In search of the actual groundwater recharge. Project plan

Transcriptie

1 In search of the actual groundwater recharge Project plan BTO (s) May 2011

2 In search of the actual groundwater recharge Project plan BTO (s) May KWR Alle rechten voorbehouden. Niets uit deze uitgave mag worden verveelvoudigd, opgeslagen in een geautomatiseerd gegevensbestand, of openbaar gemaakt, in enige vorm of op enige wijze, hetzij elektronisch, mechanisch, door fotokopieën, opnamen, of enig andere manier, zonder voorafgaande schriftelijke toestemming van de uitgever. Postbus BB Nieuwegein T F E info@kwrwater.nl I

3

4 Colofon Titel In search of the actual groundwater recharge Projectnummer BTO (s) Opdrachtnummers B , O & S Onderzoeksprogramma s PBC Bronnen & CARE Projectmanager Dr. ir. Gé van den Eertwegh Opdrachtgevers BTO & Stichting Kennis voor klimaat Kwaliteitsborger Prof. dr. Pieter Stuyfzand Auteurs Dr. ir. Ruud P. Bartholomeus, Bernard R. Voortman Msc & Prof. dr. ir. Jan-Philip M. (Flip) Witte Verzonden aan Dit rapport is verspreid onder BTO-participanten en is openbaar Foto kaft Medewerkers van team Ecologie onderzoeken in Polen een droog rivierduin langs rivier de Bierbza

5

6 List of symbols Symbol Unit Description a L/T empirical coefficient for calculation of P i b - soil cover fraction for calculation of P i E 0 L/T Penman open water evaporation E a L/T actual soil evaporation E p L/T potential soil evaporation E p_bare L/T potential evaporation of a wet bare soil ET a L/T actual evapotranspiration ET p L/T potential evapotranspiration = E p + T p + I ET p_dry L/T potential evapotranspiration of a crop with a dry canopy (I=0) ET p_wet L/T potential evapotranspiration of a crop with a wet canopy ET ref L/T reference evapotranspiration ET ref_mak L/T reference evapotranspiration according to Makkink ET ref_mak_dry L/T reference evapotranspiration of Makkink but valid for a crop with a dry canopy only ET ref_pm L/T reference evapotranspiration according to Penman-Monteith, i.e. for a crop with dry canopy ET ref_pm_avg L/T ET ref_pm under the real climatic conditions, thus including both wet and dry periods ET ref_pm_dry L/T ET ref_pm f CO2 - correction factor for higher water use efficiency due to increased CO 2 (Kruijt et al., 2008; Witte et al., 2006a; Witte et al., 2006b) f KNMI - correction factor for altered temperature and wind, given by the KNMI 06-scenarios (Van den Hurk et al., 2006) G W/m 2 soil heat flux g I - correction factor for the contribution of I in ET ref_mak g I,c - correction factor for the contribution of I in k c I L/T interception evaporation k c - crop factor according to Feddes, i.e. to be applied on ET ref_mak k c_dry - crop factor to be applied on ET ref_mak_dry LAI L 2 /L 2 leaf area index P i L/T intercepted precipitation P gross L/T gross precipitation r air T/L aerodynamic resistance r crop T/L crop resistance R n W/m 2 net radiation R s W/m 2 global (solar) radiation SC - fractional soil cover T a L/T actual transpiration T air K air temperature T p L/T potential transpiration T soil K soil temperature Κ gr - extinction coefficient for solar radiation In search of the actual groundwater recharge KWR - vi - May 2011

7 In search of the actual groundwater recharge KWR - vii - May 2011

8 Voorwoord In dit rapport beschrijven we onze plannen voor onderzoek naar de werkelijke aanvulling van het grondwatersysteem met water dat vanuit de wortelzone naar beneden percoleert. Die grondwateraanvulling is de drijvende kracht achter de stroming van grondwater. Daarom is een nauwkeurige bepaling van deze post van essentieel belang is voor betrouwbare hydrologische berekeningen en voor een juiste schatting van de hoeveelheid grondwater die kan worden gewonnen zonder schade aan landbouw en natuur te veroorzaken. Op hogere zandgronden waar de wortelzone niet capillair wordt aangevuld vanuit het grondwater, kan de vegetatie reageren op variaties in weersgesteldheid door een aantal aanpassingen van zijn verdampingseigenschappen, zoals van het aandeel kale grond. In kwantitatief opzicht is echter nauwelijks iets over deze aanpassingsmechanismen bekend. Wij hopen met dit onderzoek iets aan dit euvel te kunnen doen. De kennis die we ontwikkelen is van belang voor de bepaling van de grondwateraanvulling onder zowel het huidige klimaat, als onder het klimaat van de toekomst, wanneer de zomers misschien veel droger zijn dan nu en de vegetatie er daardoor heel anders uit ziet. Het onderzoek bestaat uit drie projecten: een project van de drinkwaterbedrijven (BTO), een project uit het fundamenteel onderzoeksprogramma van KWR (FOP), en een promotieproject gesubsidieerd door de stichting Kennis voor Klimaat (KvK). De plannen voor de eerste twee projecten werden begeleid door vertegenwoordigers van de waterbedrijven op het gebied van de hydrologie (Bas Baartmans, Harry Boukes, Jeroen Castelijns, Merel Hoogmoed, Vera Lagendijk, Karel de Mey, Nico van der Moot, Philip Nienhuis, Theo Olsthoorn, Birgitta Putters, Sjaak Rijk, Harry Rolf en Dirk de Smet). Behalve uit Ruud Bartholomeus en Flip Witte bestond de begeleiding van het promotieplan uit wetenschappers van enkele universiteiten (Sjoerd van der Zee, Peter van Bodegom, Marc Bierkens). Wij danken al deze begeleiders hartelijk voor hun commentaar en de vruchtbare uitwisseling van ideeën. Ook zijn wij Fred Daniëls en Mirja Henschel van de universiteit van Münster erkentelijk voor het delen van hun expertise op het gebied van droogteminnende vegetaties met ons. Vermeld moet ook worden Nationaal Park de Hoge Veluwe, dat zeer ruimharig medewerking aan ons heeft verleend en waarvan we, onder leiding van boswachter Evert Cock, een excursie met terreinwagen kregen aangeboden. Ten slotte betuigen wij onze dank aan de waterbedrijven en de stichting Kennis voor klimaat voor het vertrouwen dat wij van hen hebben gekregen om dit interessante onderzoek te mogen uitvoeren. Ruud Bartholomeus, Bernard Voorman, Flip Witte Nieuwegein, maandag 9 mei 2011 In search of the actual groundwater recharge KWR - viii - May 2011

9 In search of the actual groundwater recharge KWR - ix - May 2011

10 Inhoud List of symbols Voorwoord Inhoud vi viii x Inleiding 1 Part I Redefining crop factors (BTO & FOP) 9 1 Introduction Problem definition Research topics and goals BTO and FOP-fund Outline and planned deliverables 12 2 Research topics and products Vegetation cover vs. reference drought stress (step A) Unravel crop factors (step B) Water and heat balance as function of soil and vegetation cover (optional, if time allows) (step C) 23 3 Delivered products Part II PhD proposal Bernard Voortman (KvK) 27 1 Introduction Background and problem definition Research objectives Outline 31 2 Vegetation dynamics in coastal and inland sand dunes Introduction Natural Succession Internal vegetation dynamics Nitrogen deposition 39 3 Research approach and analyses of the available knowledge 41 4 The role of mosses and lichens in the soil-water balance 43 In search of the actual groundwater recharge KWR - x - May 2011

11 4.1 Background and problem definition Research objectives Methods Expected results Timeline 49 5 Modelling plant available water for dune vegetation: the effects of vegetation structure and topography Background and problem definition Research objectives Methods Expected results Timeline 55 6 A conceptual model to estimate the vegetation coverage of dry grassland vegetation of European sandy soils under changing climatic conditions Background and problem definition Research objectives Methods 58 7 The response of vegetation to climate change and its effect on evapotranspiration and groundwater recharge in coastal and inland sand dunes of The Netherlands Background and problem definition Research objectives Methods 61 8 Timetable 63 References 65 In search of the actual groundwater recharge KWR - xi - May 2011

