On the impact of 50-years-accumulated land-surface changes upon micrometeorological conditions

Transcription

1 29 On the impat of 5-years-aumulated land-surfae hanges upon mirometeorologial onditions Niole Mölders Abstrat The influene of the onurrent land-use hanges that took plae in East-Germany from the Thirties to Eighties on mirometeorologial onditions is examined. These land-use hanges lead to a slightly warmer and drier atmospheri boundary layer (ABL). Moreover, they may signifiantly modify the partitioning of inoming energy between sensible and latent heat as well as partitioning of loud-mirophysial proesses between the warm and old path of preipitation formation. The atmospheri response to onurrent land-use hanges deviates from superposition ( of the responses to simple land-use hanges) also in areas of nonsignifiant hanges. Zusammenfassung Der Einfluss der akkumulierten Landnutzungsänderungen, die zwishen den 3-er und 8-er Jahren in Ostdeutshland stattfanden, auf das Mikroklima wird untersuht. Diese Landnutzungsänderungen führen zu einer geringfügig wärmeren und trokeneren atmosphärishen Grenzshiht. Die Aufteilung der einfallendenden Energie in Flüsse fühlbarer und latenter Wärme sowie der wolkenmikrophysikalishen Prozesse in den warmen und kalten Pfad der Niedershlagsbildung ändern sih signifikant. Die atmosphärishe Response auf die akkumulierten Landnutzungsänderungen weiht auh in Gebieten ohne signifikante Änderungen von der Superposition der Auswirkungen einfaher Landnutzungsänderungen ab. 1. Introdution In reent years there has been a heightened awareness in the general publi of anthropogenially indued limate hanges. Climate hanges are the response to a sum of limate relevant aspets as astronomial, geophysial, hemial, and biospheri onditions as well as anthropogeni impats like emission of trae gases or land-use hanges. Out of the variety of possible human impats on limate, land-use hanges aused by subsidy politis, urbanization, deforestation, and afforestation seem to be of speial interest, not at least for water- and food resoure management. Thus, a lot of reent studies examine the impat of land-use hanges on global and regional limate for!arge areas, where land-use type A was onverted to B (e.g., Xue 1996), or where land-use hanged ontinental-wide (e.g., Copeland et al. 1996). Studies on limate-indued hanges of biome and their impat on monsoon exist for the!arge sale (e.g., Claussen 1997). Their results indiate that oherent regions of substantial hanges in sreen height temperature, humidity, wind speed, and preipitation may be the outome of meso-a-sale land-use hanges. The big hanges of the aforementioned quantities are losely related to the hanges of vegetation parameters. Due to ompliate interation and feedbak proesses between some of the limate determining onditions a prerequisite for better understanding limate is the knowledge of the influenes of various proesses and their responses to land-use hanges on limate. Consequently, in order to assess the impats of proesses on limate one has to look at their respetive temporal and spatial sales as well as on possible (nonlinearity of) feedbak. Reent landuse hanges usually take plae on loal sale and an alter, among other, albedo, roughness length or evaporative ondutivity. At the interfae earth-atmosphere, the fluxes of momentum, water, and energy are regulated by bio-geophysial proesses govemed by these quantities. These proesses have typial temporal sales of hours to a day. Thus, to examine the impat of land-use hanges on limate one has to understand how the altered surfae quantities

