class ITBFatalityFunction(FunctionProvider): """Indonesian Earthquake Fatality Model This model was developed by Institut Teknologi Bandung (ITB) and implemented by Dr. Hadi Ghasemi, Geoscience Australia. Reference: Indonesian Earthquake Building-Damage and Fatality Models and Post Disaster Survey Guidelines Development, Bali, 27-28 February 2012, 54pp. Algorithm: In this study, the same functional form as Allen (2009) is adopted to express fatality rate as a function of intensity (see Eq. 10 in the report). The Matlab built-in function (fminsearch) for Nelder-Mead algorithm was used to estimate the model parameters. The objective function (L2G norm) that is minimised during the optimisation is the same as the one used by Jaiswal et al. (2010). The coefficients used in the indonesian model are x=0.62275231, y=8.03314466, zeta=2.15 Allen, T. I., Wald, D. J., Earle, P. S., Marano, K. D., Hotovec, A. J., Lin, K., and Hearne, M., 2009. An Atlas of ShakeMaps and population exposure catalog for earthquake loss modeling, Bull. Earthq. Eng. 7, 701-718. Jaiswal, K., and Wald, D., 2010. An empirical model for global earthquake fatality estimation, Earthq. Spectra 26, 1017-1037. Caveats and limitations: The current model is the result of the above mentioned workshop and reflects the best available information. However, the current model has a number of issues listed below and is expected to evolve further over time. 1 - The model is based on limited number of observed fatality rates during 4 past fatal events. 2 - The model clearly over-predicts the fatality rates at intensities higher than VIII. 3 - The model only estimates the expected fatality rate for a given intensity level; however the associated uncertainty for the proposed model is not addressed. 4 - There are few known mistakes in developing the current model: - rounding MMI values to the nearest 0.5, - Implementing Finite-Fault models of candidate events, and - consistency between selected GMPEs with those in use by BMKG. These issues will be addressed by ITB team in the final report. Note: Because of these caveats, decisions should not be made solely on the information presented here and should always be verified by ground truthing and other reliable information sources. :author Hadi Ghasemi :rating 3 :param requires category=='hazard' and \ subcategory=='earthquake' and \ layertype=='raster' and \ unit=='MMI' :param requires category=='exposure' and \ subcategory=='population' and \ layertype=='raster' """ title = tr('Die or be displaced') synopsis = tr( 'To asses the impact of earthquake on population based on earthquake ' 'model developed by ITB') citations = tr( ' * Indonesian Earthquake Building-Damage and Fatality Models and ' ' Post Disaster Survey Guidelines Development Bali, 27-28 ' ' February 2012, 54pp.\n' ' * Allen, T. I., Wald, D. J., Earle, P. S., Marano, K. D., ' ' Hotovec, A. J., Lin, K., and Hearne, M., 2009. An Atlas ' ' of ShakeMaps and population exposure catalog for ' ' earthquake loss modeling, Bull. Earthq. Eng. 7, 701-718.\n' ' * Jaiswal, K., and Wald, D., 2010. An empirical model for ' ' global earthquake fatality estimation, Earthq. Spectra ' ' 26, 1017-1037.\n') limitation = tr( ' - The model is based on limited number of observed fatality ' ' rates during 4 past fatal events. \n' ' - The model clearly over-predicts the fatality rates at ' ' intensities higher than VIII.\n' ' - The model only estimates the expected fatality rate ' ' for a given intensity level; however the associated ' ' uncertainty for the proposed model is not addressed.\n' ' - There are few known mistakes in developing the current ' ' model:\n\n' ' * rounding MMI values to the nearest 0.5,\n' ' * Implementing Finite-Fault models of candidate events, and\n' ' * consistency between selected GMPEs with those in use by ' ' BMKG.\n') actions = tr( 'Provide details about the population will be die or displaced') detailed_description = tr( 'This model was developed by Institut Teknologi Bandung (ITB) ' 'and implemented by Dr. Hadi Ghasemi, Geoscience Australia\n' 'Algorithm:\n' 'In this study, the same functional form as Allen (2009) is ' 'adopted o express fatality rate as a function of intensity ' '(see Eq. 10 in the report). The Matlab built-in function ' '(fminsearch) for Nelder-Mead algorithm was used to estimate ' 'the model parameters. The objective function (L2G norm) that ' 'is minimized during the optimisation is the same as the one ' 'used by Jaiswal et al. (2010).