Journal of Hydrology 375 (2009) 204–216

Contents lists available at ScienceDirect

Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol

Towards an understanding of coupled physical and biological processes in the cultivated Sahel – 1. Energy and water David Ramier a, Nicolas Boulain a, Bernard Cappelaere a,*, Franck Timouk b, Manon Rabanit a, Colin R. Lloyd c, Stéphane Boubkraoui d, Frédéric Métayer d, Luc Descroix d, Vincent Wawrzyniak a a

IRD - Hydrosciences, Montpellier, BP 64501, 34394 Montpellier Cedex 5, France IRD/CESBIO, 18 Avenue Edouard Belin, bpi 2801, 31401 Toulouse Cedex 9, France c CEH, Crowmarsh Gifford, Wallingford OX10 8BB, UK d IRD/LTHE, BP 53, 38041 Grenoble Cedex 09, France b

a r t i c l e

i n f o

Keywords: Eddy covariance Surface energy budget Hydrologic cycle Millet Fallow savanna Semi-arid

s u m m a r y This paper presents an analysis of the coupled cycling of energy and water by semi-arid Sahelian surfaces, based on two years of continuous vertical flux measurements from two homogeneous recording stations in the Wankama catchment, in the West Niger meso-site of the AMMA project. The two stations, sited in a millet field and in a semi-natural fallow savanna plot, sample the two dominant land cover types in this area typical of the cultivated Sahel. The 2-year study period enables an analysis of seasonal variations over two full wet–dry seasons cycles, characterized by two contrasted rain seasons that allow capturing a part of the interannual variability. All components of the surface energy budget (four-component radiation budget, soil heat flux and temperature, eddy fluxes) are measured independently, allowing for a quality check through analysis of the energy balance closure. Water cycle monitoring includes rainfall, evapotranspiration (from vapour eddy flux), and soil moisture at six depths. The main modes of observed variability are described, for the various energy and hydrological variables investigated. Results point to the dominant role of water in the energy cycle variability, be it seasonal, interannual, or between land cover types. Rainfall is responsible for nearly as much seasonal variations of most energy-related variables as solar forcing. Depending on water availability and plant requirements, evapotranspiration pre-empts the energy available from surface forcing radiation, over the other dependent processes (sensible and ground heat, outgoing long wave radiation). In the water budget, preemption by evapotranspiration leads to very large variability in soil moisture and in deep percolation, seasonally, interannually, and between vegetation types. The wetter 2006 season produced more evapotranspiration than 2005 from the fallow but not from the millet site, reflecting differences in plant development. Rain-season evapotranspiration is nearly always lower at the millet site. Higher soil moisture at this site suggests that this difference arises from lower vegetation requirements rather than from lower infiltration/higher runoff. This difference is partly compensated for during the next dry season. Effects of water and vegetation on the energy budget appear to occur more through latent heat than through albedo. A large part of albedo variability comes from soil wetting and drying. Prior to the onset of monsoon rain, the change in air mass temperature and wind produces, through modulation of sensible heat, a marked chilling effect on the components of the surface energy budget. Ó 2008 Elsevier B.V. All rights reserved.

Introduction The role of rapidly changing Sahelian land surfaces in the dynamics of the West African monsoon system is generally thought to be major, but still very poorly understood. Since the pioneering work of Charney (1975), effects on climate of land use (Xue and Shukla, 1993) or soil moisture (Koster et al., 2004) in this * Corresponding author. Tel.: +33 467 149 017; fax: +33 467 144 774. E-mail address: [email protected] (B. Cappelaere). 0022-1694/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2008.12.002

semi-arid area have been suggested through the use of GCMs. A major effort towards a better understanding of these links has been undertaken under the African Monsoon Multidisciplinary Analyses project (AMMA; Redelsperger et al., 2006). Such an objective requires to significantly improve our knowledge of the functioning of the terrestrial subsystem itself, in which complex physical and biological processes are tightly coupled. To that aim, an important component of the project is to carry out a comprehensive and coherent in-situ surveying of the fluxes and stocks of energy and matter that compose the basic geo–eco cycles in this area. Data

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

sets of this kind are key to supporting in-depth analyses of the land surface exchange processes and associated model development/ validation. For understandable reasons, they are particularly scarce and difficult to acquire in Africa. In the West Niger area of the cultivated Sahel, the HAPEX-Sahel experiment of 1992 (Goutorbe et al., 1997) yielded abundant data and study results (Dolman et al., 1997), but denser, larger in area, longer in time, and more consistent sampling of more variables are now needed to progress towards objectives. HAPEX-Sahel sampling was generally limited in time to only very short periods at the end of the 1992 rain season, and spatial location of the various types of instruments was rather dispersed, making integration of variables difficult. This paper presents an analysis of year-round dynamics of the local energy and water cycles and their interactions, for the two land cover types that now dominate in this area, namely semi-natural and cultivated. This analysis is based on the first 2 years of data from the experimental setup installed in West Niger in 2005, more specifically from two surface flux stations and associated instruments located on a fallow and a millet site, respectively. HAPEX-Sahel results (e.g., Lloyd, 1995; Verhoef et al., 1996; Gash et al., 1997; Kabat et al., 1997; Lloyd et al., 1997; Braud, 1998) pointed to a strong variability in surface exchanges but suffered from the limitations mentioned above. A first driver of this flux variability is the variability of rainfall. It is particularly large in the region at any timescale including annual (Lebel et al., 2003), and is a major factor of seasonality with a short wet season and a long, totally-dry season. A dataset that covers two climatically contrasted years, hence duplicating each season, offers a first insight into this variability. Land use is another important factor of variability for these geo–eco cycles, which is investigated here through thorough, homogeneous monitoring of the two major land cover types. Besides the use of more accurate gas analysers, this extended sampling distinguishes the work presented here from the few other previous experiments in the region (e.g., Wallace et al., 1991, in Niger; Schüttemeyer et al., 2006, and Bagayoko et al., 2007, under the more humid Sudanian climate of the Volta Basin in Ghana and Burkina-Faso, respectively). The seasonal vari-

205

ation of the surface energy balance was analysed by Verhoef et al. (1999) for a tiger-bush and a savanna site in Niger, but only partially from direct measurements. In many regards, the dataset being acquired as part of the AMMA programme in West Niger stands as one of the largest and richest surface-energy data set in West Africa. Together with the energy and water cycles, those of vegetation and carbon, all in direct interaction, are investigated jointly, and presented in the companion paper by Boulain et al. (2009). While this new dataset serves here as the basis for a data-driven analysis of physical and biological processes involved in the control of the local energy and water cycles, it is also a major information source for other works of the AMMA programme in this area, including modelling (e.g., Boone and de Rosnay, 2007; Pellarin et al., 2009; Saux-Picart et al., accepted for publication; Saux-Picart et al., 2009) or scaling studies (Boulain et al., 2009; Ezzahar et al., 2009). Unlike most previous works in the region, which focused on specific components or aspects of these cycles at the land surface, this paper presents a more global and comprehensive view on their coupled dynamics, offered by the dataset. Materials and methods Study area The study area is part of the Wankama catchment (Fig. 1; 2.4°E, 13.4°N), 60 km east of Niamey, Niger. This semi-arid site belongs to the AMMA-Niger observatory (Cappelaere et al., 2009), one of three meso-sites along the West African latitudinal transect (Lebel et al., 2009). It is a 1.9 km2 endoreic catchment, typical of the cultivated Sahel (Peugeot et al., 2003). It essentially consists of a sandy hillslope (Fig. 1d) with slopes below 2%. Precipitation is limited to a short monsoon season from June to September, with an annual mean of 560 mm for the years 1905–2004 in Niamey (standard deviation: 135 mm), and a very strong year-to-year and spatial variability. Like the whole Sahel, the region was hit by a severe