12 Inleiding Hydrologen hebben in de afgelopen decennia prachtige computermodellen ontwikkeld waarmee men de stroming van water in de bodem en de diepere ondergrond kan simuleren en waarmee men het verloop in ruimte en tijd van grootheden als de grondwaterstand en het bodemvochtgehalte kan nabootsen. Ruwweg zijn er twee typen modellen beschikbaar: mechanistische modellen en tijdreeksmodellen. In het eerste type worden de fysische eigenschappen van de ondergrond ruimtelijk expliciet geschematiseerd. Een voorbeeld is het model Modflow (Harbaugh et al., 2000). In tijdreeksmodellen worden gebiedseigenschappen impliciet verwerkt in een beperkt aantal sterk gelumpte parameters waarop men het model ijkt. Een bekend voorbeeld is Menyanthes (Von Asmuth et al., 2002; Von Asmuth et al., 2006). Beide benaderingen, mechanistisch of gelumpt, hebben hun eigen voor- en nadelen die in algemene literatuur over modelleren uitvoerig is bediscussieerd (zie o.a. Baird, 1999; Guisan and Zimmermann, 2000). In beide benaderingen wordt het model gevoed door gegevens over neerslag en verdamping. Bij een fysisch model gaat het doorgaans om de werkelijke verdamping, berekend met een model voor het vochttransport in de bodemzone, zoals Swap (Van Dam et al., 2008), MetaSwap (Schaap and Dik, 2007) en Onzat (Van Drecht, 1996), bij een tijdreeksmodel meestal om de referentieverdamping, zoals verstrekt door het KNMI, maar in principe kan ook de werkelijke verdamping als invoer dienen. De fout in de berekening van de werkelijke verdamping zal, schatten we, in veel gevallen zeker 10% bedragen (zie bijvoorbeeld Tabel 6.3 in Massop et al. (2005)). Onder het Nederlandse klimaat tikt die fout gemiddeld ongeveer twee tot drie keer zo hard door in de berekende grondwateraanvulling (zonder oppervlakteafvoer: neerslag minus werkelijke verdamping). Zoals bekend is de grondwaterstand ten opzichte van de drainagebasis bij benadering rechtevenredig aan de grondwateraanvulling. Dit betekent, op zijn beurt, dat een fout in de verdamping eveneens dubbel terugkomt in de gesimuleerde grondwaterstanden. In de praktijk zien we die fout echter niet omdat men de hydrologische modellen ijkt aan gemeten stijghoogten: men verhoogt of verlaagt bijvoorbeeld het doorlaatvermogen van het eerste watervoerende pakket zodanig, dat de gesimuleerde waarden overeenstemmen met de waarnemingen. Fouten in de verdamping compenseren met een aangepast doorlaatvermogen, daar komt het op neer (bij tijdreeksmodellen wordt voor fouten in de verdamping standaard gecompenseerd via de optimalisering van de modelparameters). De vraag is echter of het model daarmee geschikt is voor extrapolaties, dus voor toepassingen onder andere omstandigheden dan die waarop het model is geijkt (zie kader). Deze vraag klemt des te meer bij klimaatprojecties, omdat de verdampingseigenschappen van de vegetatie door klimaatverandering wel eens zouden kunnen gaan veranderen 1, (Nijssen et al., 2011). Dat het klimaat in Nederland in rap tempo verandert, is met metingen aangetoond: de neerslag is de afgelopen eeuw gemiddeld met bijna 20% toegenomen (Witte et al., 2009) en de temperatuur met 1.7 o C (Anonymus, 2009). Neerslag valt steeds meer in de wintermaanden en in de vorm van buien met een hoge neerslagintensiteit. Voor Nederland publiceerde het KNMI in 2006 vier klimaatscenario s die betrekking hebben op de jaren 2050 en 2100 (Van den Hurk et al., 2006). Deze scenario s verschillen in de mate van verandering, maar hebben gemeen dat de temperatuur en de potentiële verdamping stijgen, dat de hoeveelheid neerslag in de winter stijgt en dat de intensiteit van de neerslagbuien toeneemt. 1 Ter overweging: in het onderzoek naar de winning Ter Wisscha werd een achtergrondsverdroging gevonden van ongeveer 1 cm/jr. Waarschijnlijk wordt die geheel of gedeeltelijk veroorzaakt door veranderende verdampingseigenschappen van de vegetatie. Zo is uit luchtfotoanalyse gebleken dat het Aekingerzand tussen 1950 en 2006 deels is dichtgegroeid, wat tot een afname van 35% in areaal kaal zand heeft geleid; daarvoor in de plaats is voornamelijk bos en heide gekomen. In search of the actual groundwater recharge Inleiding KWR May 2011

13 Volgens het KNMI (mondelinge mededeling Janet Bessembinder) zijn thans het W en W+ scenario s het aannemelijkst, dat wil zeggen de scenario s waarbij de temperatuur in 2050 wereldwijd met 2 o C is gestegen en waarbij de windrichting onveranderd blijft (W), dan wel in de zomer vaker uit het oosten komt (W+). Recent onderzoek duidt er op dat de intensiteit van de buien nog meer toeneemt met een stijging van de temperatuur dan in 2006, toen de klimaatscenario s werden uitgegeven, werd aangenomen (Lenderink and Van Meijgaard, 2008). De verandering van klimaat leidt bovendien tot een andere verdamping van de vegetatie. De potentiële verdamping zal door het opwarmen van de aarde stijgen, maar hoe zit het met de werkelijke verdamping? Kader: hoe het mis kan gaan met een onttrekkingskegel Dat een fout in de grondwateraanvulling substantieel kan doorwerken in de berekende onttrekkingskegel illustreren we hier aan de hand van een analytische formule voor een freatische onttrekking Q (m3/d) midden op een cirkelvormig eiland met straal L (m), grondwateraanvulling R (m/d), doorlatendheid k (m/d) en hoogte drainagebasis ten opzichte van een ondoorlatende basis H (m). Volgens Huisman (1972) geldt voor de hoogte h (m) van de grondwaterspiegel ten opzichte van de ondoorlatende basis: 2 2 R L x 2 2 ln Lx h H Q 2k k Waarin x (m) de afstand tot de winning is. Stel nu dat een hydroloog dit model eerst fit op een situatie zonder winning (Q = 0) met een grondwateraanvulling die een factor f van de werkelijke aanvulling bedraagt (R fout = f R). Na kalibreren vindt hij dan natuurlijk een doorlatendheid gelijk aan k fout = f k. Eenvoudig is in te zien dat die fout daarna, bij een winning Q >0, alleen doorwerkt in de rechterterm van bovenstaande vergelijking. Voorbeeld: stel een hydroloog wil de onttrekkingskegel op een gebied als de Veluwe uitrekenen, met een oppervlak van 1000 km 2 (L = m), een freatische pakket met een dikte H van 100 m ten opzichte van een ondoorlatende basis en een werkelijke doorlatendheid k van 5 m/d. De werkelijke grondwateraanvulling (1 mm/d) heeft hij 30% onderschat (f = 0.7), zodat hij na ijking op de gemeten grondwaterstanden (zonder winning) een doorlatendheid vindt van k fout = 3.5 m/d. Vervolgens gaat hij de kegel berekenen van een onttrekking van 3000 m 3 /d (9.1 Mm 3 /jr). De fout die hij dan maakt is hieronder weergegeven. Op 1.5 km van de winning bedraagt de fout 0.5 m, op 4.1 km afstand 0.3 m en op 7 km is de fout 0.2 m fout (m) x (m) Dit voorbeeld is natuurlijk wat gechargeerd, want verstandige hydrologen ijken hun model niet alleen aan gemeten stijghoogten, maar ook aan gemeten afvoeren. Bovendien hebben ze andere manieren om informatie over de eigenschappen van de ondergrond in te winnen, zoals pomproeven. Niettemin toont het voorbeeld aan dat het van belang is de grondwateraanvulling nauwkeurig te bepalen. In search of the actual groundwater recharge Inleiding KWR May 2011

14 Berekening van de werkelijke verdamping gebeurt in hydrologische modellen als MetaSwap door de vegetatie te schematiseren als een laag met vaste verdampingseigenschappen. Het is aannemelijk dat deze berekening tot aanzienlijke fouten kan leiden bij klimaatprojecties, vooral in natuurgebieden op droge zandgronden zoals stuwwallen en duinen: 1. Verdamping via de huidmondjes (transpiratie) is in de modellen doorgaans een functie van de drukhoogte in de wortelzone en de potentiële transpiratie van de vegetatie (Feddes et al., 1978). Waarschijnlijk kan dit concept overeind blijven staan. 2. Een deel van de neerslag bereikt niet de bodem maar wordt door bladeren opgevangen waarvan het weer verdampt. Onder het huidige klimaat is deze interceptiepost ongeveer 20% van de neerslag bij een kort groen grasland (Witte et al., 2006a) en kan hij wel 45% bedragen bij donker naaldhout als Douglas en Fijnspar (Dolman et al., 2000). Als de neerslag minder gelijkmatig valt, en steeds meer in de vorm van intensieve buien, zal de interceptieverdamping dalen zodat meer neerslagwater de bodem bereikt. In een hydrologische simulatie komt dit effect van klimaatverandering alleen dan voldoende naar voren, wanneer de interceptie expliciet en betrouwbaar wordt gesimuleerd. Door Alterra is veel onderzoek gedaan naar de interceptie van bossen (Dolman et al., 2000), maar er is vrij weinig tot niets bekend over de interceptie van andersoortige natuurlijke vegetaties, zoals de mosrijke steppe op de Veluwe (Figuur 1). 3. De potentiële verdamping van de vegetatie wordt meestal berekend door de Makkink referentiegewasverdamping van het KNMI te vermenigvuldigen met een gewasfactor (Feddes, 1987). Deze gewasfactor is empirisch bepaald: het is de verhouding tussen de met een lysimeter of meetmast gemeten verdamping en de Makkink-verdamping. In de meting van de verdamping is de interceptieterm inbegrepen. Dat betekent dat de hoogte van de gewasfactor afhangt van het neerslagpatroon ten tijde van de metingen. Dit roept de vraag op hoe houdbaar gewasfactoren zijn bij klimaatprojecties, zeker als de neerslag steeds meer in de vorm van hevige buien gaat vallen. 4. Gewasfactoren voor natuurlijke vegetaties zijn nauwelijks onderzocht (Spieksma et al., 1997). 5. Vermenigvuldiging van de Makkink-verdamping met een gewasfactor zou moeten leiden tot de potentiële verdamping van het vegetatietype waarvoor de gewasfactor is bedoeld. De potentiële verdamping is gedefinieerd als de verdamping van een gewas of vegetatietype dat optimaal wordt voorzien van zoet water. Dat is een moeilijk houdbaar concept voor een vegetatie op droge zandgronden. Een dergelijke vegetatie is immers van nature helemaal niet optimaal van water voorzien, terwijl kunstmatige toediening van voldoende water geen oplossing is om de potentiële verdamping te bepalen, want die handeling leidt tot een andere vegetatie, met nieuwe verdampingseigenschappen. Daarom zijn gewasfactoren voor droge vegetaties in de praktijk verhoudingsgetallen tussen de werkelijke verdamping (die bij droogte is gereduceerd) en de Makkink-verdamping. Net als gewasfactoren (punt 3) gelden die verhoudingsgetallen voor de tijdens de meting heersende weersomstandigheden. Bovendien is de veronderstelling achter de hydrologische rekenprocedure dat met die verhoudingsgetallen de potentiële verdamping wordt vastgesteld, onjuist. 6. Door stijging van de CO 2 concentratie gaan planten netto wat zuiniger om met water: ze transpireren minder. Dit effect is waargenomen in historische reeksen van rivierafvoeren (Gedney et al., 2006). Bovendien toont een recente analyse van fossiel bladmateriaal aan dat de dichtheid aan huidmondjes afneemt met een toenemend CO 2 gehalte (De Boer et al., 2011). Door die afname stijgt de stomatale weerstand en daalt de transpiratie. Voor Nederland is geraamd dat in 2050 (ΔCO 2 = 150 ppm) de potentiële verdamping (evapotranspiratie) gemiddeld met 2-5% zal zijn gedaald, afhankelijk van de vegetatiestructuur (Kruijt et al., 2008; Witte et al., 2006a; Witte et al., 2006b). 7. Op hogere zandgronden als de duinen en de Veluwe, nemen het aandeel kale grond en het aandeel niet-wortelende planten (mossen en korstmossen) toe naarmate de zomer droger wordt. Omdat kale grond en (korst)mossen veel minder verdampen dan wortelende vaatplanten, zorgt dit voor een daling van de werkelijke verdamping. Indicatieve berekeningen tonen aan dat in een van de klimaatscenario s (W+), het aandeel kale grond op zuidhellingen van duinen in 2050 kan zijn gestegen van 30% tot meer dan 80% (Witte et al., 2008b). Het samenspel van klimaat, bodem, water In search of the actual groundwater recharge Inleiding KWR May 2011