\n' 'The coefficients used in the indonesian model are x=0.62275231, ' 'y=8.03314466, zeta=2.15') defaults = get_defaults() parameters = OrderedDict([ ('x', 0.62275231), ('y', 8.03314466), # Model coefficients # Rates of people displaced for each MMI level ('displacement_rate', { 1: 0, 2: 0, 3: 0, 4: 0, 5: 0, 6: 1.0, 7: 1.0, 8: 1.0, 9: 1.0, 10: 1.0 }), ('mmi_range', range(2, 10)), ('step', 0.5), # Threshold below which layer should be transparent ('tolerance', 0.01), ('calculate_displaced_people', True), ('postprocessors', OrderedDict([ ('Gender', { 'on': True }), ('Age', { 'on': True, 'params': OrderedDict([('youth_ratio', defaults['YOUTH_RATIO']), ('adult_ratio', defaults['ADULT_RATIO']), ('elder_ratio', defaults['ELDER_RATIO'])]) }), ('MinimumNeeds', { 'on': True }) ])), ('minimum needs', default_minimum_needs()) ]) def fatality_rate(self, mmi): """ ITB method to compute fatality rate :param mmi: """ # As per email discussion with Ole, Trevor, Hadi, mmi < 4 will have # a fatality rate of 0 - Tim if mmi < 4: return 0 x = self.parameters['x'] y = self.parameters['y'] return numpy.power(10.0, x * mmi - y) def run(self, layers): """Indonesian Earthquake Fatality Model Input: :param layers: List of layers expected to contain, my_hazard: Raster layer of MMI ground shaking my_exposure: Raster layer of population density """ displacement_rate = self.parameters['displacement_rate'] # Tolerance for transparency tolerance = self.parameters['tolerance'] # Extract input layers intensity = get_hazard_layer(layers) population = get_exposure_layer(layers) question = get_question(intensity.get_name(), population.get_name(), self) # Extract data grids my_hazard = intensity.get_data() # Ground Shaking my_exposure = population.get_data(scaling=True) # Population Density # Calculate population affected by each MMI level # FIXME (Ole): this range is 2-9. Should 10 be included? mmi_range = self.parameters['mmi_range'] number_of_exposed = {} number_of_displaced = {} number_of_fatalities = {} # Calculate fatality rates for observed Intensity values (my_hazard # based on ITB power model R = numpy.zeros(my_hazard.shape) for mmi in mmi_range: # Identify cells where MMI is in class i and # count population affected by this shake level I = numpy.where((my_hazard > mmi - self.parameters['step']) * (my_hazard <= mmi + self.parameters['step']), my_exposure, 0) # Calculate expected number of fatalities per level fatality_rate = self.fatality_rate(mmi) F = fatality_rate * I # Calculate expected number of displaced people per level try: D = displacement_rate[mmi] * I except KeyError, e: msg = 'mmi = %i, I = %s, Error msg: %s' % (mmi, str(I), str(e)) # noinspection PyExceptionInherit raise InaSAFEError(msg) # Adjust displaced people to disregard fatalities. # Set to zero if there are more fatalities than displaced. D = numpy.where(D > F, D - F, 0) # Sum up numbers for map R += D # Displaced # Generate text with result for this study # This is what is used in the real time system exposure table number_of_exposed[mmi] = numpy.nansum(I.flat) number_of_displaced[mmi] = numpy.nansum(D.flat) # noinspection PyUnresolvedReferences number_of_fatalities[mmi] = numpy.nansum(F.flat) # Set resulting layer to NaN when less than a threshold. This is to # achieve transparency (see issue #126). R[R < tolerance] = numpy.nan # Total statistics total = int(round(numpy.nansum(my_exposure.flat) / 1000) * 1000) # Compute number of fatalities fatalities = int( round(numpy.nansum(number_of_fatalities.values()) / 1000)) * 1000 # As per email discussion with Ole, Trevor, Hadi, total fatalities < 50 # will be rounded down to 0 - Tim if fatalities < 50: fatalities = 0 # Compute number of people displaced due to building collapse displaced = int( round(numpy.nansum(number_of_displaced.values()) / 1000)) * 1000 # Generate impact report table_body = [question] # Add total fatality estimate s = format_int(fatalities) table_body.append( TableRow([tr('Number of fatalities'), s], header=True)) if self.parameters['calculate_displaced_people']: # Add total estimate of people displaced s = format_int(displaced) table_body.append( TableRow([tr('Number of people displaced'), s], header=True)) else: displaced = 0 # Add estimate of total population in area s = format_int(int(total)) table_body.append( TableRow([tr('Total number of people'), s], header=True)) # Calculate estimated needs based on BNPB Perka 7/2008 minimum bantuan # FIXME: Refactor and share minimum_needs = self.parameters['minimum needs'] needs = evacuated_population_weekly_needs(displaced, minimum_needs) # Generate impact report for the pdf map table_body = [ question, TableRow([tr('Fatalities'), '%s' % format_int(fatalities)], header=True), TableRow([tr('People displaced'), '%s' % format_int(displaced)], header=True), TableRow( tr('Map shows density estimate of ' 'displaced population')), TableRow([tr('Needs per week'), tr('Total')], header=True), [tr('Rice [kg]'), format_int(needs['rice'])], [tr('Drinking Water [l]'), format_int(needs['drinking_water'])], [tr('Clean Water [l]'), format_int(needs['water'])], [tr('Family Kits'), format_int(needs['family_kits'])], TableRow(tr('Action Checklist:'), header=True) ] if fatalities > 0: table_body.append( tr('Are there enough victim identification ' 'units available for %s people?') % format_int(fatalities)) if displaced > 0: table_body.append( tr('Are there enough shelters and relief items ' 'available for %s people?') % format_int(displaced)) table_body.append( TableRow( tr('If yes, where are they located and ' 'how will we distribute them?'))) table_body.append( TableRow( tr('If no, where can we obtain ' 'additional relief items from and ' 'how will we transport them?'))) # Extend impact