Fig. 1. The Wankama catchment in the AMMA-Niger meso-site. Location of meso-site (square) in: (a) West Africa; (b) the Niger Republic; (c) Map of Wankama catchment with location of EC systems (squares); and (d) Toposequence of Wankama catchment.

206

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

drought in the 1970s and 1980s, with an average rain shortage as high as 30%. The more recent period shows signs of a return to a more normal interannual mean, but with still only 479 mm/year in Wankama over the 1992–2006 period (standard deviation: 90 mm; Fig. 2). Rainfall occurs as short, intense convective storms, which can produce strong, ephemeral surface runoff, flowing down to a temporary endoreic pond at the catchment outlet (Fig. 1c,d). Depth to the water table varies from 15 to 60 m. Extensive rain-fed cultivation of millet is by far the main, if not the only crop grown. Millet fields presently cover 58% of the catchment’s surface area. Traditional techniques are used, with little chemical fertilizer and no pesticides. Shrubs and weeds are manually eliminated before the rain season. Some trees are preserved along field edges. Hand-hoe weeding may be repeated once or twice in the season, but some grass often persists. Sowing starts with the first 5–10 mm rainfall (enough to wet the first 5 cm of soil), and can be repeated several times as long as rainfall remains erratic. It is performed as pockets, with a 1-m to 1.4-m spacing. Mature millet is 2–3 m high at the study site. Harvest is done shortly after the end of the rain season, in September or October, removing the whole plant (ear and stem) from the field, and leaving bare soil through the dry season. Significant fallow periods are necessary in this traditional agricultural system. In the catchment, fallow now represents 23% of the total surface area. Most fallow fields are no more than five-year old, which is now typical of the Niamey region. They include a shrub layer with Guiera senegalensis as the dominant species (Meir et al., 2007), and an annual grass layer whose composition mostly

depends on the rain distribution at the season beginning. The average height of the shrub layer is 2 m, against 0.6 m for the grass layer. Shrub density averages around 700 individuals per hectare. Trees exist as isolated individuals or small groups. Fallows represent the main energy supply for the local population. More details on the vegetation cover can be found in Boulain et al. (2009). Experimental setup Instruments used for this paper are the following: (i) two eddy covariance (EC) stations, measuring surface fluxes of sensible heat, latent heat (i.e., evapotranspiration), momentum and CO2, together with 4-component radiation, soil heat flux and profiles of soil temperature and moisture, for the two dominant land cover types (millet crop and fallow savanna; Fig. 1); (ii) a tipping-bucket recording rain gauge, midway between the two stations; and (iii) eight vegetation monitoring plots, scattered over the catchment according to land cover types (for description and methodology, see Boulain et al., 2009). This setup started operating in June 2005. Table 1 summarizes the main characteristics of the duplicated EC systems and associated instruments. EC data was processed with the EdiRe software (Version 1.4.3.1167, R. Clement, University of Edinburgh), based on CarboEurope recommendations (Mauder and Foken, 2004), including despiking, double rotation, cross-correlation for derivation of time lag between the sonic anemometer and the gas analyser, spectral corrections, Webb corrections and atmospheric stability test, with no gap filling. Fluxes as well as all other energy and associated data were averaged over 30-min periods. Direct

Cumulated rain (mm)

700 600

2006

500

2005

400 300 200 100 0 april

may

june

july

august

september

october

Fig. 2. Cumulative rainfall at Wankama for the 15 rain seasons of the 1992–2006 period (2005: thick grey line; 2006: thick black line; other years: thin grey lines).

Table 1 Characteristics of the eddy covariance (EC) system and associated instruments at both sites (millet crop and fallow savanna). Instrument

Measurements

Height or depth

Storage interval

Above ground Campbell CSAT-3 sonic anemometer (Campbell Scientific, Inc., Logan, USA) LI-COR LI-7500 infrared gas analyser (LI-COR Biosciences, Lincoln, USA) Kipp & Zonen CNR1 radiometer (Kipp & Zonen, Delft, The Netherlands) Vaisala HMP45C (Vaisala Oyj, Helsinki, Finland)

3D wind speed and direction Sonic air temperature CO2 and H2O concentrations Air pressure Shortwave (0.3–2.8 lm) and longwave (5–50 lm) incoming and outcoming radiation Air temperature and relative humidity

5.1 m (millet) and 4.95 m (fallow) 5.1 m (millet) and 4.95 m (fallow) 2.5 m (millet) and 3.4 m (fallow) 2m

20 Hz

Soil measurements Campbell CS616 water content reflectometers (6)

Soil volumetric water content

Campbell T108 temperature probes (6)

Soil temperature

Hukseflux HFP01SC heat flux plates (3, averaged) (Hukseflux, Delft, The Netherlands)

Surface soil heat flux

.1, .5, 1, 1.5, 2 and 2.5 m .1, .5, 1, 1.5, 2, and 2.5 m .05 m

20 Hz 1 min 1 min 1 min 1 min 1 min

207

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

measurements of soil heat flux were checked against estimates at 5 cm derived from soil temperature gradients, using surface temperature inverted from measured outgoing longwave radiation. Study period and meteorological setting Results presented in this paper cover the mid-June 2005–midJune 2007 period, thereby including two rain seasons (2005 and 2006) as well as the two subsequent dry seasons. Annual rainfall was 495 mm from 47 events in 2005, against 572 mm in the 42 events of 2006 (Fig. 2). This makes the two wet seasons contrasted in terms of total rainfall (difference amounting to one standard deviation of the 1992–2006 period) and of mean event depth. Contrasts are even larger with respect to rain distribution along the season (Fig. 2). The 2005 season had a rather early start, while 2006 had a very late one as well as an earlier end, making the 2006 season very short (115 days, i.e., one third less than the 165 days of 2005). Hence, the 2005 wet season was characterized by comparatively smaller, sparser rain events, separated by longer

Air temperature (°C)

a

dry spells. In contrast to the millet crop, this rain pattern had significant adverse effects on the natural vegetation, as compared to 2006 (Boulain et al., 2009). Maximum event rainfall was 69 mm in 2006, against 49 mm in 2005, both in the wet month of August. Variations of air temperature, relative humidity, and vapour pressure deficit (VPD) during the study period are shown in Fig. 3. Also displayed are the rain event occurrences, showing the late onset of the 2006 monsoon. The intertropical discontinuity (ITD), materializing the shift in air mass and wind direction (from east to south-west), crossed the study area in May 2006, and around mid-April in 2007. The wet-season rise in humidity and fall in temperature start several weeks before the first rainfalls, with a lag of around two weeks. During the pre- and post-monsoon periods, sharp humidity variations are observed, linked to changes in wind direction (Pagès et al., 1988), including daily cycles. Combining variations in temperature and humidity, VPD displays wet season lows below 1 kPa, against highs above 5 kPa in early May. A relative low occurs shortly after the boreal winter solstice (December 24, 2005; January 1, 2007), due to lower temperatures.