15 en vegetatie is echter nauwelijks begrepen, laat staan goed gekwantificeerd. Overigens is het effect van droogte in hogere zandgronden op microschaal al zichtbaar: noordhellingen zijn daar meer bedekt en productiever dan zuidhellingen. Al met hebben we ernstige bedenkingen bij de manier waarop de werkelijke verdamping wordt berekend op de hogere zandgronden en bij klimaatprojecties. Dit betekent dat effecten op waterhuishouding en ecosystemen van natuurlijke variaties in het weer, nog niet goed kunnen worden voorspeld. Dit geldt nog meer voor de effecten van klimaatverandering. Een goede berekeningswijze is nodig voor een betrouwbare schatting van de grondwateraanvulling, nu en onder het toekomstige klimaat. In dit licht vinden wij het verbazingwekkend dat tot nu toe zo weinig onderzoek is gedaan naar de verdamping van natuurlijke vegetaties, terwijl van de andere kant veel geld en menskracht is geïnvesteerd in de ontwikkeling van grondwatermodellen. Het is immers de grondwateraanvulling die het hydrologische systeem aandrijft. Alle aandacht naar de motor, maar de brandstof wordt verwaarloosd. Misschien is deze discrepantie te verklaren uit verschillen in achtergrond van onze academici: Delftenaren die zich graag wiskundig uitleven op het verzadigde grondwatersysteem, en Wageningers die vooral geïnteresseerd zijn in gewasproductie en daarom de verdamping van landbouwgewassen onderzoeken. Vooral voor de drinkwaterbedrijven is een betrouwbare berekening van de grondwateraanvulling van belang. Ongeveer tweederde van het leidingwater dat de drinkwaterbedrijven in Nederland produceren, is immers onttrokken grondwater, water dat door grondwateraanvulling is ontstaan. Onttrekkingen op de stuwwallen (Brabantse wal, Veluwe, Sallandse hevelrug, etc.) zijn volledig afhankelijk van wat er vanuit de hemel uiteindelijk doorsijpelt naar het grondwater. Als we het KNMI mogen geloven zal het jaarlijkse neerslagoverschot (de neerslag minus Makkink verdamping) te De Bilt onder het W+ scenario in 2050 met bijna 30% zijn gedaald (Hermans et al., 2009) en in de Amsterdamse Waterleidingduinen met 42% (Witte et al., 2008b). Vooral de zomers worden veel droger. Wat deze verandering betekent voor de werkelijke grondwateraanvulling, valt met de huidige stand van kennis echter niet te zeggen. Zal de aanvulling, net als het neerslagoverschot, op jaarbasis afnemen of, door een aantal Figuur 1. Vegetatie op de Hoge Veluwe, vooral bestaand uit mossen en korstmossen. Een dergelijke steppevegetatie komt op de Veluwe in grote oppervlakten voor en is zeer stabiel (volgens prof. Daniëls minstens twee eeuwen oud). Van de mossen is thans niet bekend hoeveel ze verdampen en of ze capillair water uit de zandondergrond kunnen onttrekken. Misschien is het contact met de ondergrond wel zo los, dat de moslaag hydrologisch geschematiseerd kan worden als een interceptiereservoir. In search of the actual groundwater recharge Inleiding KWR -4- May 2011

16 terugkoppelingsmechanismen van de vegetatie, juist toenemen? We weten het niet. Ook voor de duinbedrijven die water winnen via infiltratie van rivierwater, kan de werkelijke grondwateraanvulling een belangwekkend vraagstuk worden. In de AWD wordt het meeste water gewonnen via infiltratie van rivierwater; slechts ongeveer 15% is afkomstig van de natuurlijke grondwateraanvulling. De verwachting is echter dat door klimaatverandering de afvoer van rivieren in de zomer dramatisch kan gaan dalen. Volgens een onderzoek in opdracht van de Deltacommissie kan de Rijnafvoer in 2050 s zomers tot 35% zijn gedaald en in 2100 zelfs tot 60% (Vellinga et al., 2009). Een dergelijk lage afvoer gaat gepaard met een slechtere waterkwaliteit (van Vliet and Zwolsman, 2008) en met een grotere concurrentie om de vraag naar zoet water door verschillende belanghebbenden. Betwijfeld kan worden of er dan voldoende over blijft voor duininfiltratie. Opslag van water in de ondergrond tijdens de natte wintermaanden (ASR) kan de nood lenigen, maar de vraag is of die maatregel voldoende is om de hele zomer door te komen. Ook pompstations in de duinen worden door veranderde rivierafvoeren meer afhankelijk van de natuurlijke grondwateraanvulling. Ten slotte is een nauwkeurige berekening van de grondwateraanvulling van groot belang voor het berekenen van de kwaliteit van het grondwater. De mate van indikking van opgeloste stoffen in het infiltrerende neerslagwater, de snelheid waarmee stoffen in de bodem omzetten en uitspoelen, de mineralisatiesnelheid van organische stof en respiratiesnelheid van de vegetatie: zij hangen allemaal samen met de bodemtemperatuur en de grondwateraanvulling en de daardoor veroorzaakte dynamiek in de grondwaterspiegel. De waterbedrijven (vertegenwoordigd in de PBC Bronnen) hebben eind 2008 een onderzoeksvoorstel naar de grondwateraanvulling op hogere zandgronden goedgekeurd (projectnummer B ). In dat voorstel werd beloofd te proberen een promovendus aan te stellen met subsidie uit het FESprogramma Kennis voor Klimaat. Dat is gelukt: binnen het KvK-programma Climate Adaptation for Rural areas (CARE; (Anonymus, 2010)) werkt Bernard Voortman sinds 15 juli 2010 als promovendus bij KWR (S ). Bovendien heeft het onderzoek meer body gekregen door subsidie uit het fundamenteel onderzoeksprogramma van KWR, waardoor vanaf november 2009 Ruud Bartholomeus voor twee jaar als postdoc aan het onderwerp kan werken (O710023). In het onderzoek werken we nauw samen met prof. Sjoerd van der Zee (Wageningen universiteit), dr. ir. Peter van Bodegom (Vrije universiteit) en prof. dr. ir. Marc Bierkens (universiteit Utrecht), onder andere voor de begeleiding van Bernard Voortman. Bovendien hebben we onlangs samenwerking gezocht met prof. dr. Fred Daniëls en promovenda Mirja Henschel (universiteit Münster), vegetatiekundigen met grote kennis van droge graslanden. Fred Daniëls doet al sinds 1980 onderzoek op de Veluwe, onder andere via herhaalde vegetatieopnamen op dezelfde locaties (Figuur 2). Alle drie de projecten hebben tot doel de grondwateraanvulling op hogere zandgronden zo nauwkeurig en klimaatrobuust mogelijk te bepalen. Voor een praktische aanpak is het werk in twee delen verdeeld, wat in de opzet van dit rapport is terug te vinden: - Het BTO-onderzoek en het onderzoek van het fundamenteel onderzoeksprogramma FOP zijn samengevoegd en worden beschreven in het eerste deel. De nadruk van dit onderzoek is het vinden van een in de praktijk eenvoudig toe te passen manier om de verdamping van hogere zandgronden vast te stellen, bijvoorbeeld via correctiefactoren voor de gewasfactoren. - Het belangrijkste doel van een promotieonderzoek is het afleveren van een goede dissertatie, het liefst in de vorm van ten minste vier artikelen die in peer reviewed tijdschriften zijn of kunnen worden geplaatst. Wetenschappelijke diepgang staat voorop, nieuwe praktische toepassingen zijn zeer gewenst, zullen ook worden nagestreefd, maar zijn geen doel op zich. Bernard Voortman heeft een onderzoeksvoorstel geschreven dat is besproken met en goedgekeurd door zijn begeleidingscommissie (Bartholomeus, Bierkens, Van Bodegom, Witte, Van der Zee). Het onderzoeksvoorstel is hier opgenomen als tweede deel. In search of the actual groundwater recharge Inleiding KWR May 2011