2 3 affet mirometeorologial (e.g., evapotranspiration) and meso-y/ß-sale proesses (e.g., reyling of water within an area). In East-Germany, urbanization, eonomi development, open-pit mining, drainage of marshland, reultivation of military areas, and modified agriultural praties appreiably altered the landsape (see Mölders 1999a Figs. 3.1, 3.3). Herein, as defined by Mölders ( 1999b) onurrent land-use hanges our, i.e. different land-use types A, B, C, et. are simultaneously onverted to various land-use types X, Y, Z et.. Thus, the impat on rnirolimate is examined for the onurrent land-use hanges that took plae in southern Brandenburg and northern Saxony between the Thirties and Eighties. 2. Model and data The non-hydrostati meteorologial model GESIMA used in this study is validated for a wide range of phenomena (e.g., Claussen 1988, Kapitza and Eppel 1992, Eppel et al. 1995, Devantier and Raabe 1996, Hinneburg and Tetzlaff 1996). Moreover, Mölders ( 1998) showed that GESIMA is able to simulate the atmospheri response to surfae harateristis usually observed. Soil wetness is determined by a fore-restore-method (Deardorff 1978). At the surfae, the fluxes of sensible and latent heat are alulated by a bulk-parameterization. Transpiration by plants is onsidered by bulk-stomatal ondution. Herein, the various landuse types are haraterized by varying plant physiology and physial parameters. Soil heat fluxes and soil temperatures are alulated by a one-dimensional heat-diffusion equation. Surfae stress and near-surfae fluxes of heat and water vapor are expressed in terms of dimensionless drag oeffiients and transfer oeffiients of heat and moisture by a parametri model (Kramm et al. 1995). The turbulent flux of momentum for the region above the surfaelayer is determined by a one-and-a-half-order losure sheme. Heterogeneity of preipitation and land-use is onsidered by an expliit subgrid sheme (Seth et al. 1994, Mölders et al. 1996). Herein, a fundamental assumption is that the subgrid-sale near-surfae meteorologial foring, whih is experiened by the surfae, is important in determining the net exhange of heat, moisture, and momentum at the earth-atmosphere interfae. Within the framework of the expliit subgrid sheme, eah atmospheri grid ell (5 x 5 km 2 ) falls into 25 subgrid ells (1 x 1 km 2 ) at the boundary earth-atmosphere and within the soil. The subgrid ells are assumed to be homogeneously overed by their individual vegetation over horizontally homogeneous soil types. Energy- and water fluxes are solved for eah of these subgrid ells with their own soil and near-surfae meteorologial foring. Arithmetially averaging the subgrid-fluxes realizes the oupling of the subgrid ells to the atmospheri grid ell. Note that the subgrid sheme allows to onsider land-use hanges on a realisti meso-y-sale size. A 5-lass (water vapor, loud water, rainwater, ie, graupel) bulk-parameterization sheme of loud mirophysis serves to desribe loud- and preipitation formation (Mölders et al. 1997). Radiative transfer is determined by a simplified two-stream method taking into aount the water substanes (Eppel et al. 1995). In nature, land-use hanges would possibly affet the profiles in the ABL. Moreover, the limate of the Thirties and Eighties slightly differs. For omparability and to avoid additional degrees of freedom, all simulations start with the same vertial profiles of wind, hurnidity, air- and soil temperature. The synopti ondition is orientated towards typial ases in spring, for whih loal reyling of previous preipitation ours. Simulations start at LT (loal time) with a geostrophi wind of 7.5 m/s from 1. Radiation is assumed for the 122nd day of a year. Surfae pressure, soil wetness fator, soil temperature in 1 m depth, and water surfae temperature are set equal to 13 hpa,.9, 285. K, and K, respetively. To determine the surfae harateristis of the Thirties histori maps are digitized (see Mölders 1999a Fig. 3.3). These maps originate from the turn of the 19th to 2th entury and were updated until Their graphi quality is partly limited for whih errors may result in digitizing. The last reently updated maps originate from the late Eighties (1986 to 1989).

3 Although more reent data would be available from satellite, the aforementioned maps are digitized (see Mölders 1999a Fig. 3.1) to ensure the same inquirement of data for the Thirties and Eighties. Hereafter, the land-use data of the Thirties and Eighties, the simulations performed with these data sets as well as their results are denoted S3 and S8, respetively. The histori terrain height is assumed to be equal to that of S8 exept for open-pit mines. In the maps of the Eighties, the terrain height before onset of mining ativities is also given for the mining areas. Thus, this terrain height is taken for S3 in those areas where mining exists in the Eighties, but not in the Thirties. The elevation of histori open-pit mines already reultivated in the Eighties is assumed to be 1 m below the average terrain height of the same 1 x 1 km 2 areas in the Eighties. Tab. 1. Perentage of the domain (3 km 2 ) that is of land-use type A in S3 (lines) and of land-use type B in S8 (olumns). Note that no marshland and heath exists on the resolution of the digitization ( 1 x 1 km 2 ). The value listed for equal land-use types in S3 and S8 represents the perentage of unhanged area overed with that land-use. The olumns and lines add up to the perentage of the various land-use types in S3 and S8, respetively. Land-use Marsh- Water Open-pit Grass- Agri- Heath Dei- Mixed Conife- Village City land mme land ulture duous f. forest rous f. Water Open-pit m Grass land Agriulture Deiduous f Mixed forest Coniferous f Village City In the north-eastern part of the domain, forest and grassland prevail, while in the south-western part, agriulturally used land dominates (see Mölders 1999a Figs. 3.1, 3.3). Open-pit mining exists in Südraum Leipzig ( outskirts of Leipzig), Bitterfelder Revier (mining distrit of Bitterfeld) and Lausitz. In the Eighties, only 54.1 % of land surfae over kept the same as in the Thirties (Tab. 1 ). In the histori maps, settlements, for instane, over an appreiably smaller fration of the domain, while grassland takes an appreiably larger fration as ompared to the maps of the Eighties (Tab. 1). The derease of overage by grassland may be explained by the limited graphi quality of the histori maps and by the fat that, in the maps of the Eighties, only water meadows are indiated as grassland. In the maps of the Eighties, deiduous and oniferous forests are seldom distinguished, for whih mixed forest ours more frequently than in the maps of the Thirties (Tab. 1). Moreover, in maps of the Eighties, heath (usually military areas) and marshland are of subgrid-sale with respet to the resolution of 1 x 1 m 2, and thus do not our in S8 (Tab. 1). In the Thirties, however, these land-use types over about 1 % of the domain eah. At that time, the river Elbe was not straightened. Thus, the Elbe takes a larger area in the Thirties than Eighties for whih the river is sometimes dominant on 1 x 1 km 2 -resolution (usually at the ost of grassland). Flooding of open-pit mines or artifiial water reservoirs lead to additional water in S8. Furthermore, mining ativity hanged from the exploitation of a lot of small (often subgrid-sale) open-pit mines to that of few, but large open-pit mines. 31