40 35 30 25 20 15

Relative humidity (%)

b

100 80 60 40 20 0

VPD (kPa)

c

6

4

2

0 15/6/05

14/9/05

14/12/05

15/3/06

15/6/06 Date

14/9/06

14/12/06

15/3/07

15/6/07

Fig. 3. Time course of daily air temperature (a), relative humidity (b), and vapour pressure deficit (c), at the fallow site. Lines are 15-day moving averages. Bars at top represent rain event occurrences.

208

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

Results Radiation

a

400

Sw in (W m -2)

Together with the seasonal cycle of potential, clear-sky solar radiation Rso (estimated using Allen et al.’s relationship, 1998), Fig. 4(a–d) shows the variations of daily components of the radiation budget at the fallow and millet stations: short wave incoming (Swin) and outgoing (Swout), long wave incoming (Lwin) and outgoing (Lwout) radiation, respectively. Due to their close proximity, incoming radiation at the two stations are nearly identical and only the fallow site is presented. The potential radiation Rso is highest (above 330 W m2) through a long period spanning from mid-April to end of August, then decreases to a low of 254 W m2 at the boreal winter solstice. The upper envelope for the Swin values follows logically the same

300

overall cycle, with (i) virtually no loss from Rso during much of the dry season (November to March, inclusive) and (ii) a loss factor from April to October that appears well in phase with the monsoon period, as timed by the seasonal relative humidity signal of Fig. 3b. Occasional sharp drops in Swin are the consequences of major rain or cloud events during the wet season, and of dust/aerosols events in the dry season. Much less scattered than Swin, Lwin (Fig. 4c) is also seasonally in phase with Rso, albeit with a few weeks lag indicating the thermal inertia of the Earth system. Except for a peak in April 2006 due to highs in Swin and albedo (see ‘‘Albedo”), Swout displays a relatively flat upper envelope (Fig. 4b), owing to the phase lag between Swin and albedo. Fluctuations of Lwout are well related to those of Lwin, except for the rain season. Fig. 5a shows the net radiation Rn at both sites, together with the combination Swin  Swout + Lwin hereafter referred to as surface forcing radiation Rf (plotted for the fallow site). The interest for Rf

200 100 0

b

120

Sw out (W m -2)

100 80 60 40 20 0

c

500

Lw in (W m -2)

450 400 350 300 250

Lw out (W m -2)

d

600 550 500 450 400 15/6/05

14/9/05

14/12/05

15/3/06

15/6/06 Date

14/9/06

14/12/06

15/3/07

15/6/07

Fig. 4. Time course of the four daily radiation components at the fallow (black dots) and millet (grey dots) sites. Line in (a) is potential solar radiation, Rso. Lines in (b) and (d) are 15-day moving averages. Bars at top represent rain event occurrences.

209

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

700

250

600

Net radiation (W m -2)

200

500 150

400 300

100

200 50

100

0

b

Forcing radiation (W m -2)

a

0

0.5

Albedo (-)

0.4

0.3

0.2

0.1 15/06/05 14/09/05 14/12/05 15/03/06 15/06/06 14/09/06 14/12/06 15/03/07 15/06/07 Date Fig. 5. (a) Daily forcing radiation Rf at fallow site (crosses) and net radiation Rn at fallow (black dots) and millet (grey circles) sites; (b) daily albedo at fallow (black) and millet (grey) sites. Lines are 15-day moving averages. Bars at top represent rain event occurrences.

arises from the fact that, compared to Rn, it is not as directly interacting with the other, complementary components of the surface energy balance (see ‘‘seasonal dynamics of surface energy cycle, and interactions between components”). Both variables display quite simple and smooth, yet distinct, seasonal patterns. Strong mono-modal seasonality results from the summertime concomitance of high incoming with low outgoing radiation, in relation to albedo and surface temperature. Whereas Rn exhibits a rather brief, sharp peak in the second half of the rainy season, the summer highs for Rf spread much longer, in closer relationship with the solar signal. In the dry season, short-timescale fluctuations in component radiations, linked to aerosol events (e.g., early January 2007), are largely attenuated in Rn and Rf, due to compensation between Swin and Lwin. A similar effect occurs in the pre-monsoon period of transition from the dry to the wet season, when a drop occurs in Swin concurrent with a peak in Lwin, presumably due to moister air. Albedo Through much of the dry season, albedo (ratio of Swout to Swin, Fig. 5b) displays values well above 0.3 at both sites, indicating that a large fraction of the surface consists of dry bare soil. During the short rain season, it is driven down by vegetation growth and soil wetting, reaching lower values at the fallow site. A strong anti-cor-

relation with Rn comes more from the seasonal concomitance of high insolation with rain and vegetation, and hence low Lwout, than from the direct effect of Swout. In the early monsoon period before rainfall actually arrives, albedo starts a gradual decrease, likely due to the spectral effect of increasing air humidity (which shifts incident radiation composition from the near infrared to the visible domain where spectral albedo is lower, Samain et al., 2008). First rains produce more scattered albedo values, related to changes in surface color from wetting and drying sequences. In 2005, and to a lesser extent in 2006, a dry spell in July produced a new short period of albedo increase. During the rain season proper, albedo follows a very markedly decreasing general trend, but with significant short time scale variability due to the wetting–drying effect. Day-to-day change in daily albedo can amount to over half the whole seasonal variation, while fluctuations can even be more pronounced at sub-daily scale: a single storm can produce a drop in semi-hourly albedo of over 0.15 in the first part of the wet season. Return to antecedent value takes from 1 day at the season beginning to a few days in the heart of the season. At this time, soil moisture and, most of all, vegetation development smooth out variations due to storm occurrences. Minimum values are reached in September (mid-August for the 2005 millet), together with the peak in leaf cover fraction (Boulain et al., 2009). Periods with no or little rain in the wet season produce new, short-lived increases in albedo at both

210

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

sites, most notably in 2005. In addition to soil drying, this can logically be interpreted as periods of partial wilting of the grass layer, and possibly of the millet or brush layer. The trend of albedo decrease resumes quickly afterwards, when a new rainstorm occurs. From the albedo data, no weeding of the millet crop is apparent during the growing season. Very little cultivation is performed on millet in this area, often being limited to just soil tillage prior to or at the start of the rain season (Peugeot et al., 2003). Because it keeps a larger fraction of bare soil, the millet site almost always displays higher albedo than the fallow, except towards the end of the dry season. Differences build up rapidly during the first part of the rain season, due to a more rapid drop of fallow albedo, to reach as much as 0.07 in the second part of the season (late August–early September). Dry-season highs are restored earlier for millet (by end of November) than for fallow, although the latter’s rise does seem to be much faster than at the more northern grassland site of Agoufou (Samain et al., 2008). Some of the dry-season short drops in Swin (Fig. 4a) can be traced in the albedo signal as very sudden and significant spikes at both sites (such as those peaking on March 8, 2006, or on January 2, April 2, and May 28, in 2007), making daily albedo rise by as much as 0.05. These short peaks are generally associated with major dust events (e.g., March 8, 2006; Slingo et al., 2006), which can impact albedo by changing the spectral composition and the direct versus diffuse proportions of incident radiation (Samain et al., 2008).