report for on-screen display table_body.extend([ TableRow(tr('Notes'), header=True), tr('Total population: %s') % format_int(total), tr('People are considered to be displaced if ' 'they experience and survive a shake level' 'of more than 5 on the MMI scale '), tr('Minimum needs are defined in BNPB ' 'regulation 7/2008'), tr('The fatality calculation assumes that ' 'no fatalities occur for shake levels below 4 ' 'and fatality counts of less than 50 are ' 'disregarded.'), tr('All values are rounded up to the nearest ' 'integer in order to avoid representing human ' 'lives as fractionals.') ]) table_body.append(TableRow(tr('Notes'), header=True)) table_body.append( tr('Fatality model is from ' 'Institute of Teknologi Bandung 2012.')) table_body.append(tr('Population numbers rounded to nearest 1000.')) # Result impact_summary = Table(table_body).toNewlineFreeString() impact_table = impact_summary # check for zero impact if numpy.nanmax(R) == 0 == numpy.nanmin(R): table_body = [ question, TableRow([tr('Fatalities'), '%s' % format_int(fatalities)], header=True) ] my_message = Table(table_body).toNewlineFreeString() raise ZeroImpactException(my_message) # Create style colours = ['#EEFFEE', '#FFFF7F', '#E15500', '#E4001B', '#730000'] classes = create_classes(R.flat[:], len(colours)) interval_classes = humanize_class(classes) style_classes = [] for i in xrange(len(colours)): style_class = dict() style_class['label'] = create_label(interval_classes[i]) style_class['quantity'] = classes[i] if i == 0: transparency = 100 else: transparency = 30 style_class['transparency'] = transparency style_class['colour'] = colours[i] style_classes.append(style_class) style_info = dict(target_field=None, style_classes=style_classes, style_type='rasterStyle') # For printing map purpose map_title = tr('Earthquake impact to population') legend_notes = tr('Thousand separator is represented by %s' % get_thousand_separator()) legend_units = tr('(people per cell)') legend_title = tr('Population density') # Create raster object and return L = Raster(R, projection=population.get_projection(), geotransform=population.get_geotransform(), keywords={ 'impact_summary': impact_summary, 'total_population': total, 'total_fatalities': fatalities, 'fatalites_per_mmi': number_of_fatalities, 'exposed_per_mmi': number_of_exposed, 'displaced_per_mmi': number_of_displaced, 'impact_table': impact_table, 'map_title': map_title, 'legend_notes': legend_notes, 'legend_units': legend_units, 'legend_title': legend_title }, name=tr('Estimated displaced population per cell'), style_info=style_info) return L
class PAGFatalityFunction(ITBFatalityFunction): """ Population Vulnerability Model Pager Loss ratio(MMI) = standard normal distrib( 1 / BETA * ln(MMI/THETA)). Reference: Jaiswal, K. S., Wald, D. J., and Hearne, M. (2009a). Estimating casualties for large worldwide earthquakes using an empirical approach. U.S. Geological Survey Open-File Report 2009-1136. :author Helen Crowley :rating 3 :param requires category=='hazard' and \ subcategory=='earthquake' and \ layertype=='raster' and \ unit=='MMI' :param requires category=='exposure' and \ subcategory=='population' and \ layertype=='raster' """ synopsis = tr('To asses the impact of earthquake on population based on ' 'Population Vulnerability Model Pager') citations = \ tr(' * Jaiswal, K. S., Wald, D. J., and Hearne, M. (2009a). ' ' Estimating casualties for large worldwide earthquakes using ' ' an empirical approach. U.S. Geological Survey Open-File ' ' Report 2009-1136.') limitation = '' detailed_description = '' title = tr('Die or be displaced according Pager model') defaults = get_defaults() # see https://github.com/AIFDR/inasafe/issues/628 default_needs = default_minimum_needs() default_needs[tr('Water')] = 67 parameters = OrderedDict([ ('Theta', 11.067), ('Beta', 0.106), # Model coefficients # Rates of people displaced for each MMI level ('displacement_rate', { 1: 0, 1.5: 0, 2: 0, 2.5: 0, 3: 0, 3.5: 0, 4: 0, 4.5: 0, 5: 0, 5.5: 0, 6: 1.0, 6.5: 1.0, 7: 1.0, 7.5: 1.0, 8: 1.0, 8.5: 1.0, 9: 1.0, 9.5: 1.0, 10: 1.0}), ('mmi_range', list(numpy.arange(2, 10, 0.5))), ('step', 0.25), # Threshold below which layer should be transparent ('tolerance', 0.01), ('calculate_displaced_people', True), ('postprocessors', OrderedDict([ ('Gender', {'on': True}), ('Age', { 'on': True, 'params': OrderedDict([ ('youth_ratio', defaults['YOUTH_RATIO']), ('adult_ratio', defaults['ADULT_RATIO']), ('elder_ratio', defaults['ELDER_RATIO'])])}), ('MinimumNeeds', {'on': True})])), ('minimum needs', default_needs)]) def fatality_rate(self, mmi): """Pager method to compute fatality rate""" N = math.sqrt(2 * math.pi) THETA = self.parameters['Theta'] BETA = self.parameters['Beta'] x = math.log(mmi / THETA) / BETA return math.exp(-x * x / 2.0) / N
