Extrapolating water and carbon fluxes of managed grasslands in time and space based on surface and airborne observations
Zusammenfassung der Projektergebnisse
Summary of results (1st and 2nd project phase; 3-6 pages) Micrometeorology According to our hypotheses and to meet the targets formulated in the Goals section a variety of tools and methods were incorporated and applied in the micrometeorology work package of P6: CO2 and H2O fluxes were measured from 2004 through 2009 by Eddy Covariance (EC) at different sites (Leymus chinensis LC; Stipa grandis SG). Over the whole experiment time, year-around EC measurements were implemented over LC at an ungrazed site fenced since 1979 (UG79). In the first phase (2004 to 2006) a second EC tower was roving between three other sites (all LC) with different grazing intensities (winter grazed WG, continuously grazed CG, heavily grazed HG) during vegetation time. In the second phase parallel measurements were done at the two steppe types (LC, SG) from 2007 to 2009 at ungrazed sites. Additionally, a third EC tower continued measurements from the first phase at HG in 2008 and 2009. Instruments were installed for year-around measurements aiming to cover at least the complete growing season. To test the applicability of the method to determine grazing differences, they were investigated vs. method sensitivity (Ketzer et al. 2008). Surface characteristics, energy fluxes, and CO2 and H2O fluxes were shown to be different for most of the grazing intensities. However, the applied management of the moderately grazed sites was hard to determine, and the data on the radiation and turbulent fluxes did not show significant differences (Ketzer et al. 2008). Considering the observed high interannual variability in precipitation, the effect of a single very dry year can last longer than was expected without the microclimatic feedback. Consequently, sustainable management of the semi-arid grassland plays an important role not only in environmental protection against irreversible vegetation changes, but also for the resulting microclimate intensifying risks in extremely dry years. Any interference in this very sensitive semi-arid grassland ecosystem could result in serious changes in energy and matter fluxes. Carbon The C flux was initially calculated and corrected following the standard methodology of CARBOEUROPE (Aubinet et al. 2000). During phase 2 of MAGIM, methods for open path instruments (as used for the study) were published to correct for the newly-discovered selfheating effect (Burba et al. 2008). All C results had to be recalculated and the correction P6 - Extrapolating water and carbon fluxes of managed grasslands in time and space based on surface and airborne observations. 128 methods had to be tested for the sites in Inner Mongolia with their extreme winter temperatures (Vetter et al. in revision). After including the self-heating correction as an option in the post-processing routines, the carbon fluxes at the ecosystem level were calculated (Fig. 4.7.1; NEE - net ecosystem exchange; TER - total ecosystem respiration, GPP - gross primary production). Fig. 4.7.1: Mean diurnal courses of incoming short wave radiation Rg, temperature T and carbon fluxes: NEE, TER and GPP of 2004 through 2006 The calculated C fluxes represent only a first estimate of the effect of differently grazed sites on C fluxes and their reaction to the varying meteorological conditions; an investigation of the inter-annual and annual variation from 2004 to 2009 should be finished by 2010. The results of the first study indicate a grazing effect at higher grazing intensities, which are visible in lower C fluxes at grazed sites (CG/WG, HG) than at the ungrazed site (UG79). The reduced vegetation cover (including fewer roots under dry conditions in 2005; Gao et al. 2008) and P6 - Extrapolating water and carbon fluxes of managed grasslands in time and space based on surface and airborne observations. 