Fig. 6 shows the variations in measured fluxes of sensible and latent heat, H and LE, for the fallow (Fig. 6a) and millet (Fig. 6b) sites. These are daytime contributions to daily fluxes (24-h averages), since nighttime boundary layer conditions are rarely suitable for EC measurements. Nighttime fluxes are generally much smaller, especially for latent heat. Also plotted is the reference daily evapotranspiration (ET0, 15-day moving average), as calculated by the FAO Penman–Monteith formula (Allen et al., 1998). In the wettest periods, measured evapotranspiration is close to ET0 for fallow but largely below for millet, and roughly follows ET0 variations. This suggests a control by the climatic demand for a given land cover type, whereas control is mostly by soil water during the dry season and dry spells of the rain season. The fallow site always evaporates significantly more water than the millet site through the growing season. This is especially so in the second part of the 2006 rain season, when, unlike the fallow vegetation, the millet crop was not able to take advantage of the abundant rainfall, because of insufficient plant development (see Boulain et al., 2009). In all cases however, LE does largely dominate H through the heart of the growing season (July–September). Higher evapotranspiration by natural over cultivated vegetation was also reported in this area by Gash et al. (1997) and Wallace et al. (1991). Conversely, rain-season H is higher at the cultivated site. At both sites, but more so at the millet site (which shows more var-

200

Heat flux - fallow (W m-2)

6 150

5 4

100 3 2

50

1

b

0

0

200

7

Heat flux - millet (W m-2)

6 150

5 4

100 3 2

50

1 0 15/6/05

14/9/05

14/12/05

15/3/06

15/6/06

14/9/06

14/12/06

15/3/07

Evaporated water depth - fallow (mm day -1)

7

Evaporated water depth - millet (mm day -1)

a

Eddy fluxes

0 15/6/07

Date Fig. 6. Daily(*) sensible heat flux (black dots) and latent heat flux (grey dots) at fallow (a) and millet (b) sites. Lines are 15-day moving averages. Thin black line on top of each graph is 15-day average reference evapotranspiration ET0. Bars at top represent rain event occurrences. Right axes express values in equivalent evaporation fluxes. ((*): daytime 24-h averages).

211

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

iability in relation to a less developed rooting system), the effect of the rainfall pattern within the wet season is clearly visible, particularly for the alternation of dry and wet spells during the 2005 season. At the end of the wet season, the rapid LE drop is counterbalanced by a concurrent H increase. The latter is very smooth at the millet site while transitions are more abrupt at the fallow site. Because there is more soil water left at the end of the wet season at the millet site (see ‘‘Soil water”), the ranking of the two sites in LE is reversed, compared to the rain season, through the whole recession phase in the first part of the dry season. LE is extremely small during the second part of the dry season, when water is accessible only to very deeply-rooted trees. Ground heat Fig. 7 shows recorded variations in daily soil temperatures (at 0.1, 1.0, and 2.5 m depths) at the fallow site, and heat fluxes (at 0.05 m depth; positive into the soil) for both sites. Temperatures are very similar at the millet site (not shown). Like air temperature, their seasonal pattern is bimodal with a summer low between spring and fall highs. Hence, the temperature profile inverts several times per year over the 2.5 m monitoring depth (downward gradient when temperatures increase, and vice-versa). The shallow

a

depth of heat flux measurements (5 cm) minimizes the deviation from surface flux G for time steps in the order of days or above. From the graph of measured daily heat flux (Fig. 7b) and differences in daily soil temperatures with depth (Fig. 7a), it can be inferred that the ground is mostly releasing heat in November–December– early January, when soil temperatures are decreasing, and storing in March–April–May when soil temperatures are rising. Things are less clear in the monsoon (and pre-monsoon) period, with greater day-to-day variability, due to the succession of wet and dry spells for the topsoil and to the variability of LE. Dry spells tend to increase ground heat (e.g., the rather long dry spell in July 2005), whereas moister soil favours heat release (e.g., mid July and August–early September of 2006). Hence, an abundant monsoon season will decrease heat storage, whereas a deficient rain season might result in increased ground storage. Smaller topsoil temperature gradients than in the rest of the year produce stronger flux intensities, especially on the negative side (soil heat release) when moisture is higher. Finally, a new short period of ground storing occurs at the switch from the wet to the dry season, in October. At the daily timescale, ground heat flux is relatively small compared to the other energy fluxes: in absolute mean, it is less than 7% of the sum H + LE. At a multi-day scale, it becomes even smaller, as can be seen in Fig. 7b for the 15-day time step. Differences between sites are

45

Soil temperature (°C)

40

35

30

25

20

b

20

Soil heat flux (W m-2)

10 0 -10 -20 -30 -40 15/6/05

14/9/05

14/12/05

15/3/06

15/6/06

14/9/06

14/12/06

15/3/07

15/6/07

Date Fig. 7. (a) Daily soil temperature at fallow site, at 0.1 m (thin black), 1 m (grey), and 2.5 m (thick black) depths. (b) Daily soil heat flux (positive downward) at fallow (black) and millet (grey) sites, lines are 15-day moving averages. Bars at top represent rain event occurrences.