17 Figuur 2. Op dinsdag 1 maart 2011, een koude dag, leidt prof. Fred Daniëls ons rond door nationaal park de Hoge Veluwe. Op de achtergrond, v.l.n.r., Ruud Bartholomeus, Bernard Voortman en boswachter Evert Cock. De samenhang tussen beide delen is schematisch weergegeven in Figuur 3. We gaan deze figuur hier niet bespreken; we hopen dat hij een houvast biedt bij het vinden van samenhang in het onderzoek dat op de volgende pagina s wordt beschreven. Het onderzoek dat we doen is ambitieus en het is de vraag of het plan dat we hier presenteren wel volledig kan worden gerealiseerd binnen het gegeven budget en de beperkte tijd. Natuurlijk moeten we ook keuzes maken in de processen waarop we ons richten. In het eerste deel zullen we zo goed mogelijk aangeven wat we waarschijnlijk wel gaan halen, en over welke producten we minder zeker zijn. De komende tijd zullen we verder gaan met pogingen ons onderzoek te versterken. We willen andere partijen ervoor proberen te interesseren, maar we vinden het vooral gewenst de drinkwaterbedrijven meer bij het onderzoek te betrekken dan we tot nu toe hebben gedaan. In search of the actual groundwater recharge Inleiding KWR May 2011

18 Figuur 3. Samenhang tussen de verschillende onderdelen van het onderzoek. KVK = promotieonderzoek van Bernard Voortman; FOP = postdoc onderzoek van Ruud Bartholomeus, BTO = BTO-onderzoek grondwateraanvulling. In search of the actual groundwater recharge Inleiding KWR May 2011

19 In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

20 Part I Redefining crop factors (BTO & FOP) In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

21 In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

22 1 Introduction 1.1 Problem definition Climate change is among the most pressing issues of our time. The global temperature rises, the atmospheric CO 2 -concentration increases and prolonged dry periods are alternated with heavy rainfall events. According to climate scenarios for the Netherlands, the summer temperature may increase with 3 C, summer rainfall may decrease with 19% and reference evapotranspiration may increase with 15%. Although the total amount of summer rainfall decreases, heavy rainfall events will occur more frequently in summer. The winter rainfall will increase (Van den Hurk et al., 2006). As both precipitation and evapotranspiration are included in the water balance, it is evident that these changes would alter the current hydrological system. However, although the reference evapotranspiration is predicted to increase, the future actual evapotranspiration is highly unknown. As precipitation and actual evapotranspiration largely determine groundwater recharge, climate change likely affects both the spatial and temporal availability of groundwater. This will alter the fresh water availability for agriculture, nature conservation and industrial and drinking water supply. Current knowledge, however, is insufficient to estimate reliably the effects of climate change on future freshwater availability. It has been stated already that increased droughts may affect groundwater recharge rates, and herewith the future availability of fresh water (Bates et al., 2008) for e.g. agriculture and drinking water. In order to predict future groundwater recharge, however, we first need to understand possible vegetation responses to climate change, because these determine the future actual evapotranspiration ET a (Stuyfzand and Rambags, 2011; Wegehenkel, 2009). Although the future potential evapotranspiration ET p may increase due to increased temperatures (Solomon et al., 2007), it is the future ET a that affects groundwater recharge (Droogers, 2000). 1.2 Research topics and goals BTO and FOP-fund Currently, groundwater modeling frameworks, both numerical ones like ModFlow-MetaSwap (used for the NHI; and e.g. analytical impulse-response models like Menyanthes (Von Asmuth et al., 2002) include a very simplistic schematization of the vegetation. In hydrological modeling, it is common practice to fit simulated to measured groundwater levels, by adjusting e.g. the soil physical parameters as these are poorly known. Parameters of ET a usually remain untouched. Generally, ET a is calculated for a vegetation layer with fixed properties, for which ET p is derived from the reference ET (ET ref ) via the crop factor approach (Feddes, 1987). Such approach may be preferred for model simplicity, but the empirical crop factors can only be applied under the environmental conditions in which they were determined. Additionally, the current crop factors implicitly include the interception of crops under the current climate only. Consequently, empirical crop factors may for instance not be applied under a future climate with altered precipitation regime. Our goal is to unravel and redefine current crop factors so that they can be used for climate projections. Recently, we found an important vegetation response to drought conditions that may affect the groundwater recharge. Based on an explorative study in the Dutch dune area we found that the cover of vascular plants might decrease due to increased drought, which could increase future groundwater recharge, as ET a decreases with decreased vegetation cover (Kamps et al., 2008; Witte et al., 2008b). Drought induced differences in vegetation characteristics affecting ET a can be observed on surfaces with different slope and aspect, where received solar radiation determines the spatial variability in vegetation characteristics due to spatially variable drought conditions (Bennie et al., 2008; Nevo et al., 1998; Witte et al., 2008b). Solar radiation is a key determinant of vegetation characteristics, not only at large spatial scales, but also at local scales where slope and aspect may vary (Bennie et al., 2008). Higher solar radiation on equator-facing slopes relative to polar-facing slopes results in a higher evaporative demand, and consequently drier equator-facing slopes (Gutiérrez-Jurado et al., 2006). As a consequence of such In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

23 dry conditions, the vegetation on equator facing surfaces is more xeric (Bennie et al., 2008; Gutiérrez- Jurado et al., 2006), has a lower aboveground biomass and plant cover (Badano et al., 2005; Broza et al., 2004; Holland and Steyn, 1975; Nevo et al., 1998; Reddy et al., 2004; Schimper, 1903; Shoshany, 2002; Witte et al., 2008b), and is more patchy (Kutiel et al., 1998; Shoshany, 2002) than on polar facing surfaces. Our goal is to relate empirically reference drought stress, i.e. a measure of the soil dryness, to vegetation cover. Changes in soil cover will affect the heat balance at the soil-atmosphere interface. Current hydrological models do consider local effects of low soil moisture pressure in the root zone on ET a by using the Feddes-function for relative root water uptake (Feddes et al., 1978). However, besides this water-limiting effect on ET a, also energy plays an important role. So far, the energy balance is only incorporated in ET ref, which has only very limited variability across the Netherlands and does not account for small-scale differences in ET a. However, besides the water balance, also the energy balance will be site-specific. A low vegetation cover, for example, will increase the soil heat flux, which will affect evaporation (Monteith and Unsworth, 1990). Our goal is to calculate the feedback of vegetation cover on the soil heat flux and ET a (optional, if time allows). Transpiration is closely correlated to photosynthesis and herewith CO 2 uptake. The increase of the atmospheric CO 2 concentration potentially affects transpiration; the water use efficiency of plants will increase due to increased CO 2. Recent research (Kruijt et al., 2008; Witte et al., 2006a; Witte et al., 2006b) showed that the stomatal aperture of plants will be lower in the future climate due to increased CO 2. Kruijt et al. (2008) and Witte et al. (2006a; 2006b) showed that future ET ref will still increase, but about 2% (for a grass cover) less than according to the KNMI 06 climate scenarios. So, together with soil moisture, interception and energy availability, [CO 2 ] will affect ET a. Our goal is to take better account of the plant stomatal response on atmospheric [CO 2 ], micro-climatic variability, interception and soil moisture, simultaneously (optional, if time allows). Overall, we intend to estimate optimally the actual evapotranspiration ET a, for current and for future climatic conditions and herewith the simulation of groundwater recharge in groundwater models. We focus on the effects of increased drought on vegetation characteristics and the vegetation feedbacks on evaporation, transpiration and interception. Finally, these factors determine groundwater recharge. The internationally established model for the unsaturated zone SWAP (Van Dam et al., 2008) will play a central role in this project. SWAP incorporates in detail the interacting processes in the soil-water-plantatmosphere continuum, and allows in-depth analysis of each of these processes. 1.3 Outline and planned deliverables The processes that determine the interactions between climate, vegetation, soil moisture and groundwater recharge are complex. Although drought stress is the most important stress factor for plant growth (Reddy et al., 2004) and primarily determines vegetation characteristics (Porporato et al., 2001b) other stresses may affect vegetation characteristics as well. Some other stresses however, like heat stress and shear (wind) stress, are often initiated by drought stress, and are therefore of secondary importance. We also recognize that the impact of e.g. salt spray in the coastal zone (Stuyfzand, 1993), and microclimatological conditions like cold stress in dune depressions (Vugts, 2002), may be important for the vegetation. However, given the period of our project and the complexity of each of these processes and their interactions, it will be infeasible to incorporate them all in our research. Therefore, this research is restricted to the analysis of climate change effects on drought stress and the resulting vegetation characteristics. Effects of salt and wind will be minimized by a deliberate selection of field plots, i.e. by investigating inland rather than coastal dunes. Additionally, in dune systems processes like water repellency, preferential flow and surface runoff affect the moisture availability for plants and the groundwater recharge (Dekker, 1998; Dekker et al., 2001; Dekker et al., 2000; Ritsema, 1998; Ritsema and Dekker, 1994). Because of the difficulty to generalize these processes and to schematize them, we do not explicitly incorporate them in our modeling procedures now. In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

