Module pyl4c.towers
Tools for representing flux tower sites and working with flux tower data.
NOTE: Tower subgrid should accurately reflect the canonical EASE-Grid 2.0 nesting, i.e., the 1-km cells returned correspond to those nested under the respective 9-km cell, which is apparent if we look at the subgrid of the top-left cell:
>>> TowerSite('test', ease2_to_wgs84((0, 0)), validate = False).subgrid
Functions
def consolidate_tower_data(file_glob, output_hdf5_path, n_sites=356, n_steps=6575, keys=('nee', 'gpp', 'reco'))-
Expand source code
def consolidate_tower_data( file_glob, output_hdf5_path, n_sites = 356, n_steps = 6575, keys = ('nee', 'gpp', 'reco')): ''' Combines the CSVs from multiple calibration tower sites into a single HDF5 dataset, e.g.: consolidate_tower_data( "/anx_lagr2/laj/smap/fluxnet2015/merged/*_smapMergedCalVal.csv", "/anx_lagr3/arthur.endsley/SMAP_L4C/calibration/Fluxnet2015_LaThuile_tower_data_for_356_sites.h5") Parameters ---------- file_glob : str GLOB expression targeting all input CSV files output_hdf5_path : str File path to output HDF5 file n_sites : int Number of tower sites n_steps : int Number of time steps keys : tuple Sequence of keys to extract from each CSV ''' file_paths = glob.glob(file_glob) assert len(file_paths) == n_sites, 'Should be one file per tower site' all_records = [] for path in file_paths: site_records = [] with open(path, 'r') as stream: reader = csv.DictReader(stream) for each in reader: site_records.append([ np.float(each[k]) if each[k] != 'NaN' else np.nan for k in keys ]) all_records.append(site_records) site_records = None shp = (n_steps, n_sites) # Output a T x N array arr = np.array(all_records) # Becomes an (N x T x k) array all_records = None with h5py.File(output_hdf5_path, 'a') as hdf: for i, field in enumerate(keys): hdf.create_dataset(field, shp, dtype = np.float32, data = arr[...,i].swapaxes(0, 1))Combines the CSVs from multiple calibration tower sites into a single HDF5 dataset, e.g.:
consolidate_tower_data( "/anx_lagr2/laj/smap/fluxnet2015/merged/*_smapMergedCalVal.csv", "/anx_lagr3/arthur.endsley/SMAP_L4C/calibration/Fluxnet2015_LaThuile_tower_data_for_356_sites.h5")Parameters
file_glob:str- GLOB expression targeting all input CSV files
output_hdf5_path:str- File path to output HDF5 file
n_sites:int- Number of tower sites
n_steps:int- Number of time steps
keys:tuple- Sequence of keys to extract from each CSV
Classes
class MetaTowerSite (*args, **kwargs)-
Expand source code
class MetaTowerSite(type): ''' Prototype for Tower classes. ''' @property def pft_1km_map(cls): if getattr(cls, '_pft_1km_map', None) is None: cls._pft_1km_map = get_pft_array('M01') return cls._pft_1km_map @property def pft_9km_map(cls): if getattr(cls, '_pft_9km_map', None) is None: cls._pft_9km_map = get_pft_array('M09') return cls._pft_9km_mapPrototype for Tower classes.
Ancestors
- builtins.type
Instance variables
prop pft_1km_map-
Expand source code
@property def pft_1km_map(cls): if getattr(cls, '_pft_1km_map', None) is None: cls._pft_1km_map = get_pft_array('M01') return cls._pft_1km_map prop pft_9km_map-
Expand source code
@property def pft_9km_map(cls): if getattr(cls, '_pft_9km_map', None) is None: cls._pft_9km_map = get_pft_array('M09') return cls._pft_9km_map
class TowerSite (id, coords, validate=True)-
Expand source code
class TowerSite(object, metaclass = MetaTowerSite): ''' Represents an eddy covariance tower site. Parameters ---------- id : str Unique name for the TowerSite coords : tuple | list The site's (longitude, latitude) coordinates validate : bool True to validate the PFT of any mapped EASE-Grid 2.0 cell ''' valid_pft_range = range(1, 9) def __init__(self, id, coords, validate = True): assert hasattr(coords[0], 'real') and hasattr(coords[1], 'real'),\ 'Coordinates must be numbers, not strings or other types' self.id = id self.coords = coords self.idx_1km = ease2_from_wgs84(coords, 'M01', exact = True) self.idx_9km = ease2_from_wgs84(coords, 'M09', exact = True) self._subgrid_idx = None self._subgrid_pft = None self._validate = validate if validate: self.locate_ease2() # Fire it up! @cached_property def pft_9km(self): 'The PFT class at 9-km scale' r, c = self.idx_9km return self.__class__.pft_9km_map[int(r), int(c)] @property def subgrid(self): 'Returns a list of 1-km subgrid indices' if self._subgrid_idx is None: # A search radius of 4 pixels yields a width, height of 9 pixels self._subgrid_idx = ease2_search_radius(self.coords, 4, 'M01') return self._subgrid_idx @property def subgrid_pft(self): 'Returns an 81-element array of the 1-km subgrid PFTs' if self._subgrid_pft is None: self._subgrid_pft = index( self.__class__.pft_1km_map, np.array(self.subgrid).swapaxes(0, 1).tolist()) return np.array(self._subgrid_pft) def locate_ease2(self): ''' Finds the EASE-Grid 2.0 row, column indices for this site. Returns ------- list A list of tuples, each a 2-element sequence of (row, column) indices ''' row, col = self.idx_9km pft = self.