【元胞自动机】基于元胞自动机模拟雪反照率更新问题附matlab代码 上传版本.zip

# CASPA : Cellular Automaton for SnowPack Albedo README written and maintained by J Cook, University of Sheffield The code is written in Matlab and is designed to model snow albedo using radiative transfer by invoking the BioSNICAR model (Cook et al., 2017) in a cellular automaton framework. BioSNICAR is used to calculate the albedo of a snowpack in each cell in a grid i x j. At each timestep the algal biomass is updated according to a user defined doubling time and a grain size evolution model adapoted from the Community Land Model and Flanner and Zender(2006) is used to update the snow grain sizes and layer thicknesses, driven by energy fluxes predicted by te radiative transfer model. This code is in active development and is not yet published. There are no permissions granted for copying, applying or using this code. # OVERVIEW This model simulates a snow surface with initial physical/optical properties defined by the user. This surface is represented as a cellular automaton, i.e. an x,y grid composed of discrete cells that update based upon certain functions as the model advances through time. The model simulates an algal bloom that grows from an initial configuration defined by the user. The user defines the % coverage at t=0. At each timestep, algal growth is simulated using a probabalistic function, where growth in situ has a given likelihood and spread of the algae into neighbouring cells (Moore neighbourhood) has a given likelihood. Growth is represented by increasing the cell value - i.e. 0 represents clean ice, 1-10 represent increasing mixing ratio of algae in the upper surface. A carrying capacity is simulated wherein the cell value cannot exceed a value of 10. Once the cell value reaches 10 the bloom can only grow by spreading. If all the neighbouring cells are also at carrying capacity, the algae will not grow. At each timestep, the grid is thereby updated. The value of each cell is associated with a call to a specific instance of the radiative transfer scheme BioSNICAR. BioSNICAR is run offline coupled to a grain size evolution model that uses the energy fluxes in each vertical layer of the snowpack predicted by BiOSNICAR. The grain size evolution model accounts for wet and dry grain growth and accumulation of liquid water. Liquid water is generated whenever there is excess energy that raises the snow temperature abive 273.15K. This water can pecolate into lower layers. If the lower layers have sub-freezing temperatures the water will refreeze. refrozen water is assumed to have a radius of 1500 microns. Liquid water that remains in situ is modelled by increasing the optical radius of the ice grains. The optical properties of the snow is therbey updated for each instance of BioSNICAR and the broadband and spectal albedo are returned for each surface class. These are added to a lookup table accessible by CASPA as it updates the cellular model each timestep. Running CASPA with the grain evolution model ON and OFF and differencing the twp outputs provides a measure of direct and indirect effects of algae on albedo and radiative forcing. A spatially-integrated spectral and broadband albedo for the entire grid is computed using a 2D-mean. Biomass is calculated from the volume, density and mixing ratio of impurities in the upper snow layer. The spectral albedo and spetcral irradiance is used to calculate the radiative forcing following Ganey et al (2017: Nat Geoscience) and Dial et al (2018: FEMS) by multipling the incoming irradiance by 1-albedo for clean and 'mixed' surfaces, the difference between which provides the RF caused by impurities. One timestep is dimensionless, except that successive SNICAR instances have 2x algal biomass in upper layer, meaning 1 timestep is equal to the doubling time of the algae, making the time dimension tractable. For a doubling time of 3 days, one timestep = 3 days. Albedo change / doubling time (days) = albedo change per day. # RUNNING THE MODEL Running the model in default mode is highly automated and controlled from a driver script (CASPA_driver.m). The CASPA repository should be cloned or downloaded and all files saved to the working directory. The user-defined variables are assigned in the driver. The grain size evolution code can be turned on or off and the properties of the snowpack defined. The scavenging rate for mineral dusts is user defined. To have a dust free model run, set the initial mixing ratio (y) to zero. To have a constant dust mixing ratio throughout the run, set the scavenging rate to 1. Running the driver first calls the setup functions. This will populate the workspace with all relevant datasets. Then CASPA is called (CASPA_V01.m at time of writing README). The script will produce plots of the snow surface showing the growing algal blooms, albedo and IRF against time, biomass and coverae against time, and biomass and albedo against time. To tweak the inputs, user-defined variables can be found all together near the start of the script. For more in-depth changes, such as SNICAR conditions, bio-optical parameters etc, refer to the main documentation provided in this repository. # Model Development Prior to Initial Commit to Github These notes detail the state of the code prior to the first upload to GitHub, for reference. ## CASPA v 0.01 This version is the cellular automaton only, i.e it evolves a surface using classifiers but the model does not use these classifiers to call SNICAR. This was the very first stage of model development. ## CASPA v 0.02 This version was the first to call SNICAR from the cellular model. It looped through each cell in the grid and called SNICAR each time. This was slow and was raised as an urgent TODO for v0.03. ## CASPA v 0.03 This version no longer runs SNICAR per cell. Instead, all the instances of SNICAR used in the model are run once at the beginning and the spectral and broadband data appended to lists. These are then used as a lookup table for the cellular model. Of course, this has drastically reduced the run time. Bug fix in final loop - previously able to increase cell values above 10. This was because when an individual cell reached carrying capacity the algae grew by spreading but was able to spread into cells whose values > 10. This is now fixed so no cell can exceed the carrying capacity. Condense loops - now single loop for if rand>50 OR cell value =10 (previously two sequential loops) ## CASPA v 0.04 Updated colormap to show clean snow as white, then grades of red as snow algae bloom. This version is the first to include radiative forcing calculations in addition to albedo. This is achieved by differencing the spectral albedo of the simulated snowpack and a hypothetical clean snowpack then multiplying 1 - albedo by incoming spectral irradiance and normalising in space/time. There were also some minor changes to variable names etc. to imporve the readability.