24 Overall, within this project we prioritize to get insight in the effects of climate change on reference drought stress, and the response of the vegetation, and their primary feedback on the groundwater recharge by alterations in evaporation, transpiration and interception. The effects of reduced cover on hydrological processes like water repellency could be important, but are unlikely to fit within the current project. We will also only use plot-scale data and do the simulations accordingly. Extrapolation to large areas requires different data (e.g. from remote sensing), which falls outside the scope of this research. We intend to deliver: - simulations of drought stress on inclined surfaces; - empirical relationship between drought stress and vegetation cover; - redefined crop factors for ET ref, to be used for climate projections with the crop factor approach; - simulations of groundwater recharge with SWAP on slopes with different soils and orientations for different climate scenarios; - simulations of groundwater levels with Menyanthes for both the current and future climatic conditions based on the redefined crop factor approach. Optional (depending on time and student projects): - reliable and non-biased estimations of fractional vegetation cover; - integrating water and heat balance; - integrating the Penman-Monteith equation for evapotranspiration and the Jarvis model for stomatal aperture in SWAP, which allows a process-based analysis of climate change effects on actual evapotranspiration. In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

25 In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

26 2 Research topics and products 2.1 Vegetation cover vs. reference drought stress (step A) We will relate reference drought stress, i.e. a measure of the soil dryness, to the fractional vegetation cover. Drought induced differences in vegetation characteristics affecting ET a can be observed on surfaces with different slope and aspect, where received solar radiation determines the spatial variability in vegetation characteristics due to spatially variable drought conditions (Bennie et al., 2008; Nevo et al., 1998; Witte et al., 2008b). Solar radiation is a key determinant of vegetation characteristics, not only at large spatial scales, but also at local scales where slope and aspect may vary (Bennie et al., 2008). Higher solar radiation on equator-facing slopes relative to polar-facing slopes results in a higher evaporative demand, and consequently drier equator-facing slopes (Gutiérrez-Jurado et al., 2006). As a consequence of such dry conditions, the vegetation on equator facing surfaces is more xeric (2008; Gutiérrez-Jurado et al., 2006), has a lower aboveground biomass and plant cover (Badano et al., 2005; Broza et al., 2004; Holland and Steyn, 1975; Nevo et al., 1998; Reddy et al., 2004; Schimper, 1903; Shoshany, 2002), and is more patchy (Kutiel et al., 1998; Shoshany, 2002) than on polar facing surfaces. Not only in arid and semi-arid regions, but also in humid areas, small-scale differences in topography and meteorological conditions cause significant differences in vegetation characteristics (Bennie et al., 2008). Field data from the Netherlands, for instance, show that the fraction of xerophytic plant species within a vegetation plot is higher on south slopes than on north slopes, and the fraction of xerophytes is higher on sandy than on clayey soils. That differences in the fraction of xerophytes are causally connected to drought stress was already indicated by Schimper (1903). Although other stresses may affect vegetation characteristics as well, drought stress is the most important stress factor for plant growth (Reddy et al., 2004) and primarily determines the vegetation characteristics (Porporato et al., 2001b). Other stresses are often initiated by drought stress. Heat stress for example, only occurs when a shortage of water limits transpiration. Hence, cooling of the plant tissue, and nutrient limitation decrease at lower water availability (Porporato et al., 2001b). To get insight in the effect of drought stress on vegetation characteristics (i.e. fraction of xerophytes and fractional vegetation cover) we will (Figure I.1): - Step A1: define a process-based measure of reference drought stress; - Step A2: gather field data of fractional vegetation cover based on digital camera survey; - Step A3: define an empirical relationship between reference drought stress to fractional vegetation cover. Figure I.1: Relationships between reference drought stress and soil and vegetation cover In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

27 2.1.1 Process-based, physiologically relevant measure of drought stress (A1) Various indices exist to characterize drought as experienced by vegetation. Aboveground biomass, for instance, has been related to mean annual precipitation (Huxman et al., 2004), and mean precipitation deficit (Ciais et al., 2005). Such drought indices, however, cannot explain local variations in vegetation characteristics for they do not account for differences in soil texture, groundwater depth and small topographic variations that impact on micro-meteorology (Bennie et al., 2008). These factors together determine the amount of soil water that is available to plants. When soil moisture availability is too low to meet the water demand for transpiration, a plant suffers from drought stress (Reddy et al., 2004; Schimper, 1903). This definition of drought stress, also called physiological drought (Schimper, 1903), implies that not only water availability but also vegetation s demand has to be considered. Starting from this ecological basis, we will focus on transpiration reduction as the vegetation response to drought stress, thus considering both the supply and demand of water. Outline: - 1D analyses (SWAP (Van Dam et al., 2008)) for a reference vegetation - adjust SWAP to allow simulations on inclined surfaces, by correcting for solar radiation, precipitation and evaporating surface - synthetic and field sites (existing data of Jansen and Runhaar (2005) and Jansen et al. (2000)) Timeline: June 2010 April 2011 Products: - process-based measure of reference drought stress - peer reviewed publication Drought stress and vegetation characteristics on sites with different slopes and orientations ; Bartholomeus, Witte, Runhaar; Ecohydrology, peer reviewed publication Need of considering vegetation feedbacks to reliably estimate groundwater recharge ; Opinion paper, Witte, Bartholomeus, Voortman; Ecohydrology, Estimation of fractional vegetation cover based on digital photography (A2) Recently, an important feedback of the vegetation to climate change was identified in dune systems: the fraction of bare soil and non-rooting species (lichens and mosses) in the dune vegetation might increase when, according to the expectations, summers become drier (Van den Hurk et al., 2006). For hydrological modeling of dune areas, the fractions of bare soil and vegetation are important input parameters. Evapotranspiration of bare soil and non-rooting species is much lower than that of vascular plants and thus the vegetation composition and cover largely determine the soil moisture conditions (Stuyfzand, 1993; Stuyfzand and Rambags, 2011). Therefore, robust and accurate estimations of the vegetation cover are required. Traditionally, vegetation cover is estimated visually, but this method introduces large biases and is prone to systematic and random errors. Furthermore, repeated sampling can give varying results. We are interested in a method that is less sensitive and less subjective and can deliver reliable estimates of bare soil, vascular plants and moss fractions. Outline: - Student project to explore possibilities of using remote sensing to estimate soil cover fractions; - Optical sensing techniques can be a possible solution, but several possibilities are available with varying spectral and spatial dimensions (e.g. ASD-Fieldspectrometer, Cropscan, Digital camera); - Practical and accurate instruments and methods should be further developed (student project) and applied to acquire additional field data with accurate cover fractions; Timeline: - Student projects: April-May 2010, and May-June 2011 In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

28 - Acquire additional field data: May/June 2011 Research partners: - CGI (Wageningen University) Selection of relevant literature: - (Corre, 1991; Curr et al., 2000; Hemmleb et al., 2005; Hu et al., 2007; Louhaichi et al., 2010) Relationships reference drought stress cover fractions (A3) The results from Step A1 and Step A2 will be combined, which results in an empirical relationship between reference drought stress and fractional vegetation cover (grass, moss, bare soil). To do so, reference drought stress will be simulated for all sites of Step A2, i.e. sites where cover fractions will be estimated. Hitherto, detailed soil physical properties of each site should be estimated. Detailed soil physical properties will improve accurateness of simulated reference drought stress. Together with unbiased fractional vegetation covers, this will result in an empirical relationship between reference drought stress and fractional vegetation cover with minimal noise and high predictive power. Such relationship will provide insight in the effect of drought stress on fractional vegetation cover under field conditions, and allows predicting the potential future fractional vegetation cover under future climatic conditions. Outline: - empirical relationship, i.e. no dynamic vegetation models, but field data; - sites at dry sands at Veluwe and Dutch coastal dunes; start with sites of Jansen and Runhaar, add new sites; - soil samples of organic top layer and subsoil to obtain Van Genuchten soil physical parameters (analysis at VU-lab); - fractional vegetation cover from Step A2, but also detailed vegetation records. Timeline: - May-December 2011 Products: - empirical relationship between reference drought stress and vegetation cover fractions; - robust method for estimation of vegetation cover fractions, with low bias compared to traditional methods; - peer reviewed publication (based on Steps A1, 2 and 3). In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