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

rather small. The orders of magnitude of ground heat fluxes found here are quite similar to those reported by Verhoef et al. (1999).

a

200

LE + H + G - fallow (W m -2 )

212

150

Energy budget

Rn ¼ LE þ H þ G

ð1Þ

However, several practical reasons preclude perfect closure (Foken et al., 2006), e.g., spatial scale discrepancies between measurements, and spatial heterogeneities (including farmer effects); storage between measurement levels (canopy and superficial soil); so-called eddy flux losses; difficulties with ground flux estimation (Heusinkveld et al., 2004; Guyot et al., 2009). Balance shortfalls commonly stand between 10% and 40% of the available energy (Foken et al., 2006; Mauder et al., 2006). The relative effect of unaccounted storage, as well as of random errors, can potentially be minimized by increasing the time step at which Eq. (1) is considered. However, because atmospheric conditions during the night are often inadequate for EC measurements, the analysis is performed here with daily daytime fluxes. For each site, Fig. 8 compares the two sides of Eq. (1). Closure is quite satisfactory, especially at the fallow site. Mean residuals (LE + H + G  Rn) are 3.6 and 14.2 W m2 at the fallow and millet sites, respectively, while root mean square residuals are 14.7 and 19.2 W m2, respectively. Absolute residuals represent, on average, 9.6% of net radiation for fallow and 12.9% for millet. For the former, 60% of residuals are below 10% (considered as very good closure by Schüttemeyer et al., 2006), and 92% are below 20%, against 45% and 77% for the latter, respectively. The negative bias is likely due to the effect of positive storage, in the shallow soil and canopy, during day time. Residuals do not seem to be particularly related to seasons. The equivalent of Fig. 8 for day-round data at 30-min timestep (not shown) also indicates rather good closure of Eq. (1) at that short timestep, with usually some positive storage during day time and destorage overnight. Given these results and what is commonly obtained with this type of instrumentation, it can reasonably be concluded that precision on the various component fluxes is satisfactory. Soil water Fig. 9 shows the variations in soil moisture measured at three depths at the two sites, namely 0.10, 0.50, and 2.5 m. At the smaller depth, there is very large short-timescale variability throughout the rain seasons, showing that significant drying occurs between most rain events. This time variability largely supersedes differences between sites and years. At the fallow site, much less water infiltrated down to the 0.50 m depth in the 2005 than in the 2006 rain season, partly because of the lower total rainfall amount but predominantly because of the longer dry spells between rain events that allowed for more soil drying. Oppositely, more time-concentrated rainfall in 2006 allowed for a very sharp and high peak in late August (Fig. 9c). Comparing sites at 0.50 m depth, the 2005 season did show higher soil moisture through the season at the millet site. This may largely be attributed to lower evapotranspiration (see ‘‘Eddy fluxes”), rather than to any higher infiltration capacity (very little cultivation work is performed). The 2006 season produced similarly high moisture levels to those at the fallow site but for a much longer period, starting much earlier and lasting into the beginning of the dry season. Virtually no signal, except for a very slight rise in the period of September–December 2006, was recorded at the 2.5 m depth of the fallow site through the two-year period. In contrast, the millet

100

50

0

0

50

100

150

200

Rn - millet (W m -2)

b LE + H + G - millet (W m -2 )

Since all four components Rn, LE, H, and G of the surface energy budget were measured, it is possible to check the closure of the theoretical energy balance equation

200

150

100

50

0

0

50

100

150

200

Rn - millet (W m -2) Fig. 8. Energy budget closure at fallow (a) and millet (b) sites, through the study period. Grey dots correspond to the monsoon period (May 15–October 15). (Values are daytime 24-h averages).

site recorded substantial propagation of the 2006 seasonal moisture signal, with a several-week delay relative to the start and end of the rain season. Again this is due to the specific conditions of the 2006 rain season, with a large cumulative rainfall in a relatively short duration in August, and to the comparatively low evapotranspiration from the millet. Less favourable rainfall in 2005, in amount and distribution, did not allow for any substantial deep infiltration even at the millet site. Discussion Seasonal dynamics of surface energy cycle, and interactions between components From the winter solstice to the pre-monsoon period, LE is very small, and H, G, and Lwout follow a very gradual rise, reflecting a tight balance in the sharing of the increasing Rf. In the pre-monsoon period, before first rains, Rf levels off but H keeps rising, as the arrival of monsoon flux makes air temperature drop and wind

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

213

Fig. 9. Time course of volumetric soil water content at three depths for the fallow (left: (a) 0.10 m, (c) 0.50 m, (e) 2.5 m) and millet (right: (b) 0.10 m, (d) 0.50 m, (f) 2.5 m) plots. Half-hourly data. Bars at top represent rain event occurrences.

speed increase (mid-May 2006 and early April in 2007). Energy conservation triggers the Lwout decrease and change in G pattern. When rain comes in, LE becomes the major driver in the partitioning of the relatively steady Rf, responding very clearly according to precipitated amounts. Initially irregular and in low quantities, water availability causes ephemeral bursts in LE and opposite variations in H. In July, when soil water starts building up, LE sets in durably (but distinctly for the two land covers), producing the sharp falls in H and Lwout. After the rain season, soil water and LE decay rapidly while Rf is still high, yielding significant new rises in H and Lwout. Hence, including the pre-monsoon period of air mass change, dry season fluctuations in H are strongly constrained by the tight direct equilibrium that binds H to Lwout (and G) through the surface temperature, in the partitioning of Rf. In comparison, LE is controlled by factors that are largely external to that thermal state of the soil surface (i.e., soil moisture, vegetation, atmospheric conditions), allowing it to vary much more freely, and reach ET0 at the fallow plot. Energy conservation in meeting Rf then causes these variations to be compensated for by sharp adjustments in H, Lwout, and G. This leads to frequent ground heat releases in the monsoon season despite high radiative inputs. This strong external control on the surface energy cycle, exerted in particular by the water and vegetation cycles, is largely responsible for the shape of the net radiation course with a sharp peak in the second part of the rain season. Essential for latent heat, rainfall turns out about as important as solar forcing for the seasonal variability of the other surface energy outputs. It is largely dominant at interannual or infra-seasonal timescales, particularly during the first half of the rain season. Water cycle dynamics Given the problems of discrepancies in spatial scale and representativeness, combined interpretation of measured water cycle

variables can only be tentative since, unlike the energy balance, closure of the water balance cannot be checked. Fig. 10 shows the time variations, from March 2006 to March 2007, of cumulative rainfall and evapotranspiration (Pc and Ec, respectively) and of the water storage (S) in the first 2.75 m of soil (abbreviated to ‘soil’ or ‘soil layer’ in this subsection), separately for the fallow and millet sites. Storage is counted relative to the residual content at the start of the period of analysis. Also plotted is the balance (Pc  Ec  S) for these three observed variables, which theoretically represents the sum of cumulative runoff, surface retention and deep storage (in variation from initial amount). Evapotranspiration data is missing for some days essentially in the first part of the period before the rain season proper, i.e., when it is particularly low, and essentially for the millet site. These few gaps, adding to nighttime flux omission, tend therefore to slightly underestimate the evapotranspiration course. Some evapotranspiration does occur even before any rainfall at the fallow site (Fig. 10a), which can be interpreted as transpiration from sparse deep-rooted trees. The first rainfalls in June are promptly and largely evaporated. When significant rain starts to occur, as of July 9, the soil wets up and evapotranspiration rises, but at different rates for the two sites through most of the season. At the fallow site, about half the rainfall readily returns to the atmosphere, leaving little water to increase soil storage. Oppositely, soil water builds up faster than cumulative evapotranspiration at the millet site. The 69-mm rainfall event of August 22 produces a marked soil storage increase at both sites, shortly before soil moisture depletion starts to occur (end of August–beginning of September). Since hardly any additional deep storage is apparent from then on at the fallow site (see balance curve, Pc  Ec  S), this soil water loss mostly corresponds to depletion by evapotranspiration, partly fed also by last rainfalls. After rainfall stops (September 24), the balance curve testifies to deep water uptake through the whole dry season, until the beginning of the next season cycle.