129 more soil compression (Krümmelbein et al. 2008) at the heavily grazed site limit regeneration even under favourable precipitation conditions. C sequestration is determined mainly by available water: the influence of soil moisture is the main driving factor in this ecosystem. Without sufficient soil moisture, other potentially limiting factors like radiation or temperature are not important. High C uptake occurs only under wet conditions and well-developed vegetation. In addition to water availability, the timing of rainfall events and the general temperature level are also important. August and September of 2004 were quite wet and had below average temperatures. In contrast, in 2005 (complete growing season) and 2006 (July and August), several months were dry and relatively warm. By utilizing these facts, we expect UG79 to be a considerable C sink in wet years, almost C neutral in moderately dry years, and a source of carbon in very dry years. The heavily grazed areas were always a source for CO2. At moderately grazed sites, C emission was lower. Overall, grazing reduces C fluxes (both uptake and emission), and steppe systems are more likely to be a CO2 source under high water stress as observed in. Water The 2006 growing season (May to September) had relatively normal precipitation (227 mm) compared with the 25-year average value (1982-2006; 286 ± 66 mm; mean ± S.D.). It was extremely dry during the 2005 growing season with only 147 mm precipitation, which was significantly below average. The water measurements of the first phase showed the three kinds of grazing intensities (CG, WG and HG) decreased evapotranspiration (ET) via removal of the vegetation under different soil water conditions. This means the effects of grazing on the ET responded slightly to the interannual variation in precipitation. At the ungrazed site (UG79), the annual ET of the hydrological year 2005-2006 was larger than the annual precipitation, and this deficit was balanced by the difference of ET and precipitation during the next hydrological year 2006-2007. However, the energy balance ratio decreased significantly from 2007 to 2009, which indicated that ET may be underestimated. Consequently, the water imbalance of this period between precipitation and ET was almost 200 mm (Fig. 4.7.2). But if we force energy balance ratio to an average level, the water balance is closed over the entire observation period over LC at UG79. P6 - Extrapolating water and carbon fluxes of managed grasslands in time and space based on surface and airborne observations. 130 Fig. 4.7.2: Comparison of cumulative precipitation and cumulative evapotranspiration for the whole measurement period and each hydrological year at the ungrazed site The determination factors of ET at the ungrazed site were investigated on different time scales. Daily ET was mostly controlled by soil water content (SWC) under extremely dry conditions. When the soil became wetter, the response of ET to SWC became weaker and ET was more controlled by net radiation and vapour pressure deficit. For longer periods, the available energy AE strongly controlled monthly ET and the relation between solar radiation and monthly ET was a little weaker (Fig. 4.7.3). On an annual basis, precipitation was the main factor determining the interannual variations in ET. Moreover, when the 5-day average ET around the LAI measurement days at differently grazed sites was combined together, there were clear relations between ET and LAI if the data was complied into three classes of SWC (Wang et al. 2010). Fig. 4.7.3: The relations between ET and available energy, solar radiation on a monthly scale at the ungrazed site P6 - Extrapolating water and carbon fluxes of managed grasslands in time and space based on surface and airborne observations. 131 Remote Sensing According to our hypotheses and to meet the targets formulated in the Goals section a variety of tools and methods were incorporated and applied in the remote sensing work package of P6: A high resolution land cover classification was developed by means of a supervised classification of Landsat 7 ETM+ multispectral data captured in June 2000. The dominant land cover types identified are heavy grazing meadow (32 %), wetland meadow (16 %), moderate grazing meadow (16 %), and sand land with sparse vegetation (12 %), saline meadows (11 %) and hay making meadow (10 %) (Fan et al. 2007). The same Landsat scene was also used to retrieve land surface variables (e.g. albedo, NDVI and surface temperature) and to compute the regional distribution of land surface energy fluxes. The heavily grazed areas show higher albedo (0.15) than moderately grazed (0.13) and hay making areas (0.12). NDVI decreases with rising grazing intensity (hay making 0.64, moderate grazing 0.49 and heavy grazing meadow 0.42). Surface temperatures rise as grazing intensity strengthens. Grazing intensity also influences energy surface fluxes: sensible heat flux increases from around 148 Wm-² at hay making meadows to approx. 