class VolcanoPolygonHazardPopulation(FunctionProvider): """Impact function for volcano hazard zones impact on population :author AIFDR :rating 4 :param requires category=='hazard' and \ subcategory in ['volcano'] and \ layertype=='vector' :param requires category=='exposure' and \ subcategory=='population' and \ layertype=='raster' """ title = tr('Need evacuation') target_field = 'population' defaults = get_defaults() # Function documentation synopsis = tr('To assess the impacts of volcano eruption on population.') actions = tr( 'Provide details about how many population would likely be affected ' 'by each hazard zones.') hazard_input = tr( 'A hazard vector layer can be polygon or point. If polygon, it must ' 'have "KRB" attribute and the valuefor it are "Kawasan Rawan ' 'Bencana I", "Kawasan Rawan Bencana II", or "Kawasan Rawan Bencana ' 'III."If you want to see the name of the volcano in the result, you ' 'need to add "NAME" attribute for point data or "GUNUNG" attribute ' 'for polygon data.') exposure_input = tr( 'An exposure raster layer where each cell represent population count.') output = tr( 'Vector layer contains population affected and the minimum needs ' 'based on the population affected.') parameters = OrderedDict([ ('distance [km]', [3, 5, 10]), ('minimum needs', default_minimum_needs()), ('postprocessors', OrderedDict([ ('Gender', { 'on': True }), ('Age', { 'on': True, 'params': OrderedDict([('youth_ratio', defaults['YOUTH_RATIO']), ('adult_ratio', defaults['ADULT_RATIO']), ('elder_ratio', defaults['ELDER_RATIO'])]) }), ('MinimumNeeds', { 'on': True }) ])) ]) def run(self, layers): """Risk plugin for volcano population evacuation :param layers: List of layers expected to contain where two layers should be present. * my_hazard: Vector polygon layer of volcano impact zones * my_exposure: Raster layer of population data on the same grid as my_hazard Counts number of people exposed to volcano event. :returns: Map of population exposed to the volcano hazard zone. The returned dict will include a table with number of people evacuated and supplies required. :rtype: dict """ # Identify hazard and exposure layers my_hazard = get_hazard_layer(layers) # Volcano KRB my_exposure = get_exposure_layer(layers) question = get_question(my_hazard.get_name(), my_exposure.get_name(), self) # Input checks if not my_hazard.is_vector: msg = ('Input hazard %s was not a vector layer as expected ' % my_hazard.get_name()) raise Exception(msg) msg = ('Input hazard must be a polygon or point layer. I got %s with ' 'layer type %s' % (my_hazard.get_name(), my_hazard.get_geometry_name())) if not (my_hazard.is_polygon_data or my_hazard.is_point_data): raise Exception(msg) if my_hazard.is_point_data: # Use concentric circles radii = self.parameters['distance [km]'] centers = my_hazard.get_geometry() attributes = my_hazard.get_data() rad_m = [x * 1000 for x in radii] # Convert to meters my_hazard = make_circular_polygon(centers, rad_m, attributes=attributes) category_title = 'Radius' category_header = tr('Distance [km]') category_names = radii name_attribute = 'NAME' # As in e.g. the Smithsonian dataset else: # Use hazard map category_title = 'KRB' category_header = tr('Category') # FIXME (Ole): Change to English and use translation system category_names = [ 'Kawasan Rawan Bencana III', 'Kawasan Rawan Bencana II', 'Kawasan Rawan Bencana I' ] name_attribute = 'GUNUNG' # As in e.g. BNPB hazard map attributes = my_hazard.get_data() # Get names of volcanos considered if name_attribute in my_hazard.get_attribute_names(): D = {} for att in my_hazard.get_data(): # Run through all polygons and get unique names D[att[name_attribute]] = None volcano_names = '' for name in D: volcano_names += '%s, ' % name volcano_names = volcano_names[:-2] # Strip trailing ', ' else: volcano_names = tr('Not specified in data') if not category_title in my_hazard.get_attribute_names(): msg = ('Hazard data %s did not contain expected ' 'attribute %s ' % (my_hazard.get_name(), category_title)) # noinspection PyExceptionInherit raise InaSAFEError(msg) # Run interpolation function for polygon2raster P = assign_hazard_values_to_exposure_data(my_hazard, my_exposure, attribute_name='population') # Initialise attributes of output dataset with all attributes # from input polygon and a population count of zero new_attributes = my_hazard.get_data() categories = {} for attr in new_attributes: attr[self.target_field] = 0 cat = attr[category_title] categories[cat] = 0 # Count affected population per polygon and total evacuated = 0 for attr in P.get_data(): # Get population at this location pop = float(attr['population']) # Update population count for associated polygon poly_id = attr['polygon_id'] new_attributes[poly_id][self.target_field] += pop # Update population count for each category cat = new_attributes[poly_id][category_title] categories[cat] += pop # Count totals total = int(numpy.sum(my_exposure.get_data(nan=0))) # Don't show digits less than a 1000 total = round_thousand(total) # Count number and cumulative for each zone cum = 0 pops = {} cums = {} for name in category_names: if category_title == 'Radius': key = name * 1000 # Convert to meters else: key = name # prevent key error pop = int(categories.get(key, 0)) pop = round_thousand(pop) cum += pop cum = round_thousand(cum) pops[name] = pop cums[name] = cum # Use final accumulation as total number needing evac evacuated = cum