29 2.2 Unravel crop factors (step B) The reference evapotranspiration ET ref is defined as the evapotranspiration of short grassland optimally supplied with water. ET ref is used to estimate the potential evapotranspiration ET p of a crop or vegetation, which reduces to the actual evapotranspiration ET a in case of water shortage. Here, we focus on the estimation of ET p from ET ref, which is generally done by applying crop factors, k c (Allen et al., 1998; Allen et al., 2005; Feddes, 1987): ET p = k c ET ref. Crop factors have been derived for different crops and vegetation types based on measurements and are available for e.g. grass, maize, deciduous forest and pine forest. In the Netherlands, crop factors according to (Feddes, 1987) are used to transpose ET ref_mak (i.e. ET ref for a reference grassland, according to Makkink (1957)) to the ET p of a crop (i.e. evapotranspiration of a crop without water stress). ET ref_mak is supplied by the KNMI. Currently used crop factors for the Netherlands are published by (Feddes, 1987), and by definition they should be used to derive ET p from ET ref_mak : ET k ET [1.1] p c ref_mak where k c is the crop factor and ET ref_mak is the potential evapotranspiration of grass according to Makkink (1957). ET ref_mak includes I, E p and T p. Consequently, k c accounts for each of these evaporation components, and thus: ETp I Ep Tp [1.2] where ET p is the potential evapotranspiration of a cropped surface, I is the sum of evaporation of intercepted water (P i ), E p is the potential soil evaporation and T p the potential transpiration. So, crop factors according to Feddes (1987) implicitly involve I. Additionally, crop factors for the Netherlands have been derived by soil water balance experiments, and especially from sprinkling experiments in the field, where water is applied in quantities that ET p is reached (Feddes, 1987). Sprinkling, however, leads to interception. Feddes (1987) also emphasizes that the presented crop factors are averages taken over a population of average, dry, and wet years, that will certainly not be homogeneously distributed. Crop factors thus correct for the combined effect of E p, T p and I. Each of these factors, however, may change differently under changing climatic conditions. This implies that crop factors can only be applied under the conditions in which they were determined. Consequently, these crop factors are inappropriate to be used for extreme weather years such as 1976 (dry) and 1998 (wet) and they cannot be applied under changing climatic conditions. The crop factor approach by Feddes (1987) is different from that of the FAO (Allen et al., 1998), as these are by definition applied to correct for E p and T p only. However, Allen et al. (1998) indicate that their crop factors should be multiplied with a factor if interception, due to sprinkling irrigation for example, is involved. This indicates that I could significantly affect evapotranspiration from a vegetated surface. Therefore, it is important to consider the contribution of I on ET p explicitly. The KNMI provides daily values of ET ref_mak. They define ET ref_mak as the evapotranspiration from a dry grass surface, i.e. I=0, optimally supplied by water (Droogers, 2009). This definition does not correspond to the definition of Feddes (1987) which should be used for the crop factor approach. Additionally, ET ref_mak includes empirical constants and it is not a physical quantity (De Bruin, 1987), which makes it inapplicable for climate projections. Makkink (1957) derived his relationship and the empirical constants on field data, i.e. implicitly involving I, at least at wet days. Consequently, ET ref_mak as provided by the KNMI is not valid for dry crops only, as mentioned by Droogers (2009), but for the mean meteorological conditions that Makkink used for his equation. All in all, some remarks can be made to the applicability of the current crop factor approach in climate projections. Therefore, we intend to improve the current calculations of ET p and I, by detailed simulations using the Penman-Monteith approach, from which correction factors for Makkink will be derived. In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

30 This includes the following steps (Figure I.2): - Step B1: Simulation of ET p and I using Penman-Monteith: derive ET ref_mak_dry - Step B2: Derive ET ref_mak_dry with Makkink only - Step B3: Improvement of calculation scheme in SWAP - Step B4: Simulation of ET p and I using Penman-Monteith and ET ref_mak_dry for KNMI 06 scenarios - Step B5: Sensitivity analysis using SWAP and Menyanthes Simulation of ET p and I (B1) SWAP (Van Dam et al., 2008), divides ET p (obtained from ET ref through crop factors) in P i (intercepted precipitation, which will (partly) evaporate (=I)), E p (potential evaporation under standing crop, with no soil heat flux) and T p (potential transpiration). Calculation of T p, however, requires both P i and E p as input. Additionally, going from ET ref to ET p via k c implicitly includes P i. Using these k c -values for climate projections of T p will be fundamentally wrong, as the interception that is involved in k c will differ from the simulated interception that is input for the calculation of T p (Kroes and Van Dam, 2003). Therefore, we will redefine the crop factors of Feddes (1987), so that they are applicable on a dry crop only. Contrary to Makkink, the Penman-Monteith equation does not include empirical constants. Penman- Monteith may thus outcompete Makkink for climate projections. In fact, the KNMI also used Penman- Monteith for the KNMI 06 scenarios (Van den Hurk et al., 2006); the effects of climate change on ET ref were first calculated with Penman-Monteith and thereafter applied on ET ref according to Makkink. This is unnecessarily elaborative, as we can use ET ref according to Penman-Monteith directly. Additionally, De Bruin (1987) writes that Penman-Monteith successfully describes the transpiration as well as the interceptive loss from different kinds of vegetation such as tall forests, arable crops, heathland and grass. This provides more confidence in the validity of Penman-Monteith than of Makkink. First, we will use the Penman-Monteith equation to simulate E p, T p and I for the current climatic conditions ( ), which together give the evapotranspiration for a reference grass under the prevailing, actual meteorological conditions. Ideally, (E p + T p + I), i.e. ET ref_pm_avg, equals ET ref_mak, but slight differences are to be expected (see Droogers, 2009). Then, we will calculate ET ref_mak for a dry crop only, by correcting for the effect of I on ET p, given by g I : g I ET ET ref_pm_dry ref_pm_avg [1.3] ET g ET [1.4] ref_mak_dry I ref_mak Figure I.2: Unravel crop factors; split ET ref into E p, T p and I In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

31 where g I is a factor that corrects the contribution of I in ET ref_mak. ET ref_mak_dry is the reference evapotranspiration of Makkink for a dry crop (I = 0). This ET ref_mak_dry should replace ET ref_mak in SWAP. Only then, T p can be simulated correctly. The reference evapotranspiration of a wet crop should be equal to the Penman evaporation of open water. NB: g I = (E p + T p ) / (E p + T p + I) = (E p + T p ) / ET ref_pm_avg will not give the correct factor, as the reduced T p when the crop is wet will not be considered then Consequently, also the crop factors should be redefined in order to get ET p_dry. Therefore, for each crop we need to derive g I, which gives g Ic. We will do so with SWAP, using the same method as for a reference grass, but with specific crop characteristics (LAI, r crop, r air ). Finally this will give: kc_dry gi,c kc [1.5] Derive ET ref_mak_dry with Makkink only (B2) As an alternative approach, we will run two SWAP-simulations based on ET ref_mak : one with P i =0 combined with optimal soil moisture availability (which gives ET p,pi=0 ), and one with the real meteorological conditions i.e. real P gross and P i (which gives ET p ). Ideally, ET p,pi=0 /ET p gives g I. First, we will do these simulations for grass only, and compare g I to that of equation [1.3]. If the results are comparable then we could use this approach for all other crops. However, as ET ref_mak is valid for the average meteorological conditions only, the current SWAP-schematization for the use of ET ref_mak is arranged accordingly. This schematization may not hold for conditions with P i =0, which may cause differences in the obtained g I s. Therefore, the usability of this approach is quite uncertain. Nevertheless, if g I for a grass from Step B2 is comparable to that of B1, we will run SWAP with ET ref_mak for different LAI s and k c s. This allows deriving a relationship between LAI, k c and g I,c, which could be used to derive k c_dry for a variety of crops Improvement of calculation scheme in SWAP (B3) Using ET ref_mak_dry and k c_dry allows an alternative calculation of the evapotranspiration of a wet surface compared to the one that is currently incorporated in SWAP (with ET ref_mak as input). By using ET ref_mak_dry we can account better for different evaporative fluxes under either wet or dry conditions. When ET ref_mak_dry will be input, we can directly calculate ET p_dry. ET p_wet should be taken equal to the Penman open water evaporation E 0. Additionally, k c will be replaced by k c_dry. This allows discriminating better between wet and dry conditions. So we need (see also List of Symbols : 1 Pi a LAI 1 b Pgross 1 a LAI ETp_dry kc_dry ETref_Mak_dry [1.7] ETp_wet E 0 [1.8] Then: gr LAI E E e [1.9] or p p_bare [1.6] In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

