wild i e E ec s. Sys em Valida ion. Eu opean Commission, DG XII, ENVIR & CLIMATE, Con ac Numbe ENV4-CT98-0716; Eu opean Commission: Luxembou g, 1999. 37. Mil on, E.J. Re iew a icle p inciples o ield spec oscopy. Remo e Sens. 1987,8, 1807–1827. [C ossRe ] 38. McCoy, R.M. Field Me hods in Remo e Sensing; Guil o d P ess: New Yo k, NY, USA, 2005; pp. 42–58. 39. Condés, S.; S e ba, H. De i a ion o compa ible c own wid h equa ions o some impo an ee species o Spain. Fo . Ecol. Manag. 2005,217, 203–218. [C ossRe ] 40. Weiss, M.; Ba e , F. S2ToolBox Le el 2 P oduc s: LAI, FAPAR, FCOVER. A ailable online: h p://s ep.esa.in /docs/ex a/ATBD_ S2ToolBox_L2B_V1.1.pd (accessed on 11 Sep embe 2020). 41. Rose o-Vlaso a, O.; Pé ez-Cabello, F.; Mon o io Llo e ía, R.; Vlasso a, L. Assessmen o labo a o y VIS-NIR-SWIR se ups wi h di e en spec oscopy accesso ies o cha ac e isa ion o soils om wild i e bu ns. Biosys . Eng. 2016,152, 51–67. 42. E ans, J.S.; Hudak, A.T. A Mul iscale Cu a u e Algo i hm o Classi ying Disc e e Re u n LiDAR in Fo es ed En i onmen s. IEEE T ans. Geosci. Remo e Sens. 2007,45, 1029–1038. [C ossRe ] 43. Mon ealeg e, A.L.; Lamelas, M.T.; de la Ri a, J. A compa ison o open sou ce LiDAR il e ing algo i hms in a Medi e anean o es en i onmen . IEEE J. Selec . Top. Appl. Ea h Obse . Remo e Sens. 2015,8, 4072–4085. [C ossRe ] 44. Renslow, M. Manual o Ai bo ne Topog aphic Lida ; The Ame ican Socie y o Pho og amme y and Remo e Sensing: Be hesda, MA, USA, 2013. 45. Mon ealeg e, A.L.; Lamelas, M.T.; de la Ri a, J. In e pola ion Rou ines Assessmen in ALS-De i ed Digi al Ele a ion Models o Fo es y Applica ions. Remo e Sens. 2015,7, 8631–8654. [C ossRe ] 46. McGaughey, R. FUSION/LDV: So wa e o LIDAR Da a Analysis and Visualiza ion; US Depa men o Ag icul u e, Fo es Se ice, Paci ic No hwes Resea ch S a ion: Washing on, DC, USA, 2009. 47. Lis opad, C.M.C.S.; Mas e s, R.E.; D ake, J.; Weishampel, J.; B anquinho, C. S uc u al di e si y indices based on ai bo ne LiDAR as ecological indica o s o managing highly dynamic landscapes. Ecol. Indic. 2015,57, 268–279. [C ossRe ] 48. Kane, V.R.; Bakke , J.D.; McGaughey, R.J.; Lu z, J.A.; Ge sonde, R.F.; F anklin, J.F. Examining coni e canopy s uc u al complexi y ac oss o es ages and ele a ions wi h LiDAR da a. Can. J. Fo . Res. 2010,40, 774–787. [C ossRe ] 49. Ande son, D.R.; Sweeney, D.J.; Williams, T.A. Es adís ica Pa a Adminis ación y Economía; In e na ional Thomson: San a Fe, México, 2001; ISBN 970-686-051-7. 50. Domingo, D.; Alonso, R.; Lamelas, M.T.; Mon ealeg e, A.L.; Rod íguez, F.; de la Ri a, J. Tempo al ans e abili y o pine o es a ibu es modeling using low-densi y ai bo ne lase scanning da a. Remo e Sens. 2019,11, 261. [C ossRe ] 51. Mille , A.J. Subse Selec ion in Reg ession. In Monog aphs on S a is ics and Applied P obabili y 95; Isham, V., Keiding, T., Louis, N., Tibshi ani, R.R., Tong, H., Eds.; Chapman & Hall/CRC: New Yo k, NY, USA, 2002; p. 256. A ailable online: h ps://ncss- wpengine.ne dna-ssl.com/wp-con en / hemes/ncss/pd /P ocedu es/NCSS/Subse _Selec ion_in_Mul iple_Reg ession.pd (ac- cessed on 16 Oc obe 2019). 52. Moun akis, G.; Im, J.; Ogole, C. Suppo ec o machines in emo e sensing: A e iew. ISPRS J. Pho og amm. Remo e Sens. 2011 , 66, 247–259. [C ossRe ] 53. Chu ieco, E. Telede ección Ambien al. La Obse ación de la Tie a Desde el Espacio; A iel Ciencia: Ba celona, Spain, 2010. 54. Chi ici, G.; Sco i, R.; Mon aghi, A.; Ba ba i, A.; Ca isano, R.; Lopez, G.; Ma che i, M.; Mc obe s, R.E.; Olsson, H.; Co ona, P. S ochas ic g adien boos ing classi ica ion ees o o es uel ypes mapping h ough ai bo ne lase scanning and IRS LISS-III image y. In . J. Appl. Ea h Obs. Geoin . 2013,25, 87–97. [C ossRe ] 55. K is ensen, T.; Næsse , E.; Ohlson, M.; Bols ad, P.V.; Kolka, R. Mapping Abo e- and Below-G ound Ca bon Pools in Bo eal Fo es s: The Case o Ai bo ne Lida . PLoS ONE 2015,10, e0138450. [C ossRe ] 56. Valbuena, R.; Mal amo, M.; Meh ä alo, L.; Packalena, P. Key s uc u al ea u es o Bo eal o es s may be de ec ed di ec ly using L-momen s om ai bo ne lida da a. Remo e Sens. En i on. 2017,194, 437–446. [C ossRe ] 57. Gelabe , P.J.; Mon ealeg e, A.L.; Lamelas, M.T.; Domingo, D. Fo es s uc u al di e si y cha ac e iza ion in Medi e anean landscapes a ec ed by i es using Ai bo ne Lase Scanning da a. GIScience Remo e Sens. 2020,57, 497–509. [C ossRe ] 58. Alonso-Beni o, A.; A oyo, L.; A belo, M.; He nández-Leal, P.; Alonso-Beni o, A.; A oyo, L.A.; A belo, M.; He nández-Leal, P. Fusion o Wo ldView-2 and LiDAR Da a o Map Fuel Types in he Cana y Islands. Remo e Sens. 2016,8, 669. [C ossRe ]