Future vegetation projection by Earth system model and its impact on nomadism

Abstract

This paper reports the progress of our efforts to apply the livestock weight model developed by ourselves in the past to future scenarios. First, I analyzed the results of future scenario experiments of global climate models that incorporate biogeochemical cycles including ecosystems (e.g., Earth system model), and confirmed that the amount of vegetation around Mongolia will basically be projected to increase, and that the degree of this increase is more emphasized in the high temperature scenarios. This can be attributed to the effects of carbon dioxide fertilization, higher temperatures, and higher nitrogen concentrations by fertilizers etc. Next, to prepare input data for the livestock weight model, I presented the results of attempting downscaling method combining an offline vegetation model and a convolutional neural network. The former method achieved a spatial resolution of 0.5°×0.5°, which could be further refined to 8 km×8 km by the latter method. Although some technical problems remain, the results showed the possibility of obtaining future vegetation distribution with sufficient spatial resolution for analyzing nomadism. Finally, using the Leaf Area Index (LAI) distribution for 2021-2030 with RCP 8.5 scenario derived by the above method, and given the assumption that livestock move to the grid with the largest LAI value in the surrounding area each month, the LAI of the grid where livestock stay was calculated. Here, the average LAI for the year preceding the month of April, when livestock weight drops, was presented. In the future, I intend to further improve the downscaling method and the livestock weight model.