tot_needs = evacuated_population_weekly_needs(evacuated) # Generate impact report for the pdf map blank_cell = '' table_body = [ question, TableRow( [tr('Volcanos considered'), '%s' % volcano_names, blank_cell], header=True), TableRow([ tr('People needing evacuation'), '%s' % format_int(evacuated), blank_cell ], header=True), TableRow( [category_header, tr('Total'), tr('Cumulative')], header=True) ] for name in category_names: table_body.append( TableRow( [name, format_int(pops[name]), format_int(cums[name])])) table_body.extend([ TableRow( tr('Map shows population affected in ' 'each of volcano hazard polygons.')), TableRow([tr('Needs per week'), tr('Total'), blank_cell], header=True), [tr('Rice [kg]'), format_int(tot_needs['rice']), blank_cell], [ tr('Drinking Water [l]'), format_int(tot_needs['drinking_water']), blank_cell ], [ tr('Clean Water [l]'), format_int(tot_needs['water']), blank_cell ], [ tr('Family Kits'), format_int(tot_needs['family_kits']), blank_cell ], [tr('Toilets'), format_int(tot_needs['toilets']), blank_cell] ]) impact_table = Table(table_body).toNewlineFreeString() # Extend impact report for on-screen display table_body.extend([ TableRow(tr('Notes'), header=True), tr('Total population %s in the exposure layer') % format_int(total), tr('People need evacuation if they are within the ' 'volcanic hazard zones.') ]) population_counts = [x[self.target_field] for x in new_attributes] impact_summary = Table(table_body).toNewlineFreeString() # check for zero impact if numpy.nanmax(population_counts) == 0 == numpy.nanmin( population_counts): table_body = [ question, TableRow([ tr('People needing evacuation'), '%s' % format_int(evacuated), blank_cell ], header=True) ] my_message = Table(table_body).toNewlineFreeString() raise ZeroImpactException(my_message) # Create style colours = [ '#FFFFFF', '#38A800', '#79C900', '#CEED00', '#FFCC00', '#FF6600', '#FF0000', '#7A0000' ] classes = create_classes(population_counts, len(colours)) interval_classes = humanize_class(classes) # Define style info for output polygons showing population counts style_classes = [] for i in xrange(len(colours)): style_class = dict() style_class['label'] = create_label(interval_classes[i]) if i == 0: transparency = 100 style_class['min'] = 0 else: transparency = 30 style_class['min'] = classes[i - 1] style_class['transparency'] = transparency style_class['colour'] = colours[i] style_class['max'] = classes[i] style_classes.append(style_class) # Override style info with new classes and name style_info = dict(target_field=self.target_field, style_classes=style_classes, style_type='graduatedSymbol') # For printing map purpose map_title = tr('People affected by volcanic hazard zone') legend_notes = tr('Thousand separator is represented by %s' % get_thousand_separator()) legend_units = tr('(people)') legend_title = tr('Population count') # Create vector layer and return V = Vector(data=new_attributes, projection=my_hazard.get_projection(), geometry=my_hazard.get_geometry(as_geometry_objects=True), name=tr('Population affected by volcanic hazard zone'), keywords={ 'impact_summary': impact_summary, 'impact_table': impact_table, 'target_field': self.target_field, 'map_title': map_title, 'legend_notes': legend_notes, 'legend_units': legend_units, 'legend_title': legend_title }, style_info=style_info) return V
class FloodEvacuationFunctionVectorHazard(FunctionProvider): """Impact function for vector flood evacuation :author AIFDR :rating 4 :param requires category=='hazard' and \ subcategory in ['flood', 'tsunami'] and \ layertype=='vector' :param requires category=='exposure' and \ subcategory=='population' and \ layertype=='raster' """ title = tr('Need evacuation') # Function documentation synopsis = tr( 'To assess the impacts of (flood or tsunami) inundation in vector ' 'format on population.') actions = tr( 'Provide details about how many people would likely need to be ' 'evacuated, where they are located and what resources would be ' 'required to support them.') detailed_description = tr( 'The population subject to inundation is determined whether in an ' 'area which affected or not. You can also set an evacuation ' 'percentage to calculate how many percent of the total population ' 'affected to be evacuated. This number will be used to estimate needs' ' based on BNPB Perka 7/2008 minimum bantuan.') hazard_input = tr( 'A hazard vector layer which has attribute affected the value is ' 'either 1 or 0') exposure_input = tr( 'An exposure raster layer where each cell represent population count.') output = tr( 'Vector layer contains population affected and the minimum needs ' 'based on evacuation percentage.') target_field = 'population' defaults = get_defaults() # Configurable parameters # TODO: Share the mimimum needs and make another default value parameters = OrderedDict([ ('evacuation_percentage', 1), # Percent of affected needing evacuation ('postprocessors', OrderedDict([ ('Gender', { 'on': True }), ('Age', { 'on': True, 'params': OrderedDict([('youth_ratio', defaults['YOUTH_RATIO']), ('adult_ratio', defaults['ADULT_RATIO']), ('elder_ratio', defaults['ELDER_RATIO'])]) }), ('MinimumNeeds', { 