4 32 3. Design of the study Simulations are arried out alternatively applying the land-use data of the Thirties and Eighties. Additionally, nine simulations are performed for whih, in the landsape of the Thirties, land-use is altered aording to the distribution of one land-use type (e.g., grassland) in the Eighties. Besides omparing the results of S3 and S8, they are statistially analyzed by probability density funtions, student-t-tests, and similarity oeffiients. Applying the priniple of superposition serves to detet areas of nonlinearly mirolimati responses to onurrent land-use hanges. The statistial behavior of the response to the altered landsape is evaluated by frequeny distributions of simulated water- and energy fluxes, state quantities as well as loudand preipitating partiles. Probability density funtions are alulated for the entire simulation time using hourly data obtained by the different simulations. The probability density funtion of a quantity is given by (Olberg and Rak6zi 1984, Mölders et al. 1996) pdf(x) = P(X ~X~ x+m) IM (1) where p is the frequeny in the interval ex, x+m). The null-hypothesis is that land-use hanges provide no differenes in the atmospheri response, while the alternative hypothesis is that they do. For eah grid-point the individual sampling distributions applied are the hourly values of wind, temperature, and humidity (all taken at referene height), surfae temperature, soil wetness, energy- and water fluxes. In the ase of loud- and preipitation partiles, the hourly olumn-integrated values of the respetive mixing ratios serve as individual sampling distributions for eah grid-olumn. At a gridpoint/olumn, a hange in the aforementioned quantities will be regarded as statistially signifiant if data gives evidene against the null-hypothesis so onviningly that it would happen in no more than 1% of the ases. Following Ogunjemiyo et al. (1997) and Friedrih (1999), the proedure, firstly suggest by Jakson et al. (1989) for omparing data fields ontaining different parameters or for omparing data fields of same parameters that are reorded at different times, is adapted for omparing the results obtained by the simulations with the parameter sets representing the landsapes of the Thirties and Eighties. In so doing, the simulated quantities, X 1, (e.g., variables of state, fluxes, et.) at the grid-point j on the distribution field are transformed into a set of values z 1, by subtrating the domain-averaged value and normalizing the differene by the standard deviation of the differenes S (see Ogunjemiyo et al. 1997) X.-X Z.=-1 J s (2) The similarity between two transformed distributions is now established on the basis of similarity in the sign of Zrpairs (Ogunjemiyo et al. 1997) as m+n.\ =---- m+n+p (3) where Cs is the similarity oeffiient, n, p and m are the numbers of Zrpairs with negative, positive, and mixed signs. The similarity oeffiient ranges from zero (no similarity) to 1 (absolute agreement). Appliation of similarity oeffiients points out land-use hanges that yield similar mirolimati onditions with respet to the histori landsape. Furthermore, the time variane in these quantities aused by various land-use hanges an be found. Note that the similarity oeffiients between the landsape of the Thirties and Eighties with respet to

5 thermal ondutivity, heat apaity, em1ss1v1ty, albedo, roughness-length, field apaity, apillary, and maximal evaporative ondutivity amount.78,.77,.79,.75,.76,.81,.84, and.83, respetively. lt has to be expeted that the atmospheri response to onurrent land-use hanges may be enhaned or diminished as ompared to that resulting from simple land-use hanges. Herein, simple land-use hanges are defined as the hanges of various land-use types A, B, or C to only one land-use type D. In a linear response, the sum of the differentials in simulated quantities (e.g., water- and energy fluxes) aused by simple land-use hanges, would equal the differentials in the atmospheri response due to onurrent land-use hanges (priniple of superposition). After some algebra, one arrives at the general formula (Mölders 1999a) 33 n { > enhanement (n - 1) X/ -I. X/+ X/= ß. = superposition i=l < diminution. (4) Here, Xj denotes a given quantity at grid-point j. The index k represents the simulation with the landsape from whih the hanges start, namely S3, i = 1,..., n stand for the simulations with the landsapes wherein only one land-use type is altered as ompared to S3, and p represents the simulation with onurrent land-use hanges, namely S8. Enhanement means a positive deviation from the priniple of superposition (ß) and diminution is haraterized by a negative one (-ß). Only those deviations from superposition will be disussed as relevant whih, absolutely regarded, amount to more than the error that typially arises in routinely measuring the quantity, X 1 (i.e., lßl 2::.2 K for air temperature, lßl 2::.5 K for ground temperature, lßl 2::.5 mls for wind, lßl 2::.5 g/kg for humidity, and lßl 2:: 35 W/m 2 for the fluxes). 4. General aspets The onurrent land-use hanges do not affet the simulated quantities of state, nor the wind above the ABL, exept in areas of deep onvetion. In S8, the daily domain-average of temperature is.2 K higher, humidity is.14 g/kg drier, the u- and v-omponent of wind vetor are.6 mls and.1 m/s stronger, the w-omponent is.2 m!s weaker than in S3. Generally, predited fluxes vary more strongly during the day, when the energeti input is high, than at night. In both landsapes, the distributions of turbulent latent heat fluxes math those of the dominant land-use leading to a higher water supply to the atmosphere over the forest-dominated Northeast than the agriulturally-dominated Southwest. At noon, air temperatures of S3 reah higher values than in S8 beause of the lower loudiness at that time. At night, however, the near-surfae layer of S3 ools more strongly (about 1-2 K) over western Niederlausitz than in S8. In the domain-average, the upper ABL of S3 is warmer than in S8. The probability density funtions of near-surfae air temperatures show a trimodal distribution for S3 and S8 (Fig. 1 ). However, in S8, the seondary minimum ourring at about 8 C nearly reahes zero on the benefit for values of about 1 C (Fig. 1). This hange of near-surfae air temperature results from the higher frequeny of surfae temperatures of about l 5 C. The probability density distribution of ground temperatures beomes steeper in response to the onurrent land-use hanges (Fig. 1 ). Around noon, on average, near-surfae atmosphere is slightly drier in S3 than S8, espeially over Fläming and Niederlausitz. Exept Oberlausitz and northern Niederlausitz, the lower ABL is less humid in S3 than S8 at night. On the ontrary, over large areas, the upper ABL of S3 is meister than in S8 at night. Therefore, at night, loudiness inreases more strongly in S3 than S8. The probability of high near-surfae speifi humidity dereases due to the land-use hanges (Fig. 1 ). The signifiantly different preipitation distribution and