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

a

700

Water depth - fallow (mm)

214

600

PC

500

EC

400 300 200

P c -E c -S

100

S

b

700

Water depth - millet (mm)

0 15/3/06

600

14/6/06

13/9/06 Date

13/12/06

15/3/07

PC

500 400

P c -E C -S 300

EC

200 100

S 0 15/3/06

14/6/06

13/9/06 Date

13/12/06

15/3/07

Fig. 10. Time course of cumulative rainfall (Pc, thick black), cumulative evapotranspiration (Ec, thin black), water storage in the first 2.75 m (S, dark grey), and balance of three previous terms (Pc  Ec  S, light grey) at the fallow (a) and millet (b) sites, for the March 15, 2006–March 15, 2007 period.

At the millet site, lighter September rainfall is still in excess of evapotranspiration, and the rising balance curve suggests that the substantial water loss from the monitored soil layer largely goes to deeper storage. This deep storage continues after rainfall stops, contributing still more than evapotranspiration to the soil layer depletion until the end of October. By the end of November, net deep storage reverses to become a major contributor to evapotranspiration, rapidly getting more significant than the soil layer, and also more so than at the fallow site through much of the dry season. Over the year, measured evapotranspiration represents some 65% of the rainfall at the fallow site and 45% at the millet site. These figures are consistent with simulation results obtained with an ecohydrological model by Boulain et al. (accepted for publication). The 2006 rain season being rather wet, with concentrated rainfall, the remainder can largely be attributed to runoff and, at least at the millet site, to some interannual deep storage. Compared to the various components of the energy cycle, including latent heat, differences between sites and years appear to be amplified for the water cycle, especially for deep infiltration. Effect of land cover type For a lot of the investigated variables, the two land cover types exhibit a quite similar behaviour, differences being often in the order of measurement precision. Major differences occur during the growing season and beginning of the dry season, for variables that are most directly linked with vegetation and/or water. Comprising a dense grass layer together with abundantly-leaved shrub stands, fallow vegetation shows clearly lower albedo, producing slightly

higher Rf. Millet is characterized by a low vegetation cover fraction, due to low density and to restriction of grass development by tillage and soil crusting. Wet season LE is noticeably higher at the fallow site, entailing somewhat lower H and Lwout. Note that lower evapotranspiration for millet, attributable to higher surface resistance (Gash et al., 1997) and lower leaf area, is not at the expense of plant productivity, thanks to higher water use efficiency (Boulain et al., 2009). A switch occurs at the end of the growing season, when millet evapotranspiration becomes larger (and conversely for H, Lwout, and shallow soil temperatures), presumably due to more water remaining in the soil. At the annual scale, evapotranspiration appears to be higher at the fallow plot. Lower rain-season evapotranspiration at the millet site translates into much more deep percolation (beneath the 2.75-m-thick monitored zone) and probably (although not measured nor estimated) more runoff than at the fallow site. Runoff increase from crop impingement on natural vegetation has been reported by several studies in the area (e.g., Leblanc et al., 2008; Séguis et al., 2004). However, since soil moisture remains higher at the millet site, this observed lower evapotranspiration is not in consequence of higher runoff. Lower evapotranspiration at the millet site is coherent with the multidecadal water table rise recorded in this area, attributed to the extensive land clearing (Favreau et al., 2009). Results from the fallow site are quite consistent with those obtained by Verhoef et al. (1999) for an older fallow savanna. Not unexpectedly, among the surface types investigated here and by Timouk et al. (2009) at the Northern Sahelian site of AMMA-Gourma (Mali), greatest resemblance in energy variables is found between our fallow site and the Agoufou grassland.

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

Interannual variability The two rain seasons 2005 and 2006 differ by their total amounts of precipitation of 495 and 572 mm, respectively, but more so by their respective durations and time distributions of events and precipitated depths along the season. The more abundant 2006 season was also by far the shortest and the one with the smallest number of events, which were thus more concentrated and intense. It resulted in a wetter soil through most of the season and over the whole monitored soil profile at both sites. This allowed for better vegetation development, as evidenced by the lower albedo values (hence, higher Rf) at both sites throughout the growing season, and especially in August–September. Together with the higher Rf, more water and vegetation produced higher evapotranspiration fluxes, most significantly at the fallow site, and thus lower H, Lwout (hence higher Rn), and soil temperatures. For millet, the rainfall abundance of 2006 was counterbalanced by the less favourable timing (season start and duration), to which it is quite sensitive (Boulain et al., 2006). In natural vegetation, species diversity, particularly in the grass layer, ensures a variety of growth cycles that makes the season timing less of a problem, therefore water abundance will generally be the limiting factor. Note that for millet, additional variability arises from human factors, such as sowing strategies in relation to uncertain monsoon onset, or from variety properties. Abundant rainfall in late August 2006 produced a late growing season peak, as seen in the water and energy cycles. Higher residual moisture at both sites led to some more evapotranspiration during much of the following dry season, resulting in slightly lower Lwout and soil temperature at the winter solstice. Longer sampling of the strong interannual variability of precipitation (Fig. 2) should help to better characterize its impact on the energy and water cycles, for the two land covers.

Conclusions In this and the companion paper (Boulain et al., 2009), the extended, multi-disciplinary data set being acquired on the Sahelian AMMA site of West Niger, has shown its capability to provide new insights into the complex, interacting physical and biological processes that control the coupled cycles of energy, water, vegetation, and carbon. At this stage, the hydrological cycle is still left partly open since runoff and deep drainage are not included in the analysis explicitly. The importance of lateral redistribution of storm water in this endoreic landscape calls for a catchment-scale approach, which is being pursued jointly to the local scale analysis presented here (Boulain et al., accepted for publication; Ezzahar et al., 2009). This analysis points to the central role of the latent heat component, and therefore of water when available, in the control of the energy balance. Sensible heat and soil heat, and, to an even greater extent, outgoing long wave radiation are largely under its direct subordination, the first two being partially buffered by air and soil over different, limited timescales. Solar forcing remains of course the first factor of seasonal variability, but rain water effects are almost as large, and dominant at all other timescales. The indirect effect of water through albedo is smaller than its ‘‘direct” effect, that is, through latent heat. Latent heat modulation by vegetation is important, as it varies markedly between land cover types, lower for the millet than for the ‘‘natural” vegetation in the growing season. As effects on latent heat are partially offset by those on albedo, overall differences are slight for a number of energy-related variables. However, because evapotranspiration dominates the water budget so largely in this area, differences are amplified for water-related variables, such as humidity in deeper soil horizons, which come in play as residuals in this budget. Although cultivated