32 P i Ep Ep_bare 1 1 SC ETp_wet and finally: [1.10] P i T 1 ET E ETp_wet p p_dry p Implementation of ET ref_mak_dry in SWAP, as described above, allows continuing using Makkink as supplied by the KNMI. However, I and its effect on T p will be handled differently compared to the current approach. By adjusting SWAP, Makkink can be used in climate scenarios, as it explicitly separates dry from wet canopy conditions. We will explore this further in step B4. [1.11] Simulation of ET p and I using Penman-Monteith and ET ref_mak_dry for KNMI 06 scenarios (B4) In this analysis, we will show the application of our improved schematization of ET p and I in groundwater models for the KNMI 06 scenarios. Current hydrological models assume fixed evaporation characteristics of the vegetation, and the effect of interception on the groundwater recharge is not considered explicitly. With climate change, however, vegetation characteristics and interception will change. Here, we will show how evaporation, transpiration and interception should be handled in groundwater models in climate projections, and how each of these factors will change due to climate change. We will use the correction factors of Kruijt et al. (2008) and Witte et al. (2006a; 2006b) (f CO2 ) to correct ET ref_mak_dry for a higher water use efficiency due to increased atmospheric CO 2 -concentration under future climatic conditions. For each KNMI 06 scenario (which includes f KNMI ), we will run SWAP and simulate ET p based on the adjusted SWAP schematization. ET p will be based on the adjusted SWAP version, with ET p_dry_cc as input. For a specific crop and climate scenario the following holds: ETp_dry_cc ETref_Mak_dry kc_dry fknmi f CO [1.12] 2 ETp_cc Ep_cc Tp_cc I_cc (output of SWAP) [1.13] Outline step B1-B4: - vegetation of dry sands: grass, heather, deciduous forest and evergreen forest - calculating interception of forests and of short vegetation requires different approaches (Gash vs. Von Hoyningen-Hüne for forests and agricultural vegetation resp.). In contrast to Von Hoyningen-Hüne, Gash accounts for the effect of rain duration and evaporation during the rain event. - no measurements, only use existing data (e.g. the Castricum lysimeter data; see Van der Hoeven (2010), and general equations for E, T and I, as used in SWAP. Only then, we are able to compare our redefined crop factors to current ones, using SWAP. Both in the Netherlands and internationally, SWAP can be considered the standard to estimate the actual evapotranspiration as function of meteorological conditions combined with crop and soil conditions. Timeline step B1-B4: - April 2011-February 2012 Products: In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

33 - Correction factors k c_dry for grass, heather, deciduous and evergreen forest, to be applied on ET ref_mak_dry to get ET p for each crop/vegetation type Sensitivity analysis using SWAP and Menyanthes (B5) We will perform a sensitivity analysis on the effect of using our redefined crop factors (k c_dry ) or the current ones (k c ) in soil moisture and groundwater simulations. First, we will simulate groundwater levels and soil moisture conditions with SWAP with the traditional crop factors as input, for both the current and future climatic conditions (with f KNMI and f CO2 ). We will compare climate-induced changes in groundwater levels with SWAP simulations based on k c_dry, f KNMI and f CO2. This will show us whether redefined crop factors significantly alter simulated groundwater levels. Then, we will analyze the effect of using either ET ref_mak (i.e. valid for average conditions), or I, E p and T p explicitly (i.e. accounting for temporal deviations from the average conditions) in groundwater level simulations. This will show us the need of considering the effects of temporal dry or wet conditions on interception and on groundwater recharge. Currently, Menyanthes uses P gross and ET ref_mak, together considered as measure of groundwater recharge, as explaining variables of the groundwater level. Temporal variability in both I, E p and T p is thus not considered. We argue that using ET ref_mak as input of groundwater simulations, without explicitly accounting for the contribution of interception, will cause systematically too low or too high groundwater levels in periods of low and high precipitation, respectively. We argue that taking account of the temporal effect of I on ET p and herewith for the minimum groundwater recharge (P gross E p - T p I), will improve the accurateness of groundwater level simulations. To show this, we will simulate E p, T p and I with SWAP and use this as input for Menyanthes. We will compare model predictions to those based on ET ref_mak of the KNMI. We will compare simulated groundwater levels to measured ones for E p, T p and I from SWAP, and ET ref_mak. Outline: - only SWAP and Menyanthes; - sites with groundwater independent vegetation, for which both groundwater levels and drainage functions are available; - simulation of E p, T p and I according to current SWAP model; no feedbacks of altered soil heat flux due to drought-induced low vegetation cover. Timeline: - January-July 2012 Products: - paper on the sensitivity of simulated groundwater levels to evaporation, interception and transpiration In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

34 2.3 Water and heat balance as function of soil and vegetation cover (optional, if time allows) (step C) In the calculation of ET ref the effect of the soil heat flux is considered to be of minor importance as the vegetation is fully covering the soil (Allen et al., 1998). Changes in soil cover (see A), however, will affect the heat balance at the soil-atmosphere interface. A low vegetation cover, for example, will increase the soil heat flux, which will increase ET p (Monteith and Unsworth, 1990). Climate change may alter both the fractional vegetation cover (see A), and the interception of the vegetated surface (see B). A low vegetation cover will affect the soil heat flux and herewith ET p. We will analyze the effect of droughtinduced changes in fractional vegetation cover (A) on E p, T p and I for different fractional vegetation covers. In the end this will show us how low vegetation cover affects the groundwater recharge through alteration of the local heat and water balance. If time allows, the following steps will be followed: - Step C1: simulate soil heat flux for partly covered surfaces with SWAP and its feedback on ET p according to Penman-Monteith - Step C2: Effect of bare soil on water balance components, through its feedback on transpiration: combine heat and water balance Soil heat flux and temperature for partly covered surfaces (C1) A general assumption in the application of the reference evapotranspiration is that it holds for a large uniform surface, so that advection can be ignored. Advection, or the transfer of heat and/or water vapor from the surroundings, provides an extra source of energy and may play a role in partly covered soils. Heat generated at a dry soil surface, for example, can be consumed through increased transpiration by the plants adjacent to the dry soil (Rosenberg et al., 1983) (p. 230). Also within a single dune surface, an open or patchy vegetation layer affects the soil heat flux, as shown by Kustas et al. (2000). The difference between net radiation R n and the soil heat flux G determines the available energy for transpiration. If the heat loss from the soil to the atmosphere is high (i.e. G is directed upward and thus G<0) there is more energy available for transpiration. This available energy will be affected by the lateral transport of heat, which especially occurs in a patchy vegetation. Therefore, we need to consider the effect of partly covered soils on the soil temperature (T soil ) and air temperature near the soil surface (T air ) and G. SWAP includes these variables, but improvements may be needed. Especially the feedback of G on T air just above the vegetation layer is not included yet. We will execute SWAP simulations on north and south facing surfaces, i.e. surfaces with differences in solar radiation R s. We will investigate the effect of different soil cover fractions on T soil, T air, G, and ET p. A sensitivity analysis will provide insight in the importance of considering both the water and heat balance to estimate ET p and ET a. We hypothesize that a low vegetation cover increases the available energy for the sparse vegetation that is present. This will increase the demand of water, i.e. T p. If the water availability does not increase proportionally, drought stress will increase. We will investigate the effect of low soil cover on T p, T a and actual drought stress, using the SWAP simulations on north and south facing surfaces. This means that we use ET ref_pm, and the reduction functions for root water uptake of Feddes et al. (1978) or of Metselaar and de Jong van Lier (2007) to describe the difference between T p and T a. Outline: - SWAP simulations on north and south surfaces in both coastal and inland dune areas in the Netherlands - adjust SWAP to account for feedback of soil heat flux on air temperature and on ET p Timeline: , possibly student project of 5 months In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

35 Products: - Insight in the effect of low soil cover and altered heat balance on ET p and its feedback on drought stress - Note or student report on effect fluxes on water balance Relevant literature: - (Kustas et al., 2000; Yang et al., 1999) Effect of bare soil on water balance components, through its feedback on transpiration: combine heat and water balance (C2) Our ultimate step could be a fully process-based simulation of the plant stomatal response on atmospheric [CO 2 ], micro-climatic variability in heat, and soil moisture, simultaneously. Hereto, we will simulate the plant stomatal aperture explicitly with the Jarvis equation (e.g. Jarvis et al., 1999), and use this as input for the Penman-Monteith equation. We intend to implement this approach in SWAP, and simulate groundwater recharge under both the current and future climatic conditions with a fully process-based approach, without the interference of a reference crop and crop factors. Ideally, this gives insight in the possibility to calculate directly the actual evapotranspiration in groundwater models. This subject could possibly be executed by a student in collaboration with Prof. dr. ir. M. Bierkens of Utrecht University and Prof. dr. ir. S. vd Zee of Wageningen University. In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

36 3 Delivered products student project CGI: insight in methods, field work and report: o - publications o - presentations: o o o o - proceedings: o o o o SOIL & VEGETATION FRACTIONS IN DUNE ECOSYSTEMS; Estimation of soil and vegetation fractions from reflectance measurements in dune ecosystems Bartholomeus, R.P., Voortman, B. and Witte, J.P.M., De toekomstige grondwateraanvulling. H2O, 17: AGU Fall Meeting, Climate change effects on vegetation characteristics and groundwater recharge. San Francisco, Calif., , oral presentation. Deltas in times of climate change, Climate change effects on vegetation characteristics and groundwater recharge. Rotterdam, The Netherlands, , oral presentation. EGU General Assembly, Climate change effects on vegetation characteristics and groundwater recharge. Vienna, Austria, , poster presentation. Latsis 2010 International Symposium on Ecohydrology, Climate change effects on vegetation characteristics and groundwater recharge. Lausanne, Switserland, , poster presentation Bartholomeus, R.P., Voortman, B., Witte, J.P.M., Climate change effects on vegetation characteristics and groundwater recharge. Abstract H43J-04 presented at 2010 Fall Meeting, AGU, San Francisco, Calif., Dec. Bartholomeus, R.P. and Witte, J.P.M., Climate change effects on vegetation characteristics and groundwater recharge. Deltas in times of climate change. Abstracts scientific programme deltas in depth. Rotterdam, pp Witte, J.P.M., Bartholomeus, R.P. and Cirkel, D.G., Climate change effects on vegetation characteristics and groundwater recharge. Latsis 2010 International Symposium on Ecohydrology. Abstract book Latsis 2010, Lausanne, Switserland, pp. 72. Witte, J.P.M., Bartholomeus, R.P. and Cirkel, D.G., Climate change effects on vegetation characteristics and groundwater recharge, EGU General Assembly Geophysical Research Abstracts, Abstract EGU , Vienna, Austria student project CGI: insight in methods, field work and report: AHN-2 combined with 4-band air pictures, both 25*25cm scale. Couple with drought stress on inclined surfaces. - student at KWR: insight in 1D (SWAP) vs 3D (Hydrus-3D) modeling of the unsaturated zone at inclined surfaces and partly covered soils. - student at CGI: Automated estimation of fractional cover of plant functional types using digital near infrared photography - publications: In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

