Biomass is an important ecological variable for understanding the responses of vegetation to the climate system and currently observed global change. In semi-arid areas it especially influences environmental processes, such as the hydrological cycle, soil erosion and degradation. It is therefore essential to develop accurate and transferable methods for biomass estimation in these areas. The quantification of biomass for large areas and long time periods is necessary to identify and monitor those areas under high risk of degradation and desertification. This can only be achieved using data collected from satellite remote sensing.rnThe impact of changes in vegetation biomass on the global ecosystem is of high relevancernand may have a critical influence on the future evolution of the climate. Modelling net primary productivity (NPP) is an important instrument for analysing carbon exchange between atmosphere and vegetation. It allows for quantification of carbon sinks and sources. The analysis of NPP time-series and their spatio-temporal patterns helps us to understand ecological functioning and potential disturbances.rnThe vegetation in the arid and semi-arid environments of Kazakhstan faces extreme climaticrnconditions. Land degradation and desertification already pose large ecological challenges. Thernregion is expected to be affected particularly strongly by future climate change. In Central Asia further rises in temperature and an intensification of aridity are predicted. In the context of diverse anthropogenic and climatic influences on the Kazakh environment, it is of great interest to observe large-scale vegetation dynamics and biomass distribution.rnIn this dissertation, previous research activities and remote-sensing-based methods forrnbiomass estimation in semi-arid regions have been comprehensively reviewed for the first time. The review revealed that the biggest challenge is the transferability of methods in time and space. Empirical approaches, which are predominantly applied, proved to be hardly transferable. This lack of transferability hinders a repeated or operational application of biomass estimation methods. Remote-sensing-based NPP models, on the other hand, allow for regional to continental modelling of NPP time-series and are potentially transferable to new regions. rnThis thesis thus deals with modelling and analysis of NPP time-series for Kazakhstan andrnpresents a methodological concept for derivation of above-ground biomass (AGB) estimatesrnbased on NPP data. The focus for the methodological development was on biomass estimationrnfor natural and semi-natural environments, which cover about 70% of the area of Kazakhstan.rnFor validation of the results, biomass field data were collected in June 2011 in three study areas in Central, South, and West Kazakhstan. Additionally, field data was also collected in Central Kazakhstan in December 2010.rnFor the selection of an appropriate model, two remote-sensing-based NPP models werernapplied to a study area in Central Kazakhstan. The first is the Regional Biomass Model (RBM), a light-use-efficiency model. The second is the Biosphere Energy Transfer Hydrology Model adapted by DLR (BETHY/DLR), which is a soil-vegetation-atmosphere-transfer model. Both models calculate the NPP on a regional scale of approximately 1 km² spatial resolution and were applied to Kazakhstan for the first time in this dissertation.rnDifferences in the modelling approaches, intermediate products, and calculated NPP, as wellrnas their temporal characteristics were analysed and discussed. The model BETHY/DLRrncalculated higher NPP (mean annual NPP 2010 and 2011: 136.9 g C m-2 and 106.7 g C m-2)rnthan the RBM (62.1 g C m-2 and 54.6 g C m-2) and showed stronger inter-annual changes. Therncomparison to field data from 2011 yielded better results for BETHY/DLR, though the resultsrnfrom both models were highly correlated to the field observations (BETHY/DLR: R=0.97,rnRMSE=8.4 g C m-2; RBM: R=0.99, RMSE=22.5 g C m-2).rnThe model BETHY/DLR was then applied with a regional land cover map to calculate NPPrnfor Kazakhstan for 2003–2011. The results were analysed regarding spatial, intra-annual, andrninter-annual variations. Such detailed analyses of NPP time-series for Kazakhstan have not beenrnavailable before in the literature. The mean annual NPP for Kazakhstan is 143 g C m-2. Thernthree most important natural land cover classes in Kazakhstan and their mean annual NPP are:rngrassland with 140 g C m-2, sparse vegetation with 120 g C m-2, and open shrubland with 112 g C m-2. The maximum productivity is reached in June. NPP anomalies occurred most often inrnagricultural areas in the North of Kazakhstan. Natural semi-arid and arid ecosystems showed a low inter-annual NPP variability.rnIn addition, the correlation between NPP and meteorological parameters was analysed.rnCorrelation between NPP and temperature as well as between NPP and photosyntheticallyrnactive radiation showed correlation coefficients of R>0.6 for more than 90% of the land area.rnThe reaction of vegetation growth to precipitation was delayed one to two months.rnIn the last part of this dissertation, a methodological concept for derivation of above-groundrnbiomass estimates of natural vegetation from NPP time-series has been developed. Thernprocedure aims at estimating the above-ground biomass of herbaceous (grass/herbs) and woody (shrubs) vegetation for Kazakhstan for the period of maximum vegetation growth. Thernmethodological concept is based on the NPP time-series, information about fractional cover ofrnherbaceous and woody vegetation, and plants’ relative growth rates (RGRs). It has been the first time that these parameters are combined for biomass estimation in semi-arid regions.rnThe developed approach was finally applied to estimate biomass for the three study areas inrnKazakhstan. The validation with field data showed the high importance of accurate fractionalrncover information. This information is especially important for estimation of woody biomass.rnWith constant fractional cover values for individual land cover classes, the resulting woodyrnbiomass showed a weak linear correlation to biomass field data. Using fractional coverrninformation from field observations, the developed approach yielded results that showed arnstrong correlation to woody biomass from field data (R=0.83, RMSE=24.4 g C m-2).rnThe validation for herbaceous biomass revealed a moderate correlation to biomass field datarn(R=0.64, RMSE=25.4 g C m-2). An underestimation of herbaceous biomass was observed forrnmost validation sites, which was accounted for by an underestimated NPP. A classification ofrndifferent herbaceous vegetation communities is needed for a better representation of the NPPrnvariability. This in turn will improve the herbaceous biomass estimates.rnThe results of this dissertation provide information about the vegetation dynamics inrnKazakhstan for 2003–2011. This is valuable information for a sustainable land managementrnand the identification of regions that are potentially affected by a changing climate.rnFurthermore, a methodological concept for the estimation of biomass based on NPP time-series is presented. The developed method is potentially transferable. Providing that the required information regarding vegetation distribution and fractional cover is available, the method will allow for repeated and large-area biomass estimation for natural vegetation in Kazakhstan and other semi-arid environments.rn