215

surfaces have been shown to produce more runoff than fallow or natural land in this area (Casenave and Valentin, 1992; Peugeot et al., 1997), our results indicate that runoff is not the prime cause for the lower return of water to the atmosphere by millet fields, since higher soil moisture levels were observed concomitantly to this lower evapotranspiration. Conversely, because of soil crusting and storm spacing, Hortonian runoff is only lightly modulated by variations in evapotranspiration (Peugeot et al., 2003). Good closure of the energy budget brings support to the reliability of the various component measurements, as does the agreement found with spatially-integrated scintillometry measurements of sensible heat at the catchment scale (Ezzahar et al., 2009). With continued acquisition over several additional wet–dry seasons cycles, this comprehensive and high-resolution data set represents a unique source of information for investigation of land processes and for calibration/validation of land surface models at different scales in this region of Africa. Acknowledgments This work was financially supported through the AMMA programme,1 (including the AMMA-Catch O.R.E.), the French ECCOPNRH programme (project ‘‘Eau et Végétation au Niger”), and by IRD. Stimulating exchanges with J.-L. Rajot, F. Lohou, M. Lothon, L. Kergoat, J.-M. Cohard, E. Ceschia, and F. Said are warmly acknowledged. I. Zin and the anonymous reviewers are thanked for their helpful comments. References Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration: Guidelines for Computing Crop Requirements. Irrigation and Drainage. FAO, Rome. 300 pp. Bagayoko, F., Yonkeu, S., Elbers, J., van de Giesen, N., 2007. Energy partitioning over the West African savanna: multi-year evaporation and surface conductance measurements in Eastern Burkina Faso. Journal of Hydrology 334 (3–4), 545– 559. Boone, A., de Rosnay, P., 2007. AMMA forcing data for a better understanding of the West African monsoon surface–atmosphere interactions. In: Boegh, E. et al. (Eds.), Quantification and Reduction of Predictive Uncertainty for Sustainable Water Resource Management, vol. 313. IAHS Publ., pp. 231–241. Boulain, N., Cappelaere, B., Seguis, L., Gignoux, J., Peugeot, C., 2006. Hydrologic and land use impacts on vegetation growth and NPP at the watershed scale in a semi-arid environment. Regional Environmental Change 6, 147–156. Boulain, N., Cappelaere, B., Séguis, L., Favreau, G., Gignoux, J., accepted for publication. Water balance and vegetation change in the Sahel: a case study at the watershed scale with an eco-hydrological model. Journal of Arid Environments. Boulain, N., Cappelaere, B., Ramier, D., Issoufou, H.B.A., Halilou, O., Seghieri, J., Guillemin, F., Gignoux, J., Timouk, F., 2009. Towards an understanding of coupled physical and biological processes in the cultivated Sahel - 2. vegetation and carbon dynamics. Journal of Hydrology (AMMA-Catch Special Issue) 375 (1–2), 190–203. Braud, I., 1998. Spatial variability of surface properties and estimation of surface fluxes of a savannah. Agricultural and Forest Meteorology 89 (1), 15–44. Cappelaere, B., Descroix, L., Lebel, T. et al., 2009. The AMMA-Catch experiment in the cultivated Sahelian area of South-West Niger – Strategy, implementation, site description, main results. Journal of Hydrology (AMMA-Catch Special Issue) 375 (1–2), 34–51. Casenave, A., Valentin, C., 1992. A runoff capability classification-system based on surface-features criteria in semiarid areas of West Africa. Journal of Hydrology 130 (1–4), 231–249. Charney, J.G., 1975. The dynamics of deserts and droughts. Quarterly Journal of the Royal Meteorological Society 101, 193–202. Dolman, A.J., Gash, J.H.C., Goutorbe, J.P., Kerr, Y., Lebel, T., Prince, S.D., Stricker, J.N.M., 1997. The role of the land surface in Sahelian climate: HAPEX-Sahel results and future research needs. Journal of Hydrology 189 (1–4), 1067–1079. Ezzahar, J., Chehbouni, A., Hoedjes, J.C., Ramier, D., Boulain, N., Boubkraoui, S., Cappelaere, B., Descroix, L., Mougenot, B., Timouk, F., 2009. Combining 1 Based on a French initiative, AMMA was built by an international scientific group and is currently funded by a large number of agencies, especially from France, the UK, the US and Africa. It has been the beneficiary of a major financial contribution from the European Community’s Sixth Framework Research Programme. Detailed information on scientific coordination and funding is available on the AMMA International website http://www.ammainternational.org.

216

D. Ramier et al. / Journal of Hydrology 375 (2009) 204–216

scintillometer and an aggregation scheme to estimate area-averaged latent heat flux during AMMA experiment. Journal of Hydrology (AMMA-Catch Special Issue) 375 (1–2), 217–226. Favreau, G., Cappelaere, B., Massuel, S., Leblanc, M.J., Boucher, M., Boulain, N., Leduc, C., 2009. Land clearing, climate variability, and water resources increase in semiarid south-west Niger: a review. Water Resources Research, in press. doi:10.1029/2007WR006785. Foken, T., Wimmer, F., Mauder, M., Thomas, C., Liebethal, C., 2006. Some aspects of the energy balance closure problem. Atmospheric Chemistry and Physics 6, 4395–4402. Gash, J.H.C., Kabat, P., Monteny, B.A., Amadou, M., Bessemoulin, P., Billing, H., Blyth, E.M., deBruin, H.A.R., Elbers, J.A., Friborg, T., Harrison, G., Holwill, C.J., Lloyd, C.R., Lhomme, J.P., Moncrieff, J.B., Puech, D., Soegaard, H., Taupin, J.D., Tuzet, A., Verhoef, A., 1997. The variability of evaporation during the HAPEX-Sahel intensive observation period. Journal of Hydrology 189 (1–4), 385–399. Goutorbe, J.P., Lebel, T., Dolman, A.J., Gash, J.H.C., Kabat, P., Kerr, Y.H., Monteny, B., Prince, S.D., Stricker, J.N.M., Tinga, A., Wallace, J.S., 1997. An overview of HAPEXSahel: a study in climate and desertification. Journal of Hydrology 189 (1–4), 4– 17. Guyot, A., Cohard, J.M., Anquetin, S., Galle, S., Lloyd, C.R., 2009. Combined analysis of energy and water balances to estimate latent heat flux of a sudanian small catchment. Journal of Hydrology (AMMA-Catch Special Issue) 375 (1–2), 227– 240. Heusinkveld, B.G., Jacobs, A.F.G., Holtslag, A.A.M., Berkowicz, S.M., 2004. Surface energy balance closure in an arid region: role of soil heat flux. Agricultural and Forest Meteorology 122 (1–2), 21–37. Kabat, P., Dolman, A.J., Elbers, J.A., 1997. Evaporation, sensible heat and canopy conductance of fallow savannah and patterned woodland in the Sahel. Journal of Hydrology 189 (1–4), 494–515. Koster, R.D., Dirmeyer, P.A., Guo, Z., Bonan, G., Chan, E., Cox, P., Gordon, P.T., Kanae, S., Kowalczyk, E., Lawrence, D., Liu, P., Lu, C.-H., Malyshev, S., McAvaney, B., Mitchell, K., Mocko, D., Oki, T., Oleson, K., Pitman, A., Sud, Y.C., Taylor, C.M., Verseghy, D., Vasic, R., Xue, Y., Yamada, T., 2004. Regions of strong coupling between soil moisture and precipitation. Science 305, 1138–1140. Lebel, T., Diedhiou, A., Laurent, H., 2003. Seasonal cycle and interannual variability of the Sahelian rainfall at hydrological scales. Journal of Geophysical Research 108 (D8), 8389. doi:10.1029/2001JD001580. Lebel, T., Cappelaere, B., Hanan, N., Levis, S., Descroix, L., Galle, S., Kergoat, L., Mougin, E., Peugeot, C., Séguis, L., Vieux, B., 2009. AMMA-CATCH studies in the Sahelian region of West-Africa: an overview. Journal of Hydrology (AMMACatch Special Issue) 375 (1–2), 3–13. Leblanc, M.J., Favreau, G., Massuel, S., Tweed, S.O., Loireau, M., Cappelaere, B., 2008. Land clearance and hydrological change in the Sahel: SW Niger. Global and Planetary Change 61 (3–4), 135–150. Lloyd, C.R., 1995. The effect of heterogeneous terrain on micrometeorological flux measurements – a case study from HAPEX-Sahel. Agricultural and Forest Meteorology 73 (3–4), 209–216. Lloyd, C.R., Bessemoulin, P., Cropley, F.D., Culf, A.D., Dolman, A.J., Elbers, J., Heusinkveld, B., Moncrieff, J.B., Monteny, B., Verhoef, A., 1997. A comparison of surface fluxes at the HAPEX-Sahel fallow bush sites. Journal of Hydrology 188–189 (1–4), 400–425. Mauder, M., Foken, T., 2004. Documentation and instruction manual of the eddy covariance software package TK2. Report Nr. 26, U. Bayreuth – Abt. Mikrometeorologie. (). Mauder, M., Liebethal, C., Göckede, M., Leps, J.-P., Beyrich, F., Foken, T., 2006. Processing and quality control of flux data during LITFASS-2003. BoundaryLayer Meteorology 121, 67–88.

