Wildfires are a regular thread in many regions on Earth, so also to areas underlain by permafrost. Often the uppermost layer of permafrost contains huge amounts of organic material or completely consists of organic material like peat, moss or lichen (e.g., Beringer et al., 2001). Wildfires can burn this organic material. The degree to which this material is burned depends, among other things, on fire intensity, fire duration, and total soil water content of the material. Fires heat the soil and huge amounts of soil water evaporate during the fire. In permafrost, fire-induced changes in soil temperature go along with changes in total soil water content and the partitioning of the water phases (e.g., Hinzman et al., 2003). Consequently, infiltration, soil volumetric heat capacity and hydraulic conductivity before and after a fire differ appreciably. As compared to pre-fire soil conditions, post-fire soils are warmer. Such modified hydro-thermodynamic states of soil remain detectable long after the fire events. Due to their impact for soil temperature regime, active layer depth and soil surface temperature on the short and long-term it would be important to consider the impact of wildfire on soil temperature in the soil models of atmospheric models.
Currently, the impact of wildfires on permafrost is not considered in routine weather forecasts, CTMs, GCMs or ESMs. In CTMs, wildfire impacts on permafrost are currently neglected even when the CTMs are applied for wildfire smoke forecasts in areas underlain by permafrost. The neglecting wildfire impacts on permafrost in ESMs is despite some ESMs consider random aerosol release from wildfires and wildfire related land-cover changes in the biogeochemical cycles. One application that considered the impact of wildfires by land-cover changes and soil-temperature and moisture changes was performed with an NWPM by Molders and Kramm (2007). Their results showed that the relatively warmer burned areas may increase atmospheric buoyancy and hence locally convection.
The lack of horizontally and vertically high resolved soil data for organic and mineral soil, uncertainty in soil parameters, and organic soils are among the biggest challenges in modeling permafrost in atmospheric applications.
Due to the lack of 3D data on the soil type distribution most modern soil models used in NWPMs, CTMs, GCMs or ESMs assume one soil-type - typically that of the uppermost soil -for the entire soil column (e.g., Slater et al., 1998, Schlosser et al., 2000). Investigations show that this simplification/assumption results in missing many details in predicted soil-temperature patterns that result from the vertical soil type profile (Fig. 7). Due to neglecting vertical soil type profile characteristics the variance of simulated and observed temporal evolution of soil temperature can differ significantly with consequences for predicted active layer depth that may be dislocated about ±0.4 m or so (e.g., Molders and Romanovsky 2006). Off-line simulations (i.e., the feedback processes between the atmosphere and surface are not considered) with different soil hydraulic models that were run with and without ensuring consistent soil hydraulic parameters, demonstrated that uncertainty in soil hydraulic parameters overwhelms that in the theory of soil hydraulic models (Shao and Irannjad, 1999).
Evaluating soil temperature and moisture conditions simulated by NWPMs, ESMs, or GCMs has been a high priority of the third PILPS phase (e.g., Henderson-Sellers et al., 1995). PILPS demonstrated that results obtained from LSMs coupled to GCMs differ on the same order of magnitude as off-line PILPS experiments; differences in LSM complexity may cause statistically significant differences in temperature, pressure, and turbulent fluxes over land (e.g., Sato et al., 1989, Thompson and Pollard 1995, Yang et al., 1995, Qu and Henderson-Sellers 1998). The results of PILPS also suggested that a soil model must be able to capture soil-temperature conditions well when run offline with observed atmospheric forcing and known site-specific parameters (necessary condition), and it must be re-evaluated when being implemented in a NWPM, GCM or ESM (sufficient condition).
Soil models of NWPMs are typically evaluated by assuming that the soil temperature and moisture measurements at a site are representative for the grid-cell within which the site is located (e.g., Chen and Dudhia 2002, Narapusetty and Molders 2005). It is well known that some discrepancies may arise due to the fact that the model grid-cell represents a volume-average condition for several square-kilometers of several centimeters thickness. For GCMs or ESMs, however, simulated soil temperature and moisture states represent even larger volumes. Due to the large area of several 100 square-kilometers covered by GCM or ESM grid-cells often several sites exist within the same grid-cell. Thus, a comparison like performed for NWPMs becomes highly ambiguous. Therefore, GCM and ESM simulations of soil regimes are typically evaluated using gridded climatologies that are derived from point observations projected on to a grid by some kind of interpolation methods (e.g., Li 2007, PaiMazumder et al., 2008).
Recently, the digital versions of the Ground Ice Conditions map and the International Permafrost Association (IPA) Circum-Arctic Map of Permafrost (known as IPA map) were combined with ancillary data sets of the Global Land One-kilometer Base Elevation data base and the global land-cover characteristics data base to provide a gridded distribution of northern hemispheric permafrost and ground ice (Zhang et al., 2000). Such gridded data can serve for evaluation of 20th century simulations of GCMs and ESMs. This dataset, however, does not contain soil temperature or moisture conditions.
Any gridded data sets bear some uncertainty from various sources. First the data stem from routine monitoring that typically has less accuracy than specialized field campaigns. Furthermore, these data have been collected for other reasons than evaluation of GCMs or ESMs. Thus, the monitoring networks may not be representative for the landscape that a GCMs or ESMs is to cover. For evaluation of CCSM3 simulated soil-temperature climatologies in Siberia PaiMazumder et al. (2008) used gridded data based on over 400 agricultural monitoring sites. They found December-biases in soil-temperature climatology for CCSM3 of up to 6 K at 0.2 m depth of which they could explain about 2.5 K by incorrect simulated atmospheric forcing. It is obvious that the soil conditions represented by the gridded data derived there from are biased with respect to the conditions in well drained, fertile soils with other density than non-plowed soils. Moreover, agriculture is typically made on soils that have a relative deep active layer depth. Thus, great care is needed in interpreting simulated soil conditions when using these kinds of gridded data. Investigations by PaiMazumder and Molders (2008) showed that such bias in representing the soil distribution can lead to overestimation of soil-temperature amplitudes of more than 1 K and difficulties in capturing the phase. These findings also suggest that some of the discrepancies found for GCM or ESM soil temperature simulations may be explained the networks on which the gridded climatologies are based. Taking the errors resulting from incorrect forcing and the gridded data into account, only about 1.5 K of the bias found by PaiMazumder et al. (2008) for the CCSM3 soil-temperature simulations may stem from model deficits or other error sources.
Was this article helpful?