184 Wm-² at heavy grazing meadows. However, latent heat fluxes decrease with increasing grazing intensity from 406 Wm-² (hay making meadows) to 295 Wm-² (heavy grazing meadows). Latent heat fluxes account for 382 Wm-² at moderate grazing meadows (Fan et al. 2007). High spectral surface measurements by spectrometer covering the spectrum of 380-1100 nm with a 0.2 mm resolution and the estimation of leaf area index (LAI) by collecting leaf samples allowed for the establishment of individual relationships between NDVI and LAI at differently grazed sites (UG79, WG, and HG). Thus, by applying NDVI measurements and the relationship found, the vegetation status and grass yield can be rapidly and nondestructively estimated in the area (Fan et al. 2009). Spectrometer measurements also revealed variations of albedo during growing season are biggest at the ungrazed site (UG79). At the heavily grazed (HG) site changes are small but the albedo is always higher than at UG79 (Fan et al 2010). Finally, the spatiotemporal variability of LAI and surface temperature in the Xilin river catchment was determined using MODIS data for a period between 2000 and 2008. The study showed precipitation has a strong spatial and interannual variability and is the main driving force for vegetation development in the catchment. Vegetation growth is impeded in dry years like 2001 and 2005; the mean LAI accounts for 0.5 during the growing season. In years with high and above average precipitation, e.g. 2003, vegetation can recover which is P6 - Extrapolating water and carbon fluxes of managed grasslands in time and space based on surface and airborne observations. 132 shown by an increase of mean LAI to 0.7 (Fig 4.7.4). The high grazing intensity in the vicinity of the Xilin river is also obvious in wet years - LAI values stay low (0.4). A mean coefficient of variation of 48 % reflects the high temporal variability of LAI. The study also showed large precipitation events are the main trigger for a substantial gain in leaf area and that the field sites used for experiments on the plot scale sufficiently represent the average conditions in the Xilin catchment (Schaffrath et al. 2010). Fig. 4.7.4: The relationship between annual mean precipitation and MODIS LAI in the Xilin river catchment from 2000 to 2008 The modelling of soil moisture with a high spatial resolution (approx. 1 to 2 m) from thermal infrared (TIR) data is still in process as the Ultralight flight campaign had to be re-scheduled from 2008 to 2009 as the Chinese government denied flight permission due to the Olympic Games in Beijing in 2008. However, the campaign was carried out in 2009 and spectrometer measurements, grass sampling, soil moisture and meteorological measurements supported the TIR data collection (Fig 4.7.5). P6 - Extrapolating water and carbon fluxes of managed grasslands in time and space based on surface and airborne observations. 133 Fig. 4.7.5: Summary of precipitation and soil moisture conditions at UG79 site and sampling during the flight campaign 2009 As a general and common contribution of P6 to MAGIM, remote sensing and micrometeorological data were provided to the MAGIM intranet and GIS data base.
Projektbezogene Publikationen (Auswahl)
-
(2007): Regional land surface energy fluxes by satellite remote sensing in the Upper Xilin River Watershed (Inner Mongolia, China). Theoretical and Applied Climatology, 88, 231-245
Fan, L., Liu, S.H., Bernhofer, C., Liu, H., Berger, F.H.
-
(2008): Sensitivity of micrometeorological measurements to detect surface characteristics of grasslands in Inner Mongolia. International Journal of Biometeorology, 52, 563-574
Ketzer, B., Bernhofer, C., Liu, H.
-
(2009): Investigating the relationship between NDVI and LAI in semiarid grassland in Inner Mongolia using in-situ measurements. Theoretical and Applied Climatology, 95, 151–156
Fan, L., Gao, Y., Brück, H., Bernhofer, C.
-
(2010): Grazing effects on seasonal dynamics and interannual variabilities of spectral reflectance in semi-arid grassland in Inner Mongolia. Plant and Soil, 340 (1-2), 169-180
Fan, L., Ketzer, B., Liu, H., Bernhofer, C.H.
-
(2010): Spatiotemporal variability of grassland vegetation cover in a catchment in Inner Mongolia, China, derived from MODIS data products. Plant and Soil, 340 (1-2), 181-198
Schaffrath, D., Bartholdt, F.K., Bernhofer, C.