__class__.pft_9km_map[int(row), int(col)] # Expensive if self._validate and pft not in self.valid_pft_range: # Get the closest valid EASE-Grid 2.0 cell instead try: row, col = self.locate_ease2_alternatives(self.coords) # Chain forward and reverse transforms to get the grid # cell indices in floating point row, col = ease2_from_wgs84( ease2_to_wgs84((row, col)), 'M09', exact = True) except ValueError: raise ValueError('Could not locate suitable EASE-Grid 2.0 cell') # Update the 9-km grid coordinates, as needed self.idx_9km = row, col return self.idx_9km def locate_ease2_alternatives(self, coords, grid = 'M09'): ''' Finds nearest EASE-Grid 2.0 cell w/ valid PFT up to 1 grid cell away. Parameters ---------- coords : tuple or list 2-element sequence of (longitude, latitude) coordinates grid : str EASE-Grid 2.0 cell size designation, e.g, "M01" or "M09" (Default: "M09") Returns ------- list A list of tuples, each a 2-element sequence of (row, column) indices ''' # Round the row, column indices of this point so as to check for # collisions in haversine(), below row, col = ease2_from_wgs84(coords, grid, exact = False) indices = ease2_search_radius(coords, 1, grid) indices_pfts = index( self.__class__.pft_9km_map, np.array(indices).swapaxes(0, 1).tolist()) options = [] # Make sure the choices also have valid PFTs for i, idx in enumerate(indices): if indices_pfts[i] in self.valid_pft_range: options.append(idx) dists = [ # Calculate great circle distance haversine( coords, ease2_to_wgs84((r,c), grid) ) for r, c in options if not (r == row and c == col) ] return options[np.argmin(dists)]Represents an eddy covariance tower site.
Parameters
id:str- Unique name for the TowerSite
coords:tuple | list- The site's (longitude, latitude) coordinates
validate:bool- True to validate the PFT of any mapped EASE-Grid 2.0 cell
Class variables
var valid_pft_range
Instance variables
var pft_9km-
Expand source code
def __get__(self, obj, cls): if obj is None: return self if asyncio and asyncio.iscoroutinefunction(self.func): return self._wrap_in_coroutine(obj) value = obj.__dict__[self.func.__name__] = self.func(obj) return valueThe PFT class at 9-km scale
prop subgrid-
Expand source code
@property def subgrid(self): 'Returns a list of 1-km subgrid indices' if self._subgrid_idx is None: # A search radius of 4 pixels yields a width, height of 9 pixels self._subgrid_idx = ease2_search_radius(self.coords, 4, 'M01') return self._subgrid_idxReturns a list of 1-km subgrid indices
prop subgrid_pft-
Expand source code
@property def subgrid_pft(self): 'Returns an 81-element array of the 1-km subgrid PFTs' if self._subgrid_pft is None: self._subgrid_pft = index( self.__class__.pft_1km_map, np.array(self.subgrid).swapaxes(0, 1).tolist()) return np.array(self._subgrid_pft)Returns an 81-element array of the 1-km subgrid PFTs
Methods
def locate_ease2(self)-
Expand source code
def locate_ease2(self): ''' Finds the EASE-Grid 2.0 row, column indices for this site. Returns ------- list A list of tuples, each a 2-element sequence of (row, column) indices ''' row, col = self.idx_9km pft = self.__class__.pft_9km_map[int(row), int(col)] # Expensive if self._validate and pft not in self.valid_pft_range: # Get the closest valid EASE-Grid 2.0 cell instead try: row, col = self.locate_ease2_alternatives(self.coords) # Chain forward and reverse transforms to get the grid # cell indices in floating point row, col = ease2_from_wgs84( ease2_to_wgs84((row, col)), 'M09', exact = True) except ValueError: raise ValueError('Could not locate suitable EASE-Grid 2.0 cell') # Update the 9-km grid coordinates, as needed self.idx_9km = row, col return self.idx_9kmFinds the EASE-Grid 2.0 row, column indices for this site.
Returns
list- A list of tuples, each a 2-element sequence of (row, column) indices
def locate_ease2_alternatives(self, coords, grid='M09')-
Expand source code
def locate_ease2_alternatives(self, coords, grid = 'M09'): ''' Finds nearest EASE-Grid 2.0 cell w/ valid PFT up to 1 grid cell away. Parameters ---------- coords : tuple or list 2-element sequence of (longitude, latitude) coordinates grid : str EASE-Grid 2.0 cell size designation, e.g, "M01" or "M09" (Default: "M09") Returns ------- list A list of tuples, each a 2-element sequence of (row, column) indices ''' # Round the row, column indices of this point so as to check for # collisions in haversine(), below row, col = ease2_from_wgs84(coords, grid, exact = False) indices = ease2_search_radius(coords, 1, grid) indices_pfts = index( self.__class__.pft_9km_map, np.array(indices).swapaxes(0, 1).tolist()) options = [] # Make sure the choices also have valid PFTs for i, idx in enumerate(indices): if indices_pfts[i] in self.valid_pft_range: options.append(idx) dists = [ # Calculate great circle distance haversine( coords, ease2_to_wgs84((r,c), grid) ) for r, c in options if not (r == row and c == col) ] return options[np.argmin(dists)]Finds nearest EASE-Grid 2.0 cell w/ valid PFT up to 1 grid cell away.
Parameters
coords:tupleorlist- 2-element sequence of (longitude, latitude) coordinates
grid:str- EASE-Grid 2.0 cell size designation, e.g, "M01" or "M09" (Default: "M09")
Returns
list- A list of tuples, each a 2-element sequence of (row, column) indices