Meir, P., Levy, P.E., Grace, J., Jarvis, P.G., 2007. Photosynthetic parameters from two contrasting woody vegetation types in West Africa. Plant Ecology 192 (2), 277– 287. Pagès, J.-P., Frangi, J.-P., Durand, P., Estournel, C., Druilhet, A., 1988. Etude de la couche limite sahélienne – Expérience Yantala. Boundary-Layer Meteorology 43, 183–203. Pellarin, T., Laurent, J.-P., Decharme, B., Descroix, L., Cappelaere, B., Ramier, D., 2009. Hydrological modelling and associated microwave emission of a semi-arid region in South-Western Niger. Journal of Hydrology (AMMA-Catch Special Issue) 375 (1–2), 262–272. Peugeot, C., Esteves, M., Galle, S., Rajot, J.L., Vandervaere, J.P., 1997. Runoff generation processes: results and analysis of field data collected at the East Central Supersite of the HAPEX-Sahel experiment. Journal of Hydrology 188– 189 (1–4), 179–202. Peugeot, C., Cappelaere, B., Vieux, B.E., Seguis, L., Maia, A., 2003. Hydrologic process simulation of a semiarid, endoreic catchment in Sahelian West Niger. 1. Modelaided data analysis and screening. Journal of Hydrology 279 (1–4), 224–243. Redelsperger, J.L., Thorncroft, C., Diedhiou, A., Lebel, T., Parker, D., Polcher, J., 2006. African monsoon multidisciplinary analysis (AMMA): an international research project and field campaign. Bulletin of the American Meteorological Society 87 (12), 1739–1746. Samain, O., Kergoat, L., Hiernaux, P., Guichard, F., Mougin, E., Timouk, F., Lavenu, F., 2008. Analysis of the in situ and MODIS albedo variability at multiple time scales in the Sahel. Journal of Geophysical Research 113, D14119. Saux-Picart, S., Ottlé, C., Perrier, A., Decharme, B., Coudert, B., Zribi, M., Boulain, N., Cappelaere, B., Ramier, D., accepted for publication. SEtHyS_Savannah: a multiple source land surface model applied to Sahelian landscapes. Agricultural and Forest Meteorology. Saux-Picart, S., Ottlé, C., Decharme, B., André, C., Zribi, M., Perrier, A., Coudert, B., Boulain, N., Cappelaere, B., Descroix, L., Ramier, D., 2009. Water and energy budgets simulation over the Niger super site spatially constrained with remote sensing data. Journal of Hydrology (AMMA-Catch Special Issue) 375 (1–2), 287– 295. Schüttemeyer, D., Moene, A.F., Holtslag, A.A.M., de Bruin, H.A.R., van de Giesen, N., 2006. Surface fluxes and characteristics of drying semi-arid terrain in West Africa. Boundary-Layer Meteorology 118, 583–612. Séguis, L., Cappelaere, B., Milesi, G., Peugeot, C., Massuel, S., Favreau, G., 2004. Simulated impacts of climate change and land-clearing on runoff from a small Sahelian catchment. Hydrological Processes 18, 3401–3413. Slingo, A., Ackerman, T.P., Allan, R.P., Kassianov, E.I., McFarlane, S.A., Robinson, G.J., Barnard, J.C., Miller, M.A., Harries, J.E., Russell, J.E., Dewitte, S., 2006. Observations of the impact of a major Saharan dust storm on the atmospheric radiation balance. Geophysical Research Letters 33, L27817. Timouk, F., Kergoat, L., Mougin, E., Lloyd, C.R., Ceschia, E., Cohard, J.M., de Rosnay, P., Hiernaux, P., Demarez, V., Taylor, C.M., 2009. Response of sensible heat flux to water regime and vegetation development in a central Sahelian landscape. Journal of Hydrology (AMMA-Catch Special Issue) 375 (1–2), 178–189. Verhoef, A., Allen, S.J., De Bruin, H.A.R., Jacobs, C.M.J., Heusinkveld, B.G., 1996. Fluxes of carbon dioxide and water vapour from a Sahelian savanna. Agricultural and Forest Meteorology 80 (2–4), 248. Verhoef, A., Allen, S.J., Lloyd, C.R., 1999. Seasonal variation of surface energy balance over two Sahelian surfaces. International Journal of Climatology 19, 1267–1277. Wallace, J.S., Wright, I.R., Stewart, J.B., Holwill, C.J., 1991. The Sahelian energy balance experiment (SEBEX): ground based measurements and their potential for spatial extrapolation using satellite data. Advances in Space Research 11 (3), 131–141. Xue, Y., Shukla, J., 1993. The influence of land surface properties on Sahel climate. Part I: desertification. Journal of Climate 6, 2232–2245.