'on': True }), ])), ('minimum needs', default_minimum_needs()) ]) def run(self, layers): """Risk plugin for flood population evacuation Input: layers: List of layers expected to contain my_hazard : Vector polygon layer of flood depth my_exposure : Raster layer of population data on the same grid as my_hazard Counts number of people exposed to areas identified as flood prone Return Map of population exposed to flooding Table with number of people evacuated and supplies required """ # Identify hazard and exposure layers my_hazard = get_hazard_layer(layers) # Flood inundation my_exposure = get_exposure_layer(layers) question = get_question(my_hazard.get_name(), my_exposure.get_name(), self) # Check that hazard is polygon type if not my_hazard.is_vector: msg = ('Input hazard %s was not a vector layer as expected ' % my_hazard.get_name()) raise Exception(msg) msg = ('Input hazard must be a polygon layer. I got %s with layer ' 'type %s' % (my_hazard.get_name(), my_hazard.get_geometry_name())) if not my_hazard.is_polygon_data: raise Exception(msg) # Run interpolation function for polygon2raster P = assign_hazard_values_to_exposure_data(my_hazard, my_exposure, attribute_name='population') # Initialise attributes of output dataset with all attributes # from input polygon and a population count of zero new_attributes = my_hazard.get_data() category_title = 'affected' # FIXME: Should come from keywords deprecated_category_title = 'FLOODPRONE' categories = {} for attr in new_attributes: attr[self.target_field] = 0 try: cat = attr[category_title] except KeyError: cat = attr['FLOODPRONE'] categories[cat] = 0 # Count affected population per polygon, per category and total affected_population = 0 for attr in P.get_data(): affected = False if 'affected' in attr: res = attr['affected'] if res is None: x = False else: x = bool(res) affected = x elif 'FLOODPRONE' in attr: # If there isn't an 'affected' attribute, res = attr['FLOODPRONE'] if res is not None: affected = res.lower() == 'yes' elif 'Affected' in attr: # Check the default attribute assigned for points # covered by a polygon res = attr['Affected'] if res is None: x = False else: x = res affected = x else: # there is no flood related attribute msg = ('No flood related attribute found in %s. ' 'I was looking fore either "Flooded", "FLOODPRONE" ' 'or "Affected". The latter should have been ' 'automatically set by call to ' 'assign_hazard_values_to_exposure_data(). ' 'Sorry I can\'t help more.') raise Exception(msg) if affected: # Get population at this location pop = float(attr['population']) # Update population count for associated polygon poly_id = attr['polygon_id'] new_attributes[poly_id][self.target_field] += pop # Update population count for each category try: cat = new_attributes[poly_id][category_title] except KeyError: cat = new_attributes[poly_id][deprecated_category_title] categories[cat] += pop # Update total affected_population += pop affected_population = round_thousand(affected_population) # Estimate number of people in need of evacuation evacuated = (affected_population * self.parameters['evacuation_percentage'] / 100.0) total = int(numpy.sum(my_exposure.get_data(nan=0, scaling=False))) # Don't show digits less than a 1000 total = round_thousand(total) evacuated = round_thousand(evacuated) # Calculate estimated minimum needs minimum_needs = self.parameters['minimum needs'] tot_needs = evacuated_population_weekly_needs(evacuated, minimum_needs) # Generate impact report for the pdf map table_body = [ question, TableRow([ tr('People affected'), '%s%s' % (format_int(int(affected_population)), ('*' if affected_population >= 1000 else '')) ], header=True), TableRow([ tr('People needing evacuation'), '%s%s' % (format_int(int(evacuated)), ('*' if evacuated >= 1000 else '')) ], header=True), TableRow([ TableCell(tr('* Number is rounded to the nearest 1000'), col_span=2) ], header=False), TableRow([ tr('Evacuation threshold'), '%s%%' % format_int(self.parameters['evacuation_percentage']) ], header=True), TableRow( tr('Map shows population affected in each flood' ' prone area')), TableRow( tr('Table below shows the weekly minium needs ' 'for all evacuated people')), TableRow([tr('Needs per week'), tr('Total')], header=True), [tr('Rice [kg]'), format_int(tot_needs['rice'])], [ tr('Drinking Water [l]'), format_int(tot_needs['drinking_water']) ], [tr('Clean Water [l]'), format_int(tot_needs['water'])], [tr('Family Kits'), format_int(tot_needs['family_kits'])], [tr('Toilets'), format_int(tot_needs['toilets'])] ] impact_table = Table(table_body).toNewlineFreeString() table_body.append(TableRow(tr('Action Checklist:'), header=True)) table_body.append(TableRow(tr('How will warnings be disseminated?'))) table_body.append(TableRow(tr('How will we reach stranded people?'))) table_body.append(TableRow(tr('Do we have enough relief items?'))) table_body.append( TableRow( tr('If yes, where are they located and how ' 'will we distribute them?'))) table_body.append( TableRow( tr('If no, where can we obtain additional ' 'relief items from and