37 o o - presentations: o - proceedings: o Bartholomeus, R.P., Witte, J.P.M., Runhaar, H Drought stress and vegetation characteristics on sites with different slopes and orientations, Ecohydrology. Witte, J.P.M., Bartholomeus, R.P., Voortman, B., Need of considering vegetation feedbacks to reliably estimate groundwater recharge, Ecohydrology. EGU General Assembly, Drought stress and vegetation characteristics on sites with different slopes and orientations. Vienna, Austria, , poster presentation. Bartholomeus, R.P. and Witte, J.P.M., Drought stress and vegetation characteristics on sites with different slopes and orientations, EGU General Assembly Geophysical Research Abstracts, Abstract EGU , Vienna, Austria. In search of the actual groundwater recharge Part I (BTO & FOP) KWR May 2011

38 Part II PhD proposal Bernard Voortman (KvK) In search of the actual groundwater recharge Part II (KvK) KWR May 2011

39 In search of the actual groundwater recharge Part II (KvK) KWR May 2011

40 The future groundwater recharge in coastal and inland sand dunes: evapotranspiration response of natural vegetation to climate change Research proposal February 2011 B.R. Voortman Supervisors: Prof. Dr. ir. S.E.A.T.M. van der Zee Prof. Dr. ir. M.F.P. Bierkens Prof. Dr. ir. J.P.M. Witte Dr. ir. P.M. van Bodegom Dr. ir. R.P. Bartholomeus In search of the actual groundwater recharge Part II (KvK) KWR May 2011

41 In search of the actual groundwater recharge Part II (KvK) KWR May 2011

42 1 Introduction 1.1 Background and problem definition Global temperatures are expected to increase in the coming century. Higher temperatures will lead to a larger water-holding capacity of the atmosphere which favours increased climate variability, with more intense precipitation events and more droughts (Trenberth et al., 2003). The timing and amount of rainfall events may shift throughout the year (Meehl et al., 2007). These changes will alter the amount of water that percolates to the saturated zone, i.e. the groundwater recharge, as well as the size and dynamics of fresh groundwater bodies. Changes in the supply of fresh groundwater will affect ecosystems, agricultural productivity and the supply of drinking water. Current knowledge, however, is insufficient to reliably estimate the amount of groundwater recharge under changing climatic conditions (Wegehenkel, 2009). The major drawback of current methods to estimate groundwater recharge is the use of static predefined vegetation characteristics in simulation models. It is likely that vegetation characteristics such as the species abundance and composition will adjust to future weather conditions (Kreyling, 2008; Miller et al., 2010) which will affect the energy and water balance. The potential of climate change to affect vegetation characteristic is especially large in water limited ecosystems where soil moisture availability is a constraining factor for plant survival (Porporato et al., 2001a). For these regions the response of vegetation should be included in predictions of the future groundwater recharge. Coastal and inland sand dunes are often considered systems where strong relationships between soil moisture conditions and vegetation characteristics are present (Maun, 1994; Park, 1990; Tazaki, 1960). Even in temperate regions with a large annual precipitation amount, soil moisture conditions affect vegetation characteristics in sand dune systems (Park, 1990). These regions are of interest in this study. In an exploratory study Witte et al. (2008a) showed that increased summer drought might lead to a dieback of vascular plants and an increase in the cover fraction of non-rooting species (mosses and lichens) and barren surfaces in coastal dunes of The Netherlands. Because the evapotranspiration of mosses, lichens and barren soil is much less than that of vascular plants this vegetation response will lead to a lower evapotranspiration and to an increase in groundwater recharge. This interaction between climate, vegetation and evapotranspiration needs further attention as it has a major effect on estimates of the future groundwater recharge. 1.2 Research objectives The primary aim of this research is to improve simulations of evapotranspiration and groundwater recharge in hydrological models in the context of climate change by considering vegetation as a dynamic component. Modelling vegetation dynamics can be achieved on different spatial scales and different levels of complexity. In this study species will be categorised in plant functional types such as bare soil, mosses and lichens, grasses and shrubs. Key interest of this study will lie in quantifying the cover percentage of these plant functional types under changing climatic conditions on groundwater independent sandy soils. This study aims to give answer to the following research questions: How are soil moisture conditions related to the cover fractions of plant functional types such as bare soil, mosses, lichens, grasses and shrubs? How do changes in cover fractions of plant functional types affect soil evaporation, transpiration, interception water loss and groundwater recharge? What is the effect of changing meteorological conditions on the distribution of plant functional types and how does this affect the future groundwater recharge? 1.3 Outline This research proposal consists of four topics which should lead to four scientific publications. Before these are introduced, a brief review of the vegetation distribution and dynamics in coastal and inland In search of the actual groundwater recharge Part II (PhD) KWR May 2011

43 sand dunes is given in chapter 2. After this chapter, an analysis of the available data and the research approach will be given (chapter 3). The first topic about the effects of mosses and lichens on the soil water balance will be discussed in chapter 4. The second topic is about modelling plant available water in one and three spatial dimensions (chapter 5). This chapter mainly deals with unsaturated flow processes and the effects of the vegetation structure and topography on the water balance. The third topic is about the relationship between soil moisture conditions and the coverage of dry grassland vegetation. A database of 3000 relevés of Corynephorus canescens grassland vegetation following a West- East gradient from The Netherlands to Slovakia will be used to relate soil moisture conditions to the cover percentage of dry grassland vegetation (chapter 6). The last topic will be about the succession of dune vegetation under changing climatic conditions and its effect on the water balance (chapter 7). This research proposal ends with a timetable summarizing the amount of time needed to perform this research per topic. In search of the actual groundwater recharge Part II (PhD) KWR May 2011

44 2 Vegetation dynamics in coastal and inland sand dunes 2.1 Introduction An analysis of historical data on active drift sand areas may reveal which processes affect vegetation development. Pollen analyses revealed that the first phase of aeolian reformation of coastal dunes started between the 10 th and 12 th century and ended in the 13 th century (Jelgersma et al., 1970). During the same period inland drift-sands were formed which represent reactivated deposits of Youger Cover Sands (Koster, 1978; Koster et al., 1993). Land cultivation and deforestation are frequently mentioned as the primary cause of increased aeolian activity at the end of the first millennium (Berendsen, 2005; Oostra, 2006). Increased human activity during medieval times coincided with a relatively warm and dry period, the so-called climate optimum (ca. 950 until 1250 AD) (Figure II.3). The climate of this period was continental with warm summers and cold winters. Geological reconstruction of dried-out fen deposits on the Veluwe indicate that annual precipitation amounts decreased during this period with 60 to 180 mm with respect to the preceding climate (Heidinga, 2006). We believe that the shift from an oceanic to a more continental climate during the climate optimum might have promoted aeolian activity, namely through a drought induced vegetation dieback. It is, however, difficult to unravel the effects of climate change from anthropogenic effects, because both have a potential to reduce the vegetation coverage and promote aeolian activity. Figure II.3. Past 1.3 ka northern hemisphere temperature variation reconstructed using different climate proxies (source Jansen et al. (2007)) Aerial photographs and historical maps show a decrease of barren surfaces in Dutch drift-sand areas in the past century (Figure II.4, Figure II.5, Figure II.6). The colonization of barren surfaces may be the result of natural succession. Some authors suggest that the speed of succession is accelerated by atmospheric nitrogen deposition (Bakker et al., 2003; Nijssen et al., 2010). Most natural vegetation in The Netherlands is maintained or affected by management practices which also altered the vegetation development in the past century. The purpose of this study is partly to asses the impact of climate change on cover fractions and the distribution of plant functional types such as bare soil, mosses, lichens, grasses and shrubs. This assessment can only be carried out successfully if the effect of meteorological conditions on observed changes in vegetation coverage and species composition in time and space can be separated from alterations caused by anthropogenic influence and other factors such as game In search of the actual groundwater recharge Part II (PhD) KWR May 2011

45 tramping, grazing and sand burial. Natural succession and ageing of vegetation can also lead to changes in coverage and species composition, which blurs the effect of weather conditions. This review will point out the significance of different factors affecting the spatial distribution and dynamics of vegetation in coastal and inland sand dunes Figure II.4. Vegetation development at the Kootwijkerzand, The Netherlands. in the period (Source upper photo: staatsbosbeheer, source lower photographs: Topografische Dients Emmen; Eurosense, 2003 vide Oostra (2006)) In search of the actual groundwater recharge Part II (PhD) KWR May 2011

46 Figure II.5. Vegetation development at the Hulshorsterzand between 1900 and 1971 (after Van Ree (1993) vide Berendsen (2005)) In search of the actual groundwater recharge Part II (PhD) KWR May 2011