6 34 intensity of S8 (see also Fig. 2) lead to a shift of soil wetness fators towards lower values as ompared to S3 (Fig. 1)..5.4 ';:?.3 '-"... -g_.2.1 ~ Air temperature ( C). s.--ß:t...~~:...~~~-.,_,,i _ '-" a._ Soil wetness '-"... -g_ r-i.i61:'!:.'. ',_,_... ~~... ::~~' """" Ground temperature ( C) Water vapor mixing ratio (g/kg) Fig. 1. Probability density funtions of air temperature at referene height, TR. soil wetness fator, f, ground temperature, Ta, and speifi humidity at referene height, qv, (upper left to lower right) as determinedfor S3 (grey irles) and S8 (diamonds) _ E ~ 12 1 ü (l).._ _ :i (/) 2 : Lukenwald... Rl so... ~ ~! 1 :.~~:f~;,;~ Dresden west-east diretion (km) Fig h-aumulated preipitation as obtained by S3 ( grey shaded), and S8 ( dotted lines). Maximum values are 4.3 mm and 1.4 mmfor S3 and S8, respetively. Usually, afforestation, open-pit mining, and urbanization inrease, while deforestation redues surfae roughness, i.e. the onurrent land-use hanges alter the turbulent state of the atmosphere in both diretions. Nevertheless, none of the omponents of wind vetor is modified signifiantly. Urbanization, however, intensifies the vertial mixing. On the ontrary, in S8, the lower fration of forest leads to a redution of vertial mixing as ompared to S3. Although the distribution of vertial motions is strongly modified by the land-use hanges probability density funtions are hardly modified (therefore not shown). The land-use hanges

7 slightly alter the horizontal wind field leading to a lower probability of high v-omponents and a higher probability of high u-omponents. 35 (a), N 1 E 25 ~ Q).!: n Q) X ~ u j8=.1 _:" B= ~= Fluxes f sensible heat (wm- 2 ), N 1 E Q).!: 15 ~ '-" 2... n Q) X ~ t;: Fluxes f sensible heat (wm- 2 ) (b) Fig. 3. Jsolines of joint probability density funtions for the suifae fluxes of latent and sensible heat in ( a) 53 and (b) S8. Lines of onstant Bowen ratios are superimposed..15 ~.1 ""O...15 S' ~.1 ""O Net radiatin (w m- 2 ) Soil heat flux (W m- 2 ) Fig. 4. Probability densityfuntions ofnet radiation, Q (left), and soil heatfluxes, G (right) as detennined by S3 ( grey irles) and S8 ( diamonds ).

8 36 On domain-average, more water vapor is supplied to the atmosphere in 24 hours in S3 than S8. Usually, in S3, lower values our over heath or small open-pit mines than over the same areas in S8. Those areas overed by water in S3 and that are sealed or altered to openpit mines evapotranspire less in S8. Evapotranspiration rates also go down for a hange from grassland to agriulture. On the ontrary, the opposite land-use hange inreases the water vapor supply to the atmosphere. Evapotranspiration also differs where great disrepanies our in the diurnal ourse of loudiness resulting from the land-use hanges. The maximum 24h-aumulated evapotranspiration of S3 (2.93 mm) exeeds that of S8 (1.98 mm), and the minimum 24h-aumulated evapotranspiration of S3 (.11 mm) is less than in S8 (.2 mm). Maximum derease and inrease amounts -2.6 mm and 2.5 mm, respetively. The probability of intermedium evapotranspiration rates dereases, while that of high and low evapotranspiration rates slightly inreases for S8 as ompared to S3 (Fig. 3). Compared with S3, the probability density funtions of sensible heat fluxes slightly shift towards lower values for S8 (Fig. 3 ). This means that the inoming energy is differently partitioned between sensible and latent heat in S3 and S8 (Figs. 3, 4). In S8, the probability density funtions of soil heat fluxes shift towards lower values as ompared to S3 (Fig. 4). While in S8, the probability density distribution of net radiation is smoothed, that of soil heat fluxes gets more strutured (Fig. 4). This effet partly results from the altered loudiness. The probability density funtions of soil heat fluxes and net radiation are more strongly altered by the land-use hanges than those of latent and sensible heat fluxes (Figs. 3, 4). Looking on the distributions of net radiation shows that net radiation reats very sensitively to altered loudiness ;;- 1. u-...,... ""O Water vapor mixing ratio (g/kg) Cloud water mixing ratio (g/kg) ;:. 1..., u-..., O'"... ""O.1 :.1.. a o~ * <> '% le mixing ratio (g/kg) Rainwater mixing ratio (g/kg) Fig. 5. Probability density funtions ofwater vapor, qv, loud water, q, ie, qi, and rainwater, qn within olumn (upper left to lower right) as determinedfor S3 (grey irles) and S8 (diamonds). Note that y-axis is logarithmi. At night in S8, louds exist only over the water meadows of Elbe at Riesa and over Oberlausitz. In S3, however, two largely East-West-orientated loud bands exist in the northern and southern part of the domain at that time. These loud bands still ontain slight amounts of rainwater at some loations, while no rainwater exists in S8.

