data_miner.rb in earth-0.11.1

- old
+ new
@@ -1,34 +1,48 @@
 EgridSubregion.class_eval do
   data_miner do
-    # FIXME TODO for some reason this doesn't work...
-    # import "eGRID regions and electricity emission factors derived from eGRID 2007 data",
-    #        :url => 'http://www.epa.gov/cleanenergy/documents/egridzips/eGRID2007_Version1-1.zip',
-    #        :filename => 'eGRID2007_Version1-1/eGRID2007V1_1_year05_aggregation.xls',
-    #        :sheet => 'SRL05',
-    #        :skip => 3,
-    #        :select => lambda { |row| row['eGRID2007 2005 file eGRID subregion location (operator)-based sequence number'].to_i.between?(1, 26) } do
-    #   key   'abbreviation', :field_name => 'eGRID subregion acronym'
-    #   store 'name', :field_name => 'eGRID subregion name associated with eGRID subregion acronym'
-    #   store 'nerc_abbreviation', :field_name => 'NERC region acronym associated with the eGRID subregion acronym'
-    #   store 'net_generation', :field_name => 'eGRID subregion annual net generation (MWh)', :units => 'megawatt_hours'
-    #   store 'electricity_co2_emission_factor', :field_name => 'eGRID subregion annual CO2 output emission rate (lb/MWh)', :units => 'pounds_per_megawatt_hour'
-    #   store 'electricity_co2_biogenic_emission_factor', :static => '0.0', :units => 'kilograms_per_kilowatt_hour'
-    #   store 'electricity_ch4_emission_factor', :field_name => 'eGRID subregion annual CH4 output emission rate (lb/GWh)', :units => 'pounds_per_gigawatt_hour'
-    #   store 'electricity_n2o_emission_factor', :field_name => 'eGRID subregion annual N2O output emission rate (lb/GWh)', :units => 'pounds_per_gigawatt_hour'
-    # end
+    import "eGRID 2010 subregions and electricity emission factors",
+           :url => 'http://www.epa.gov/cleanenergy/documents/egridzips/eGRID2010_Version1-1_xls_only.zip',
+           :filename => 'eGRID2010V1_1_year07_AGGREGATION.xls',
+           :sheet => 'SRL07',
+           :skip => 4,
+           :select => lambda { |row| row['SEQSRL07'].to_i.between?(1, 26) } do
+      key   'abbreviation', :field_name => 'SUBRGN'
+      store 'name', :field_name => 'SRNAME'
+      store 'nerc_abbreviation',               :field_name => 'NERC'
+      store 'net_generation',                  :field_name => 'SRNGENAN', :units => :megawatt_hours
+      store 'electricity_co2_emission_factor', :field_name => 'SRCO2RTA', :units => :pounds_per_megawatt_hour
+      store 'electricity_co2_biogenic_emission_factor', :static => '0.0', :units => :kilograms_per_kilowatt_hour
+      store 'electricity_ch4_emission_factor', :field_name => 'SRCH4RTA', :units => :pounds_per_gigawatt_hour
+      store 'electricity_n2o_emission_factor', :field_name => 'SRN2ORTA', :units => :pounds_per_gigawatt_hour
+    end
     
-    import "eGRID subregion data",
-           :url => 'https://spreadsheets.google.com/pub?key=0AoQJbWqPrREqdGRORTJNSWRMQ1puRVprYlAtZHhDaFE&hl=en&gid=0&output=csv' do
+    # FIXME TODO once 'US' subregion is no longer needed remove it from this source file
+    import "eGRID subregion to region associations",
+           :url => 'https://spreadsheets.google.com/pub?key=0AoQJbWqPrREqdGRORTJNSWRMQ1puRVprYlAtZHhDaFE&output=csv' do
       key 'abbreviation'
-      store 'name'
-      store 'nerc_abbreviation'
       store 'egrid_region_name'
-      store 'net_generation', :units => 'megawatt_hours'
-      store 'electricity_co2_emission_factor', :field_name => 'electricty_ef_co2', :units => 'pounds_per_megawatt_hour'
-      store 'electricity_co2_biogenic_emission_factor', :static => '0.0', :units => 'kilograms_per_kilowatt_hour'
-      store 'electricity_ch4_emission_factor', :field_name => 'electricity_ef_ch4',  :units => 'pounds_per_gigawatt_hour'
-      store 'electricity_n2o_emission_factor', :field_name => 'electricity_ef_n2o',  :units => 'pounds_per_gigawatt_hour'
+    end
+    
+    # DEPRECATED but don't remove until confirmed that all emitters use EgridSubregion.fallback rather than EgridSubregion.find_by_abbreviation 'US'
+    process "Calculate national averages" do
+      us_average = find_by_abbreviation 'US'
+      subregions = where("abbreviation != 'US'")
+      
+      us_average.name = 'United States'
+      us_average.nerc_abbreviation = 'US'
+      us_average.egrid_region_name = 'US'
+      us_average.net_generation                           = subregions.sum(:net_generation)
+      us_average.electricity_co2_emission_factor          = subregions.weighted_average(:electricity_co2_emission_factor, :weighted_by => :net_generation)
+      us_average.electricity_co2_biogenic_emission_factor = subregions.weighted_average(:electricity_co2_biogenic_emission_factor, :weighted_by => :net_generation)
+      us_average.electricity_ch4_emission_factor          = subregions.weighted_average(:electricity_ch4_emission_factor, :weighted_by => :net_generation)
+      us_average.electricity_n2o_emission_factor          = subregions.weighted_average(:electricity_n2o_emission_factor, :weighted_by => :net_generation)
+      us_average.net_generation_units                           = 'megawatt_hours'
+      us_average.electricity_co2_emission_factor_units          = 'pounds_per_megawatt_hour'
+      us_average.electricity_co2_biogenic_emission_factor_units = 'kilograms_per_kilowatt_hour'
+      us_average.electricity_ch4_emission_factor_units          = 'pounds_per_gigawatt_hour'
+      us_average.electricity_n2o_emission_factor_units          = 'pounds_per_gigawatt_hour'
+      us_average.save!
     end
     
     process "Convert co2 emission factor to metric units" do
       conversion_factor = 1.pounds.to(:kilograms) / 1_000.0 # kg / lbs * MWh / kWh
       where(:electricity_co2_emission_factor_units => 'pounds_per_megawatt_hour').update_all(%{