how will we ' 'transport them to here?'))) # Extend impact report for on-screen display table_body.extend([ TableRow(tr('Notes'), header=True), tr('Total population: %s') % format_int(total), tr('People need evacuation if in area identified ' 'as "Flood Prone"'), tr('Minimum needs are defined in BNPB ' 'regulation 7/2008') ]) impact_summary = Table(table_body).toNewlineFreeString() # Create style # Define classes for legend for flooded population counts colours = [ '#FFFFFF', '#38A800', '#79C900', '#CEED00', '#FFCC00', '#FF6600', '#FF0000', '#7A0000' ] population_counts = [x['population'] for x in new_attributes] classes = create_classes(population_counts, len(colours)) interval_classes = humanize_class(classes) # Define style info for output polygons showing population counts style_classes = [] for i in xrange(len(colours)): style_class = dict() style_class['label'] = create_label(interval_classes[i]) if i == 0: transparency = 100 style_class['min'] = 0 else: transparency = 0 style_class['min'] = classes[i - 1] style_class['transparency'] = transparency style_class['colour'] = colours[i] style_class['max'] = classes[i] style_classes.append(style_class) # Override style info with new classes and name style_info = dict(target_field=self.target_field, style_classes=style_classes, style_type='graduatedSymbol') # For printing map purpose map_title = tr('People affected by flood prone areas') legend_notes = tr('Thousand separator is represented by \'.\'') legend_units = tr('(people per polygon)') legend_title = tr('Population Count') # Create vector layer and return V = Vector(data=new_attributes, projection=my_hazard.get_projection(), geometry=my_hazard.get_geometry(), name=tr('Population affected by flood prone areas'), keywords={ 'impact_summary': impact_summary, 'impact_table': impact_table, 'target_field': self.target_field, 'map_title': map_title, 'legend_notes': legend_notes, 'legend_units': legend_units, 'legend_title': legend_title }, style_info=style_info) return V
class FloodEvacuationFunction(FunctionProvider): """Impact function for flood evacuation :author AIFDR :rating 4 :param requires category=='hazard' and \ subcategory in ['flood', 'tsunami'] and \ layertype=='raster' and \ unit=='m' :param requires category=='exposure' and \ subcategory=='population' and \ layertype=='raster' """ title = tr('Need evacuation') defaults = get_defaults() # Function documentation synopsis = tr( 'To assess the impacts of (flood or tsunami) inundation in raster ' 'format on population.') actions = tr( 'Provide details about how many people would likely need to be ' 'evacuated, where they are located and what resources would be ' 'required to support them.') detailed_description = tr( 'The population subject to inundation exceeding a threshold ' '(default 1m) is calculated and returned as a raster layer. In ' 'addition the total number and the required needs in terms of the ' 'BNPB (Perka 7) are reported. The threshold can be changed and even ' 'contain multiple numbers in which case evacuation and needs are ' 'calculated using the largest number with population breakdowns ' 'provided for the smaller numbers. The population raster is resampled ' 'to the resolution of the hazard raster and is rescaled so that the ' 'resampled population counts reflect estimates of population count ' 'per resampled cell. The resulting impact layer has the same ' 'resolution and reflects population count per cell which are affected ' 'by inundation.') hazard_input = tr( 'A hazard raster layer where each cell represents flood depth ' '(in meters).') exposure_input = tr( 'An exposure raster layer where each cell represent population count.') output = tr( 'Raster layer contains population affected and the minimum needs ' 'based on the population affected.') limitation = tr( 'The default threshold of 1 meter was selected based on consensus, ' 'not hard evidence.') # Configurable parameters # TODO: Share the mimimum needs and make another default value parameters = OrderedDict([ ('thresholds [m]', [1.0]), ('postprocessors', OrderedDict([ ('Gender', { 'on': True }), ('Age', { 'on': True, 'params': OrderedDict([('youth_ratio', defaults['YOUTH_RATIO']), ('adult_ratio', defaults['ADULT_RATIO']), ('elder_ratio', defaults['ELDER_RATIO'])]) }) ])), ('minimum needs', default_minimum_needs()) ]) def run(self, layers): """Risk plugin for flood population evacuation Input layers: List of layers expected to contain my_hazard: Raster layer of flood depth my_exposure: Raster layer of population data on the same grid as my_hazard Counts number of people exposed to flood levels exceeding specified threshold. Return Map of population exposed to flood levels exceeding the threshold Table with number of people evacuated and supplies required """ # Identify hazard and exposure layers my_hazard = get_hazard_layer(layers) # Flood inundation [m] my_exposure = get_exposure_layer(layers) question = get_question(my_hazard.get_name(), my_exposure.get_name(), self) # Determine depths above which people are regarded affected [m] # Use thresholds from inundation layer if specified