9 In the morning, early afternoon, and at night, loudiness of S3 exeeds that of S8. On the ontrary, in the late afternoon, it is usually less than in S8. The enhaned evapotranspiration of S3 ourring in the afternoon ontributes to the inreased preipitation in the late evening hours and to inreased loudiness at night as ompared to S8. In the morning, the larger loudiness of S3 results from the loally more unstable stratifiation as ompared to S8. The altered degree of heterogeneity and related modified aerodynami roughness also ontribute to great differenes in the distributions of louds and preipitation. The domainaveraged aumulated preipitation of S3 exeeds that of S8 (Fig. 2). Nevertheless, S8 provides a more shower-like preipitation with Ioally higher intensities as ompared to S3. The latter finding indiates that besides a redution of loud- and preipitating partiles the mirophysial paths of preipitation formation are affeted by onurrent landuse hanges. Herein, the altered heating, vertial mixing, and moisture supply ontribute to these differenes. At day, more ie is formed in the moister and ooler ABL of S3 than S8 (Fig. 5). Consequently, in S3, sedimentation of ie inreases. Therefore, ie quikly reahes areas of the ABL with temperatures warmer than the freezing point and melts. Thus, more rainwater is built via the old path of preipitation formation in S3.than S8. Consequently, at some loations, preipitation sets on about 2 h earlier in S3 than S8. Sine in the warmer louds of S8 oalesene and riming are favored, as ompaped with S3, in S8, a more shower-like preipitation may establish in areas of strong moisture onvergene. The probability density funtions of water vapor, loud- and preipitating partiles shift towards lower values for the onurrent land-use hanges from S3 to S8 (Fig. 5). Consequently, the probability density funtions of preipitation and soil wetness shift towards lower values, too (see also Fig. 1). 5. Similarity, enhanements, and diminution On average, the greatest dissimilarity of air temperatures at referene height ours for simple land-use hanges in favor of agriulture or oniferous forest, followed by onurrent land-use hanges (Fig. 6). At referene height humidity of S3 is the most dissimilar to that of the simulations with simple land-use hanges to water, grassland, agriulture or deiduous forest, and S8 (Fig. 6). Speifi humidity and air temperatures at referene height are less similar in the late afternoon and after sunset than at the other hours of the day (Fig. 6). These greater dissimilarities result from seondary effets, namely evaporation of rainwater and related evaporative ooling in S3. These phase transition proesses our less intensive in S8 than S3 at that time. Applying the priniple of superposition points out a strong nonlinear behavior of air temperatures at referene height in Niederlausitz at noon (Fig. 7). Here, diminution and enhanement alternate on a distane of less than 1 km. One, diminution ours over areas dominated by agriulture and grassland in the Thirties and by agriulture only in the Eighties (Fig. 7). In the other ase, diminution ours over areas mainly overed by forest in the Thirties that partly hanged to lakes, agriulture, and settlements (Fig. 7). On the ontrary, at referene height air temperatures are enhaned in an area where the forests still existing in the Thirties are already deforested in the Eighties. Enhanement ours in an area ( enter at 12 km, 11 km) where only slight afforestation takes plae in the meanwhile from the Thirties to Eighties. At night at referene height, air temperatures positively deviate (up to 1.1 K) from superposition in Fläming, while negative deviations (loally up to -5.9 K) our in nearly the rest of the domain. Note that, in an environment dominated by agriulture, urbanization tends to ause slightly higher temperatures than expeted from superposition (Fig. 7). 37

