Upscaling of soil methane fluxes from topographic attributes derived from a digital elevation model in a cold temperate mountain forest

Abstract. Forest soils are generally considered a sink for atmospheric methane (CH4), but their uptake rate can vary considerably in space and time. This study investigated the role of topography and vegetation on soil CH4 fluxes and the temporal patterns of spatially upscaled soil CH4 fluxes in a topographically complex cold-temperate mountain forest in central Japan. We measured soil CH4 fluxes nine times during the snow-free season at multiple locations within a 40 ha area in a forested watershed. Non-waterlogged soils were a sink of CH4, while small wetland patches emitted CH4 consistently throughout the study period. We used a machine-learning approach to upscale the measured soil CH4 fluxes to the landscape scale for non-waterlogged soils at each date of measurement, using topographic and vegetation attributes derived from a digital elevation model and aerial images. The accuracy of predicted fluxes varied seasonally, with the highest model performance observed in early autumn (R2=0.67) and the lowest in mid-summer (R2=0.31). Predicted CH4 fluxes varied significantly across topographic positions, with greater uptake on ridges and slopes than on the plain and foot slopes. Topography played a predominant role compared to vegetation in the spatial variability of CH4 fluxes. Predicted CH4 fluxes at the landscape scale in the non-waterlogged area ranged from −0.34 to −0.60 gCH4ha-1h-1 in spring, −0.39 to −1.28 gCH4ha-1h-1 in summer, and −0.48 to −0.89 gCH4ha-1h-1 in autumn. Seasonal fluxes were highly correlated with the 20 d antecedent precipitation index (R2=0.70), revealing the importance of seasonal moisture conditions in regulating CH4 flux dynamics. This study highlighted the importance of topography in controlling soil CH4 fluxes and the efficiency of remote sensing and machine learning approaches to scale field measurements to the landscape level, enabling visualization of spatial patterns of fluxes across the landscape over time, despite high uncertainty on some measurement dates, particularly for low elevation pixels.