Title: Estimating aboveground biomass in the forest-steppe zone in Mongolian rangelands using spectroradiometer and remote sensing data

Rangelands cover 70% of Mongolia and support forage for livestock and wildlife. Their aboveground biomass absorbs carbon and carries out photosynthesis, and the soil stores a large amount of carbon. Calculating aboveground biomass and mapping vegetation cover provides important background information for the implementation of rangeland management and using remote sensing for the biomass monitoring is crucial to save manpower, costs, and time. In this project, the aim was to develop a model to estimate aboveground biomass in the forest-steppe of Mongolia. This was done by analysing the relationship between the integrated data obtained from a Sentinel-2 Multispectral Instrument dataset, a ASD FieldSpec 4 Standard-Res portable spectroradiometer data, DJI Phantom 4 multispectral drone data and field sampling data, using the Random Forest algorithm. NDVI and MSAVI calculated from the Sentinel-2 Multispectral Instrument dataset and portable spectroradiometer data gave the best correlation with the field data and were thus used to develop the Random Forest model. Furthermore band 8 (NIR), band 4 (Red), band 3 (Green), and band 2 (Blue) of the Sentinel-2 Multispectral Instrument dataset were used for the Random Forest model. The Random Forest algorithm randomly selected 20% of all sampling data for testing and 80% of all sampling data for training in Python. The assessment, built on the Random Forest model results, are: RMSE = 36.3 kg/ha, RMSE% = 0.165, and R² = 0.94. I mapped the spatial distribution of predicted aboveground biomass in the test area using the Random Forest model. The study results showed that vegetation indices and spectral bands derived from the Sentinel-2 Multispectral Instrument dataset and portable spectroradiometer data used in the Random Forest model have good potential for estimating aboveground biomass in the forest-steppe zone of Mongolia.