Burning trees in frozen soil: Simulating fire, vegetation, soil, and hydrology in the boreal forests of Alaska, 2022.

Boreal ecosystems account for 29% of the world's total forested area and contain more carbon than any other terrestrial biome. Over the past 60 years, Alaska has warmed twice as rapidly as the contiguous U.S. and wildfire activity has increased, including the number of fires, area burned, and frequency of large wildfire seasons. These recent and rapid changes in climate and wildfire have implications for future vegetation composition, structure, and biomass in interior Alaska, given that the vegetation is highly dependent on active layer thickness, soil moisture, organic layer depth, and plant-available nutrients. Here we developed a new succession extension (DGS) of the LANDIS-II forest landscape model which integrates a vegetation dynamics model (NECN) with a soil carbon model (DAMM-McNiP), a hydrologic model (SHAW), and a deep soil profile permafrost model (GIPL) in a spatially-explicit framework. DGS Succession uses the algorithms in the NECN Succession extension of LANDIS-II to simulate growth, mortality and reproduction of vegetation but has three major improvements. First, the simple bucket model in NECN was replaced with a physically-based model (SHAW) that simulates energy and water fluxes (e.g. snow depth, evapotranspiration, soil moisture) at multiple levels in the canopy and soil. Second, the active, slow, and passive soil pools in NECN were replaced by seven soil pools that are measurable in the field, with carbon and nitrogen dynamics dictated by DAMM-McNiP. Finally, soil temperature and soil moisture are simulated only at one depth in NECN, but in DGS, soil temperature (and hence permafrost dynamics) are simulated at as many as 50 user-defined depths down to 4 meters (m) with SHAW and 75 m with GIPL. During the initial calibration phase, DGS was applied at three inventory sites at the Bonanza Creek Long Term Ecological Research area in Interior Alaska where climate forcings, species biomass, soil temperature, and/or soil moisture were available. For the landscape-scale simulations, DGS was run with the SCRPPLE fire extension of LANDIS-II under two scenarios of climate using a ∼400,000 hectare (ha) landscape that included the inventory sites. Across all three sites, DGS generally captured the variation in soil moisture and temperature across depths, seasons, and years reasonably well, though there were some discrepancies at each site. DGS had better agreement with field measurements of soil moisture and temperature than its predecessor NECN which produced unrealistically low soil moisture and unrealistically high seasonal fluctuations in soil temperature. At the landscape scale, ignitions, area burned, and soil temperature increased under climate change, as expected, while soil moisture was relatively unchanged across climate scenarios. Biomass tended to decline under climate change, which differs from other modeling studies in this region but is consistent with the browning trends observed from remote sensing data. Simulating climate, vegetation succession, hydrology, permafrost, carbon and nutrient cycling, and wildfire in an integrated, spatially-explicit framework like LANDIS-II will allow us to disentangle the drivers and ecosystem responses in this rapidly changing ecosystem, as well as other forested systems with complex hydrologic, biochemical, cryospheric, and vegetation feedbacks.

北方森林生态系统（Boreal ecosystems）占全球森林总面积的29%，其储存的碳量超过其他任何陆地生物群系。过去60年间，阿拉斯加的升温速率是美国本土的两倍，野火活动也随之加剧，具体表现为火灾数量、过火面积以及大型野火季发生频率均有所增加。鉴于阿拉斯加内陆的植被高度依赖活动层厚度（active layer thickness）、土壤湿度、有机层深度以及植物有效养分，近期气候与野火发生的快速变化，将对该区域未来的植被组成、结构与生物量产生深远影响。本研究开发了LANDIS-II森林景观模型（LANDIS-II forest landscape model）的全新演替扩展模块DGS，该模块在空间显式框架下，整合了植被动态模型（NECN）、土壤碳模型（DAMM-McNiP）、水文模型（SHAW）以及深层冻土剖面模型（GIPL）。DGS演替模块沿用了LANDIS-II的NECN演替扩展模块中的算法，以模拟植被的生长、死亡与繁殖过程，同时具备三项核心改进：其一，将NECN中的简单水桶模型替换为基于物理过程的SHAW模型，该模型可在冠层与土壤的多个层级模拟能量与水通量（如积雪深度、蒸散发、土壤湿度）；其二，将NECN中的活性、缓性与惰性土壤碳库替换为7种野外可测量的土壤碳库，其碳氮动态由DAMM-McNiP驱动；最后，NECN仅能模拟单一深度的土壤温度与湿度，而DGS则通过SHAW模型可模拟最多50层用户自定义深度的土壤温度（进而模拟冻土动态），最深可达4米；依托GIPL模型则可模拟至75米深度。在校准初始阶段，研究团队将DGS模块应用于阿拉斯加内陆博纳扎溪长期生态研究区（Bonanza Creek Long Term Ecological Research area）的3处调查样地，这些样地均具备气候强迫因子、物种生物量、土壤温度以及/或土壤湿度的观测数据。针对景观尺度模拟，研究团队以包含上述调查样地的约40万公顷（ha）研究区为对象，结合LANDIS-II的SCRPPLE火灾扩展模块，在两种气候情景下运行DGS模块。在全部3处样地中，DGS模块均能较好地复现不同深度、季节与年份下的土壤湿度与温度变化，尽管各样地仍存在一定偏差。相较于其前身模型NECN，DGS与野外实测的土壤湿度、温度数据吻合度更高——NECN会得出异常偏低的土壤湿度以及异常偏高的土壤温度季节波动。在景观尺度上，正如预期，气候变化情景下火点数量、过火面积与土壤温度均有所上升，而土壤湿度在不同气候情景下相对稳定。气候变化情景下，区域生物量整体呈下降趋势，这与该区域其他建模研究的结果存在差异，但与遥感观测到的褐变趋势一致。通过在LANDIS-II这类空间显式框架中整合气候、植被演替、水文、冻土、碳氮循环以及野火过程，本研究可厘清这一快速变化的生态系统（以及其他具备复杂水文、生物化学、冰冻圈与植被反馈的森林生态系统）的驱动因子与生态系统响应。