10 ~._.~ '< Q) (.).75 1::- ;::.5 E (/') prelpltotion 1i--~~~~~~~~~--~~,_ ~::::-:: :.:. :...:..: :..; ~ :.:.;..-~-~- speifi humidity ot referene height oir temperature ot referene height 1 r---~-----~~~.;:;: Simulation time (h) Fig. 6. Temporal development of similarity oeffiients as obtained for preipitation, speifi humidity and air temperature both at referene height for the onurrent land-use hanges from S3 to S8 (grey solid lines), for simple land-use hanges in favor of water (grey dotted lines), grassland (grey dashed lines), agriulture (blak solid lines), deiduous forest (blak dashed lines), and oniferousforest (blak dotted lines), respetively., E 12.::;. '---' 1 8 :E :i "' 2 Leipzig <> --., /\ west-east diretion (km) 2 Fig. 7. Differentials ß aording to Eq. 4 for air temperatures at referene height (in K) at 12 LT. Dark-grey boxes indiate grid ells with hanges at the 9%- or better statistialsignifiane Level, showing that the differenes in air temperature result from the onurrent land-use hanges. At noon at referene height, humidity diminishes (up to -.4 g/kg) around Lindenberg. Here, deforestation, espeially in favor of agriulture, and settlements, took plae. Moreover, seondary differenes result from modified loudiness. In the agriulturally dominated part of the domain, however, humidity at referene height hardly deviated from superposition. At night, in Fläming, near-surfae humidity is up to.9 g/kg lower than expeted from superposition (Fig. 8). In Niederlausitz, the environments of Leipzig and Lindenberg, humidity at referene height exeeds the values expeted for a linear behavior (up to 1.6 g/kg at maximum, Fig. 7). Differenes in evapotranspiration, evaporation of raindrops, and related ooling mainly ontribute to these differenes due to the nonlinear relationship of speifi saturation and air temperature. Positive deviation from superposition ours for air

11 temperatures at referene height in Fläming at night and in Niederlausitz at day. Negative deviation is deteted along the river Elbe at day and night and in Niederlausitz at night. 39,, E..,..::.: 12 1 ü <ll... "'O 8 E L -:::i 4 (/) west-east diretion (km) Fig. 8. Like Fig. 7, butfor humidity at referene height (in glkg) at 24 LT. While there is linear behavior for soil wetness before onset of preipitation, there is nonlinearity afterwards. After 24h of integration soil wetness of SSO is enhaned as ompared to the superposition of the differenes of simple land-use hanges. The predited wind fields show higher similarity before and around noon than after that time. In S3, the evaporative ooling resulting from evaporation of rainwater and settling of rainwater ontributes to downward motions (interation mirophysis dynamis) after onset of preipitation. Consequently, then the similarity of the fields of vertial motions dereases. At that time seondary differenes in the horizontal wind field result by outflow of air in the desending areas of preipitating louds and redue similarity between S3 and SSO.,, E..,..::.: 12 1 ü <ll... "'O 8 L L 4 -:::i (/) west-east diretion (km) 2 Fig. 9. Like Fig. 7, butfor the latent heatfluxes (in W/m 2 ). Until the late afternoon, S8 shows the lowest similarity to S3 for the horizontal omponents of wind vetor. Lateran, however, the simulations with a simple land-use hange in favor of oniferous forest or agriulture provide lower similarities to S3 in the ase of the v- and w omponent of wind vetor than the onurrent land-use hanges. This means that the land-use hanges in favor of oniferous forest to a high degree ontribute to the great hanges in vertial motions and that their effet is mitigated by the other onurrent land-use hanges in

12 4 S8. As ompared to superposition, in Fläming, the u-omponent and, in Niederlausitz, the v omponent of wind vetor are enhaned where espeially afforestation ourred. At noon, latent and sensible heat fluxes are diminished (about more than 21 W/m 2 ) in the southern Niederlausitz (e.g., Fig. 9). At night, sensible heat fluxes positively deviate (up to 39 W /m 2 ) from superposition in Sähsishe Bergland. On the ontrary, latent heat fluxes negatively deviate (up to -69 W/m 2 ) near Dresden. Note that there exists nonlinear behavior in areas of nonsignifiant hanges and vie versa. On average, the predited distributions of sensible and latent heat fluxes are more similar than those of net radiation or soil heat fluxes (Fig. 1). For the latter the greatest dissimilarities exist between S3 and S8. Note that net radiation and soil heat fluxes reat more sensitive to altered loudiness than to land-use hanges or than the latent and sensible heat fluxes. Looking on the similarity oeffiients obtained for omparing S3 to the results of simulations with simple land-use hanges shows that land-use hanges in favor of agriulture and oniferous forest mainly ontribute to the dissimilarities (Fig. 1). On average, the obtained fluxes agree better with those yielded by S3 in the early morning than in late afternoon and early evening (Fig. 1) r"=~-.:::::::- latent heot flux ::-::-~-e::~ ~~~--~.. ~:::::~.::::~:.:::.~:...-~ -::-:.-::-::".::=:.::~:g,::,~~:...:z. ~-#.::,;;,:;::...;,;-;:;;;; -1;, ~~ sensible heat flux "::: ~. 7~ ~--:!3-~"":::"':: ""~ "'::: -~--~ :e:"' ~--:e<::~.:::.y!-.::::": :.::.!: ::.:~~ ~..:.-:-=-~:::.:..:=:~:::- ~=-:::. ::..:;;~-~:::;.,:;;::..,_~ ~----1.,_ ~.5 (.) 2:: ;:.75.E o.5 soil heot flux - -.,,,..,... ~ net-radiotion Fig. 1. Like Fig. 6, but for the suifae fluxes Simulation time (h) ie.75.5 Q) :Q rainwoter.,_.,_ Q) (.).75 ;:.E V1.5 loud water - ~ Simulation time (h) Fig. 11. Like Fig. 6, but for the loud- and preipitating partiles.

