**7. Modeling of CH4 emission**

To improve the prediction of climate models, it is important to understand the mechanisms by which microorganisms regulate the flow of terrestrial GHG. This involves considering the complex interactions that occur between microorganisms and other biotic and abiotic factors in the environment. The potential to mitigate climate change by reducing GHG emission through the management of terrestrial microbial processes is a perspective of high importance for the future [18].

Despite this importance, however, tropical flood areas are poorly represented in global models to predict global CH4 emission. A first step in the development of a process-based model of CH4 emission from tropical flood areas for global applications was documented in 2014. To this end, the LPX-Bern Dynamic Global Vegetation Model (LPX) was slightly modified to represent the hydrology of the floodplain, vegetation, and associated CH4 emission. The extent of tropical floodplains was prescribed using the production of the spatially explicit PCR-GLOBWB hydrology model. Several variables were introduced to this model, such as vegetation, ground cover (through remote sensing), not to mention that simulated CH4 flow densities were evaluated against field observations and regional flow inventories. However, soil microbiota was not considered as a component in the modeling. Simulated CH4 emissions at the Amazon basin scale were compared to simulations of previously performed models. Thus, it was found that this LPX model reproduces the average magnitude of the net flow densities of CH4 observed for the Amazon basin. However, the model does not reproduce the temporal and spatial variability between sampling sites, considering that site information is too limited to attest or refute some resources of the model. At the Amazon basin scale, the results obtained with the promotion of this model highlighted the great uncertainty in the magnitude of CH4 emission from floodable areas.

The sensitivity analysis provided clarification on the main drivers of CH4 emission from the floodplain and their associated uncertainties. Due to an intrinsic limitation of the LPX to consider seasonality in floodplain extension, the model failed to reproduce the total dynamics of CH4 emission, raising several scientific questions. Although this model includes more specific mechanisms for tropical floodplains, it was not possible to reduce the uncertainty in the magnitude of CH4 emission from the Amazon basin, thus justifying the need for further research to restrict CH4 emission and their temporal variability [15].

In the same year, Potter et al. [76] developed a new model that sought to seasonally estimate the carbon dynamics and CH4 emission of floodable ecosystems in the Amazon. The Amazon wetland simulation model took into account three main components: (a) details of the type of vegetation in the wetlands and changes in the level of water, temperature, and dissolved oxygen; (b) primary production, mass accumulation, and decay of the litter layer in soils and sediments; and (c) routes for production and transport of CH4 through the water column to the atmosphere.

The presented model is based on the input of the following data for simulations in a given flooded environment in the Amazon: latitude and longitude; vegetation types such as area cover fractions; daily surface temperature; solar irradiance flux; wind speed; precipitation; daily water depth; biomass production values for floating macrophytes; and satellite vegetation index data for flooded forest ecosystems. In order to improve the generality and use of this model, the incorporation of mechanical simulations of vertical mixing, horizontal exchanges, and various biogeochemical processes is necessary. In addition, the microbiota component is not directly reported.

In 2016 [77], when evaluating the atmospheric concentrations of CH4 in the Amazon basin in 2010 and 2011, besides a 3D atmospheric chemical transport model (TOMCAT), two emission models in wetlands have been used [78–79] to reduce the uncertainty about CH4 emission. The first set of wetland and rice paddy emission derived from the Bloom et al. method [79]. The method uses a satellite to evaluate the carbon variation available for methanogenesis, which leads to a more accurate representation of the timing of CH4 emission. However, satellite data cannot distinguish between microbial CH4 emission from natural wetlands and anthropogenic emission from rice cultivation. The second model [the Joint UK Land Environment Simulator (JULES), version 3.4.1] [78] simulates the Earth's land surface in terms of carbon, water, and energy variations and includes a methane flux in wetlands as a component, based on Gedney et al. [80]. The flow of CH4 is dependent on the available carbon substrate, the temperature, and the fraction considered wet. The estimates used through the two wetlands emission models are based on processes and showed similar behaviors when the atmospheric model is compared to observations, regardless of which model was used [77].

In the same year (2016), another research on CH4 modeling was carried out, bringing to the fore the discussion of how beneficial the improvements in CH4 models would be for terrestrial system models and for the additional simulation of climatecarbon cycle feedbacks. Over the past four decades, several empirical models have been developed to quantify the magnitude, investigate spatial and temporal variations, and understand the underlying mechanisms and environmental controls of CH4 (CH4 flows in terrestrial ecosystems). These CH4 models are also used for the integration of multiple-scale CH4 data, such as laboratory-based incubation and molecular analysis, field observational experiments, remote sensing, and aircraft-based measurements in various terrestrial ecosystems. The authors noted that there are large discrepancies between models in terms of representation of CH4 processes and their environmental controls, and significant data, such as model incompatibilities, are partially attributed to different representations of landscape characterization and flood dynamics.

However, it should be noted that CH4 models should represent more explicitly the mechanisms underlying the exchange of Earth-atmosphere CH4, with emphasis on the improvement and validation of individual CH4 processes over depth and horizontal space, and models capable of simulating CH4 emissions at highly heterogeneous spatial and temporal scales, particularly in hotspots, should be developed; besides that, efforts should be made to develop benchmarking models (a modeling based on comparative analysis) that can be easily used for improvement, evaluation, and integration with data from molecular to global scales [81].

Widely applicable and robust prediction models should be developed from large data sets generated through collaboration with scientists around the world. To achieve high predictive accuracy, these data sets should cover a wide variety of information and variables at the most different scales of tropical floodplains within regions and globally.
