Open Data Schema for Energy
Docs Blog GitHub Contact

Cross-City Benchmarking

Compare energy and emissions profiles across cities using consistent methodology and the same underlying dataset.

When to Use

You need apples-to-apples comparisons between geographies — Atlanta vs Houston vs Charlotte — using the same simulation methodology, weather models, and building stock characterization. This eliminates the problem of comparing numbers collected with different standards, different reporting years, and different definitions.

Pipeline

Download ResStock/ComStock baseline for each target state
    → filter to target counties
    → weight and aggregate per county
    → normalize (per dwelling, per sqft, or per capita)
    → side-by-side comparison

Prerequisites

Download ResStock baseline data for each state you want to compare, per the Accessing the Data guide.

Geographic Identifiers

ResStock uses NHGIS GISJOIN codes for counties. Here are the codes for common metro areas:

City County State GISJOIN Code
Atlanta Fulton County GA G1312100
Houston Harris County TX G4820100
Charlotte Mecklenburg County NC G3711900
Miami Miami-Dade County FL G1208600
Nashville Davidson County TN G4703700
New Orleans Orleans Parish LA G2207100
Birmingham Jefferson County AL G0107300
Jacksonville Duval County FL G1203100

To find other GISJOIN codes: the format is G + 2-digit state FIPS (zero-padded) + 5-digit county FIPS (zero-padded with trailing zero). Example: Fulton County, GA = FIPS 13121 → G1312100.

Complete Code

Step 1: Load data for multiple states

import pandas as pd

# Download these first — see "Accessing the Data" guide
states = {
    "GA": "georgia_resstock_baseline.csv",
    "TX": "texas_resstock_baseline.csv",
    "NC": "north_carolina_resstock_baseline.csv",
}

dfs = {}
for state, filepath in states.items():
    dfs[state] = pd.read_csv(filepath, low_memory=False)
    print(f"{state}: {len(dfs[state]):,} buildings loaded")

# Combine into one dataframe
df = pd.concat(dfs.values(), ignore_index=True)

Step 2: Filter to target counties

cities = {
    "Atlanta":   "G1312100",  # Fulton County, GA
    "Houston":   "G4820100",  # Harris County, TX
    "Charlotte": "G3711900",  # Mecklenburg County, NC
}

city_data = {}
for city, gisjoin in cities.items():
    city_df = df[df["in.county"] == gisjoin].copy()
    city_data[city] = city_df
    print(f"{city}: {len(city_df):,} building models")

Step 3: Compare total emissions

emissions_col = "out.emissions.co2e.lrmer_mid_case_15.all_fuels.total.lb"
energy_col = "out.site_energy.total.energy_consumption.kwh"

print(f"\n{'City':<12} {'Total CO₂ (tons)':>18} {'Total Energy (GWh)':>20} {'Dwellings':>12}")
print("-" * 64)

for city, cdf in city_data.items():
    total_co2_tons = (cdf[emissions_col] * cdf["weight"]).sum() / 2_204.6
    total_energy_gwh = (cdf[energy_col] * cdf["weight"]).sum() / 1_000_000
    total_dwellings = cdf["weight"].sum()

    print(f"{city:<12} {total_co2_tons:>18,.0f} {total_energy_gwh:>20,.1f} {total_dwellings:>12,.0f}")

Step 4: Normalize for fair comparison

Raw totals are misleading because cities have different population sizes. Normalize per dwelling:

print(f"\n{'City':<12} {'CO₂/dwelling (tons)':>20} {'kWh/dwelling':>14} {'Dwellings':>12}")
print("-" * 60)

for city, cdf in city_data.items():
    total_co2 = (cdf[emissions_col] * cdf["weight"]).sum() / 2_204.6
    total_energy = (cdf[energy_col] * cdf["weight"]).sum()
    total_dwellings = cdf["weight"].sum()

    print(f"{city:<12} {total_co2/total_dwellings:>20,.2f} "
          f"{total_energy/total_dwellings:>14,.0f} {total_dwellings:>12,.0f}")

Step 5: Compare building stock composition

Understanding why cities differ is often more useful than the raw numbers:

print("\nBuilding stock composition (% of dwellings by type):")
print(f"{'Type':<30}", end="")
for city in cities:
    print(f" {city:>12}", end="")
print()
print("-" * (30 + 13 * len(cities)))

all_types = sorted(df["in.geometry_building_type_recs"].unique())
for btype in all_types:
    print(f"{btype:<30}", end="")
    for city, cdf in city_data.items():
        type_weight = cdf[cdf["in.geometry_building_type_recs"] == btype]["weight"].sum()
        total_weight = cdf["weight"].sum()
        pct = type_weight / total_weight if total_weight > 0 else 0
        print(f" {pct:>11.1%}", end="")
    print()

Step 6: Compare vintage distribution

print("\nBuilding stock by construction era (% of dwellings):")
print(f"{'Vintage':<20}", end="")
for city in cities:
    print(f" {city:>12}", end="")
print()
print("-" * (20 + 13 * len(cities)))

for vintage in sorted(df["in.vintage"].unique()):
    print(f"{vintage:<20}", end="")
    for city, cdf in city_data.items():
        v_weight = cdf[cdf["in.vintage"] == vintage]["weight"].sum()
        total_weight = cdf["weight"].sum()
        pct = v_weight / total_weight if total_weight > 0 else 0
        print(f" {pct:>11.1%}", end="")
    print()

This reveals actionable differences — for example, if Atlanta has significantly more pre-1960 housing stock than Charlotte, envelope retrofits (insulation, air sealing) will have a proportionally larger impact there.

Step 7: Compare upgrade potential across cities

Combine cross-city data with upgrade scenarios to rank where interventions matter most:

# Load heat pump upgrade for each state (upgrade 04)
# (Download these the same way as baselines — see Accessing the Data guide)
upgrade_files = {
    "GA": "georgia_resstock_upgrade04.csv",
    "TX": "texas_resstock_upgrade04.csv",
    "NC": "north_carolina_resstock_upgrade04.csv",
}

print(f"\nHeat pump retrofit impact by city:")
print(f"{'City':<12} {'Baseline CO₂':>14} {'Post-Upgrade':>14} {'Savings':>14} {'Reduction':>10}")
print("-" * 66)

for city, gisjoin in cities.items():
    state = gisjoin[1:3]  # Extract state FIPS from GISJOIN
    state_abbr = {"13": "GA", "48": "TX", "37": "NC"}[state]

    baseline_city = city_data[city]
    upgrade_df = pd.read_csv(upgrade_files[state_abbr], low_memory=False)
    upgrade_city = upgrade_df[upgrade_df["in.county"] == gisjoin]

    merged = baseline_city[["bldg_id", emissions_col, "weight"]].merge(
        upgrade_city[["bldg_id", emissions_col]], on="bldg_id", suffixes=("_b", "_u"))

    base_tons = (merged[f"{emissions_col}_b"] * merged["weight"]).sum() / 2_204.6
    upgrade_tons = (merged[f"{emissions_col}_u"] * merged["weight"]).sum() / 2_204.6
    savings_tons = base_tons - upgrade_tons
    pct = savings_tons / base_tons if base_tons > 0 else 0

    print(f"{city:<12} {base_tons:>14,.0f} {upgrade_tons:>14,.0f} "
          f"{savings_tons:>14,.0f} {pct:>9.1%}")