13 41 At noon, soil heat fluxes of S8 are up to 195 W /m 2 higher than expeted from superposition in southem Niederlausitz. Positive deviations from superposition also exist in the Lindenberg area. At night, soil heat fluxes are enhaned (more than 35 W/m 2 ) in Sähsishe Bergland. They are dirninished in Fläming (not shown). Despite the great hanges in loudiness, no relevant deviations from superposition exist for net radiation. Comparing S3 and S8, the olumn-averaged similarity oeffiients of loud- and preipitating partiles indiate that the predited ie mixing ratios are more similar than those of loud water and rainwater (Fig. 11). The similarity oeffiients indiate the strong impat of onurrent land-use hanges and simple land-use hanges in favor of oniferous forest, grassland or agriulture, on loud-water (Fig. 11). The dissimilarities grow with inreasing simulation time after onset of preipitation (Fig. 11 ). After onset of preipitation, the preipitation fields predited by the simulations with simple land-use hanges in favor of mixed forest, open-pit mines, water, grassland, deiduous forest, village, and ity broadly agree with those of S3 (see also Fig. 6). On the ontrary, there is little agreement between the preipitation obtained by S3 and that yielded by the simulations with simple land-use hanges in favor of oniferous forest or agriulture and by S8 (Fig. 6). Positive and negative deviations from superposition our in Niederlausitz for ie whih may be explained by seondary effets. The altered surfae onditions of S8, namely, lead to a redued moisture and an enhaned heat supply to the atmosphere. Thus, loudiness dereases. Espeially, the partitioning between the warm and old path of loud mirophysis shifts towards a higher preferene of the warm phase path. The modified loud partiles affet insolation and again surfae fluxes. If loudiness dereases in S8 as ompared to S3 evapotranspiration may inrease, but will be often less than in S3. Due to the ompliate feedbak between evapotranspiration and loudiness the temporal development of these quantities appreiably differs, espeially at the ost of the ie phase. The high nonlinearity results from the nonlinear relation of temperature and saturation as well as from the different (temperature dependent) saturation above ie and water, respetively. 6. Conlusions Numerial simulations alternatively assuming landsapes of the Thirties and Eighties are arried out at time-sales relevant for mirometeorologial proesses. The impat of these onurrent land-use hanges on mirolimatially relevant proesses is examined. Additionally, sensitivity studies are performed wherein, in the landsape of the Thirties, one land-use type is altered aording to its distribution in the Eighties. There is a 9% (and better) onfidene that the onurrent land-use hanges lead to hanges in loud- and preipitating partiles, soil wetness fators, latent and sensible heat fluxes, air- and surfae temperatures within the lee-side regions of their ourrene. Out of all examined quantities, loud- and preipitation partiles reat the most sensitive to the land-use hanges, followed by preipitation, and soil moisture. Moreover, their modifiation is not restrited to the areas of altered land-use. Generally, soil wetness fators grow signifiantly where preipitation is more plentiful in one of the landsapes or where it only ours in one of them. In these areas, other quantities experiene signifiant hanges due to ompliate feedbak proesses, i.e. seondary differenes establish. Consequently, the onurrent landuse hanges lead to distintly different near-surfae mirometeorologial onditions. Appliation of the priniple of superposition shows that Lausitz and Fläming (both dorninated by high vegetation and some grassland) are more sensitive to land-use hanges than the other parts of the domain. Looking on the areas that experiene great deviation from superposition suggests that the path-size of equal land-use resulting from the land-use hanges is deisive to get nonlinear response.

