The hydrological effects of changes in forest area and species composition in the federal state of Brandenburg, Germany

Das übergreifende Ziel der vorliegenden Arbeit ist es, die Interaktion zwischen Landnutzungsänderung und dem Landschaftswasserhaushalt zu quantifizieren. Das Untersuchungsgebiet für die Analyse ist das Land Brandenburg. Bedingt durch seine Kombination geringer Sommerniederschläge mit der Dominanz sandiger Böden und hoher Verdunstungsraten, insbesondere von den großflächigen Wäldern und Forsten, ist es besonders empfindlich gegenüber Landnutzungsänderung. Waldflächen sind Schlüsselelemente im Landschaftswasserhaushalt, da sie den Bodenwasserspeicher effizienter mit der Atmosphäre koppeln als die meisten anderen Vegetationsformen. Im ersten Teil der Arbeit war es daher notwendig, ein geeignetes Modellkonzept zu finden. Der Ansatz sollte in der Lage sein, die hydrologischen Effekte auf Landschaftsebene zu modellieren, ohne dabei die Datenverfügbarkeit in diesem Anwendungsbereich zu überschreiten.Das entwickelte Modellkonzept wurde in das ökohydrologische Einzugsgebietsmodell SWIM (Soil Water Integrated Model) integriert. Nach einer Test- und Entwicklungsphase konnte das Modell für die integrierte Analyse der Wirkung von zwei Szenarien auf den Landeswasserhaushalt verwendet werden. Das erste Szenario beschäftigt sich mit der möglichen Zunahme der Waldfläche als Folge der Neuausrichtung der Agrarsubventionspolitik der Europäischen Union. Die Waldflächenzunahme führt zu einer Steigerung der Evapotranspiration im langjährigen Mittel. Das zweite Szenario behandelt die Auswirkung des Brandenburger Waldumbauprogramms und hat eine vergleichsweise geringe Abnahme der langjährigen mittleren Verdunstung zur Folge. Der lineare mittlere Verlauf überdeckt ein ausgeprägtes räumliches und saisonales Muster der Veränderung. Die Zonen starker Effekte der beider Szenarien überlappen sich nur in einigen Fällen, so ist es möglich, dass die positiven Wirkungen des Waldumbauprogramms in einigen Regionen durch eine mögliche Ausweitung der Waldfläche aufgehoben werden.Die vorgestellten Ergebnisse zeigen deutlich, dass Landnutzungsänderungen, die durch politische oder administrative Entscheidungen ausgelöst werden, Auswirkungen auf elementare Landschaftsfunktionen wie den Wasserhaushalt haben. Es wird deutlich, dass ein integrativer Modellierungsansatz, der die wahrscheinlichen Wirkungen administrativer Entscheidungen in Betracht zieht, Grundlagen für eine nachhaltige Entwicklung liefern kann. Diese Ergebnisse werden umso relevanter, je stärker die betroffene Ressource bereits eingeschränkt ist. In Bezug auf die Wasserressourcen im Land Brandenburg ist das der Fall und aktuelle Studien zum Globalen Wandel in der Region prognostizieren eine Verschärfung dieser Situation.

This thesis aims to quantify the human impact on the natural resource water at the landscape scale. The drivers in the federal state of Brandenburg (Germany), the area under investigation, are land-use changes induced by policy decisions at European and federal state level. The water resources of the federal state are particularly sensitive to changes in land-use due to low precipitation rates in the summer combined with sandy soils and high evapotranspiration rates. Key elements in landscape hydrology are forests because of their unique capacity to transport water from the soil to the atmosphere. Given these circumstances, decisions made at any level of administration that may have effects on the forest sector in the state are critical in relation to the water cycle. It is therefore essential to evaluate any decision that may change forest area and structure in such a sensitive region. Thus, as a first step, it was necessary to develop and implement a model able to simulate possible interactions and feedbacks between forested surfaces and the hydrological cycle at the landscape scale. The result is a model for simulating the hydrological properties of forest stands based on a robust computation of the temporal and spatial LAI (leaf area index) dynamics. The approach allows the simulation of all relevant hydrological processes with a low parameter demand. It includes the interception of precipitation and transpiration of forest stands with and without groundwater in the rooting zone. The model also considers phenology, biomass allocation, as well as mortality and simple management practices. It has been implemented as a module in the eco-hydrological model SWIM (Soil and Water Integrated Model). This model has been tested in two pre-studies to verify the applicability of its hydrological process description for the hydrological conditions typical for the state. The newly implemented forest module has been tested for Scots Pine (Pinus sylvestris) and in parts for Common Oak (Quercus robur and Q. petraea) in Brandenburg. For Scots Pine the results demonstrate a good simulation of annual biomass increase and LAI in addition to the satisfactory simulation of litter production. A comparison of the simulated and measured data of the May sprout for Scots pine and leaf unfolding for Oak, as well as the evaluation against daily transpiration measurements for Scots Pine, does support the applicability of the approach. The interception of precipitation has also been simulated and compared with weekly observed data for a Scots Pine stand which displays satisfactory results in both the vegetation periods and annual sums.After the development and testing phase, the model is used to analyse the effects of two scenarios. The first scenario is an increase in forest area on abandoned agricultural land that is triggered by a decrease in European agricultural production support. The second one is a shift in species composition from predominant Scots Pine to Common Oak that is based on decisions of the regional forestry authority to support a more natural species composition. The scenario effects are modelled for the federal state of Brandenburg on a 50m grid utilising spatially explicit land-use patterns.The results, for the first scenario, suggest a negative impact of an increase in forest area (9.4% total state area) on the regional water balance, causing an increase in mean long-term annual evapotranspiration of 3.7% at 100% afforestation when compared to no afforestation. The relatively small annual change conceals a much more pronounced seasonal effect of a mean long-term evapotranspiration increase by 25.1% in the spring causing a pronounced reduction in groundwater recharge and runoff. The reduction causes a lag effect that aggravates the scarcity of water resources in the summer. In contrast, in the second scenario, a change in species composition in existing forests (29.2% total state area) from predominantly Scots Pine to Common Oak decreases the long-term annual mean evapotranspiration by 3.4%, accompanied by a much weaker, but apparent, seasonal pattern. Both scenarios exhibit a high spatial heterogeneity because of the distinct natural conditions in the different regions of the state. Areas with groundwater levels near the surface are particularly sensitive to changes in forest area and regions with relatively high proportion of forest respond strongly to the change in species composition. In both cases this regional response is masked by a smaller linear mean effect for the total state area. Two critical sources of uncertainty in the model results have been investigated. The first one originates from the model calibration parameters estimated in the pre-study for lowland regions, such as the federal state. The combined effect of the parameters, when changed within their physical meaningful limits, unveils an overestimation of the mean water balance by 1.6%. However, the distribution has a wide spread with 14.7% for the 90th percentile and -9.9% for the 10th percentile. The second source of uncertainty emerges from the parameterisation of the forest module. The analysis exhibits a standard deviation of 0.6 % over a ten year period in the mean of the simulated evapotranspiration as a result of variance in the key forest parameters. The analysis suggests that the combined uncertainty in the model results is dominated by the uncertainties of calibration parameters. Therefore, the effect of the first scenario might be underestimated because the calculated increase in evapotranspiration is too small. This may lead to an overestimation of the water balance towards runoff and groundwater recharge. The opposite can be assumed for the second scenario in which the decrease in evapotranspiration might be overestimated.