thresholds = self.parameters['thresholds [m]'] verify(isinstance(thresholds, list), 'Expected thresholds to be a list. Got %s' % str(thresholds)) # Extract data as numeric arrays D = my_hazard.get_data(nan=0.0) # Depth # Calculate impact as population exposed to depths > max threshold P = my_exposure.get_data(nan=0.0, scaling=True) # Calculate impact to intermediate thresholds counts = [] # merely initialize my_impact = None for i, lo in enumerate(thresholds): if i == len(thresholds) - 1: # The last threshold my_impact = M = numpy.where(D >= lo, P, 0) else: # Intermediate thresholds hi = thresholds[i + 1] M = numpy.where((D >= lo) * (D < hi), P, 0) # Count val = int(numpy.sum(M)) # Don't show digits less than a 1000 val = round_thousand(val) counts.append(val) # Count totals evacuated = counts[-1] total = int(numpy.sum(P)) # Don't show digits less than a 1000 total = round_thousand(total) # Calculate estimated minimum needs # The default value of each logistic is based on BNPB Perka 7/2008 # minimum bantuan minimum_needs = self.parameters['minimum needs'] tot_needs = evacuated_population_weekly_needs(evacuated, minimum_needs) # Generate impact report for the pdf map # noinspection PyListCreation table_body = [ question, TableRow([(tr('People in %.1f m of water') % thresholds[-1]), '%s%s' % (format_int(evacuated), ('*' if evacuated >= 1000 else ''))], header=True), TableRow(tr('* Number is rounded to the nearest 1000'), header=False), TableRow(tr('Map shows population density needing evacuation')), TableRow( tr('Table below shows the weekly minium needs for all ' 'evacuated people')), TableRow([tr('Needs per week'), tr('Total')], header=True), [tr('Rice [kg]'), format_int(tot_needs['rice'])], [ tr('Drinking Water [l]'), format_int(tot_needs['drinking_water']) ], [tr('Clean Water [l]'), format_int(tot_needs['water'])], [tr('Family Kits'), format_int(tot_needs['family_kits'])], [tr('Toilets'), format_int(tot_needs['toilets'])] ] table_body.append(TableRow(tr('Action Checklist:'), header=True)) table_body.append(TableRow(tr('How will warnings be disseminated?'))) table_body.append(TableRow(tr('How will we reach stranded people?'))) table_body.append(TableRow(tr('Do we have enough relief items?'))) table_body.append( TableRow( tr('If yes, where are they located and how ' 'will we distribute them?'))) table_body.append( TableRow( tr('If no, where can we obtain additional relief items from and how ' 'will we transport them to here?'))) # Extend impact report for on-screen display table_body.extend([ TableRow(tr('Notes'), header=True), tr('Total population: %s') % format_int(total), tr('People need evacuation if flood levels exceed %(eps).1f m') % { 'eps': thresholds[-1] }, tr('Minimum needs are defined in BNPB regulation 7/2008'), tr('All values are rounded up to the nearest integer in order to ' 'avoid representing human lives as fractionals.') ]) if len(counts) > 1: table_body.append(TableRow(tr('Detailed breakdown'), header=True)) for i, val in enumerate(counts[:-1]): s = (tr('People in %(lo).1f m to %(hi).1f m of water: %(val)i') % { 'lo': thresholds[i], 'hi': thresholds[i + 1], 'val': format_int(val) }) table_body.append(TableRow(s, header=False)) # Result impact_summary = Table(table_body).toNewlineFreeString() impact_table = impact_summary # check for zero impact if numpy.nanmax(my_impact) == 0 == numpy.nanmin(my_impact): table_body = [ question, TableRow([(tr('People in %.1f m of water') % thresholds[-1]), '%s' % format_int(evacuated)], header=True) ] my_message = Table(table_body).toNewlineFreeString() raise ZeroImpactException(my_message) # Create style colours = [ '#FFFFFF', '#38A800', '#79C900', '#CEED00', '#FFCC00', '#FF6600', '#FF0000', '#7A0000' ] classes = create_classes(my_impact.flat[:], len(colours)) interval_classes = humanize_class(classes) style_classes = [] for i in xrange(len(colours)): style_class = dict() if i == 1: label = create_label(interval_classes[i], 'Low') elif i == 4: label = create_label(interval_classes[i], 'Medium') elif i == 7: label = create_label(interval_classes[i], 'High') else: label = create_label(interval_classes[i]) style_class['label'] = label style_class['quantity'] = classes[i] if i == 0: transparency = 100 else: transparency = 0 style_class['transparency'] = transparency style_class['colour'] = colours[i] style_classes.append(style_class) style_info = dict(target_field=None, style_classes=style_classes, style_type='rasterStyle') # For printing map purpose map_title = tr('People in need of evacuation') legend_notes = tr('Thousand separator is represented by %s' % get_thousand_separator()) legend_units = tr('(people per cell)') legend_title = tr('Population density') # Create raster object and return R = Raster(my_impact, projection=my_hazard.get_projection(), geotransform=my_hazard.get_geotransform(), name=tr('Population which %s') % get_function_title(self), keywords={ 'impact_summary': impact_summary, 'impact_table': impact_table, 'map_title': map_title, 'legend_notes': legend_notes, 'legend_units': legend_units, 'legend_title': legend_title }, style_info=style_info) return R