14 42 The results of simulations with simple land-use hanges show that the magnitude of atmospheri response to land-use hanges does not neessarily depend on the fration of domain that experienes land-use onversions. In addition to the size of pathes where the individual hanges take plae, the ontrast in hydrologi and thermal behavior of the hanges is deisive for the magnitude of response. Here, espeially, the altered albedo and roughness length play a role. Diminution and enhanement our also in areas without a signifiant hange in the respetive quantity. Deviations from superposition of surfae temperature, sensible heat flux, soil heat flux, and net radiation also our in areas where large (and sometimes signifiant) differenes our in loudiness. The temporal ourse of the energy- and water yle may be nonlinearly influened by onurrent land-use hanges and also by seondary effets. The same kind of land-use hanges may provide another atmospheri response in different surroundings (see Niederlausitz). Thus, one has to onlude that the prevailing land-use adjaent to onverted areas influenes the effets of land-use hanges. Consequently, the landuse that prevails in the environment of foreseen land-use hanges has also to be onsidered in authorization proedures. The signifiant differenes found for near-surfae air- and surfae temperatures, preipitation, and loudiness suggest that in areas of big land-use onversions, hanges in these quantities ould be detetable if stations were installed at these sites. Thus, here and in regions of great nonlinear response, limate stations should be installed to evidene the impat of land-use hanges on limate by observational data. Unfortunately, histori data are rare. Nevertheless, the deadal means of air temperature observed at Dresden and Görlitz (Lausitz) deliver the same ratio between the value of the Thirties and Eighties. The deadal mean preipitation, however, dereases for Görlitz, while it inreases for Dresden (Meteorologisher Dienst, 1955, ). To improve an understanding of the impat of onurrent land-use hanges on loal limate, future studies should examine whether onurrent land-use hanges provide a different response to the atmosphere under different limate onditions. Moreover, it has to be examined to whih degree land-use hanges ontribute to reent limate hanges. Therefore, one should examine whether the nonlinearity and signifiane of the impats also exist in the long-term. If suh hints are deteted, regional limate simulations will urgently require sophistiated biome models to orretly evaluate limati impats on water resoures. Additionally, unertainty analysis on the influene of anthropogeni land-use hanges and examination of the relative ontribution of these land-use hanges to limate hange are urgently needed. 7. Aknowledgments 1 would like to express my thanks to K. Fröhlih and K. Friedrih for digitizing the land-use data. 1 thank G. Kramm, K.E. Erdmann and M. Simmel for fruitful disussions and helpful omments. Thanks also the DFG for finanial support of this study under ontrats Mo77/l- 1 and Mo77/l-2, respetively. Referenes Claussen, M., On the surfae energy budget of oastal zones with tidal flats. Contrib. Atmos. Phys. 61, Claussen, M., Modeling bio-geophysial feedbak in the Afrian and Indian monsoon region. Climate Dynamis 13, Copeland, J.H., R.A. Pielke, and T.G.F. Kittel, Potential limati impats of vegetation hange: A regional modeling study. J. Geophys. Res. 11D, Deardorff, J.W Effiient predition of ground surfae temperature and moisture, with inlusion of a layer of vegetation. J. Geophys. Res. 84C,

15 43 Devantier, R. and A. Raabe, Appliation of a quasispetral loud parameterization sheme to a mesosale snowfall event over the Balti Sea. Contrib. Atlnos. Phys. 69, Eppel, D.P., H. Kapitza, M. Claussen, D. Jaob, W. Koh, L. Levkov, H.-T. Mengelkamp, and N. Werrmann, The non-hydrostati mesosale model GESIMA. Part II: Parameterizations and appliations. Contrib. Atlnos. Phys. 68, Friedrih, K., Numerishe Untersuhungen zur Sensitivität des Bowen-Verhältnisses. Master Thesis, Inst. Meteorologie, Univ. Leipzig (available from the author, in German). Hinneburg, D. and G. Tetzlaff, Calulated wind limatology of the South-Saxonian/North-Czeh mountain topography inluding an improved resolution of mountains. Ann. Geophysiae 14, Jakson, D. A., K.M. Somers, H.H. Harvey, Similarity oeffiients: Measures of oourrene and assoiation or simply measures of ourrene? Am. Natur. 133, Kapitza, H. and D.P. Eppel, The non-hydrostati mesosale model GESIMA. Part I: Dynamial equations and tests. Contrib. Phys. Atlnos. 65, Kramm, G., R. Dlugi, G.J. Dollard, T. Foken, N. Mölders, H. Müller, W. Seiler, and H. Sievering, On the dry deposition of ozone and reative nitrogen ompounds. Atlnos. Environ. 29, Meteorologisher Dienst, 195, Mölders, N., A. Raabe, and G. Tetzlaff, A omparison of two strategies on land surfae heterogeneity used in a mesosale ß meteorologial model. Tellus 48A, Mölders, N., G. Kramm, M. Laube, and A. Raabe, On the influene of bulk-parameterization shemes of loud mirophysis on the predited water-yle relevant quantities - a ase study. Met. Zeitshr. 6, Mölders, N., Landsape hanges over a region in Bast Germany and their impat upon the proesses of its atmospheri water-yle. Meteor. Atlnos. Phys. 68, Mölders, N., 1999a. Einfahe und akkumulierte Landnutzungsänderungen und ihre Auswirkungen auf Evapotranspiration, Wolken- und Niedershlagsbildung. Wiss. Mitt. Leipzig 15, Habil.Shrift. Mölders, N., 1999b. On the atmospheri response to urbanization and open-pit mining under various geostrophi wind onditions. Meteor. Atlnos. Phys. 71, Ogunjemiyo, S., P.H. Shuepp, I. MaPherson, and R.L. Dejardins, Analysis of flux maps versus surfae harateristis from Twin Otter grid flights in BOREAS J. Geophys. Res. 12, Olberg, M., F. Rak6zi, Informationstheorie in der Meteorologie und Geophysik. Akademie Verlag, Berlin. Seth, A., F. Giorgi, and R.E. Dikinson, Simulating fluxes from heterogeneous land surfaes: expliit subgrid method employing the biosphere-atmosphere transfer sheme (BATS). J. Geophys. Res. 99D, Xue, Y., The impat of desertifiation in the Mongolian and the inner Mongolian grassland on the regional limate. J. Climate 9, Dr. habil. Niole Mölders, Institut für Meteorologie Universität Leipzig Stephanstraße Leipzig moelders@urie.meteo.uni-leipzig.de