diff --git a/README.md b/README.md
index d6ca677..342fd59 100644
--- a/README.md
+++ b/README.md
@@ -40,12 +40,23 @@ Compared to the US average, data center host communities are:
 - **Better connected**: 94.9% broadband (vs. 89%)
 
 ### Infrastructure Insights
-- **89% of data centers are in metropolitan tracts** (vs. 80% of all US tracts)
+- **89% of data centers are in metropolitan tracts** (vs. 80% of all US tracts) - only 1.11× over-index
 - **Non-metro data centers (11%)** are dominated by hyperscalers:
   - AWS (67), Meta (22), Microsoft (10), Google (4) = 55% of non-metro facilities
   - 66% are in Oregon + Washington (Columbia River hydro corridor)
-- **Energy infrastructure**: 4 states have >2/3 of generation within 50 km of a data center:
+- **Grid saturation**: 4 states have >2/3 of generation within 50 km of a data center:
   - New Jersey: 83%, Nevada: 75%, Tennessee: 70%, Oregon: 68%
+- **Hyperscaler energy strategies** (non-metro sites):
+  - AWS: 114 GW wind + 66 GW hydro
+  - Microsoft: 13 GW nuclear (Palo Verde co-location)
+  - Meta: 16 GW solar
+
+### Clustered vs. Isolated Facilities
+Facilities in DBSCAN clusters differ significantly from isolated sites:
+- **$35K income gap**: Clustered sites in tracts with median income $108K vs. $73K for isolated
+- **+18 pp education**: 51% bachelor's+ vs. 33%
+- **More diverse**: 25 pp less non-Hispanic white
+- **2× energy infrastructure**: 89 vs. 40 generators within 50 km
 
 ### Submarine Cables
 - **Data centers are NOT systematically closer to cables** than ordinary US cities
@@ -194,10 +205,18 @@ python3 make_internet_cables_map.py
 ## Data Quality Notes
 
 1. **Incomplete power ratings**: Only 5.9% of data centers have power ratings (108/1,833)
-2. **Operator fragmentation**: String variations ("Meta" vs. "Meta, Inc.") inflate distinct-operator counts
+2. **Operator fragmentation**: String variations ("Meta" vs. "Meta, Inc.", AWS variants) inflate distinct-operator counts
 3. **45 facilities** use city-precision fallback coordinates (approximate tract assignment)
 4. **7 facilities** failed RUCA join (Puerto Rico / non-US)
 5. **Broadband subscribers** are a coarse benefit proxy (actual cloud users are global)
+6. **EIA longitude correction**: 2008-2010 generator coordinates had sign errors, corrected in flat-table build
+
+## Known Limitations
+
+- **Power capacity**: Only 5.9% populated - nearby EIA generator capacity used as proxy
+- **Operator strings**: Need deduplication (50 of 190 non-metro facilities have null operator)
+- **Benefit measurement**: Broadband subscribers are an imperfect proxy for cloud computing benefits
+- **Universe**: Limited to 46 DC-host states (excludes DC-free states from ACS comparison)
 
 ## Research Ideas & Future Work
 
diff --git a/database-tables.md b/database-tables.md
index 116e226..89987f2 100644
--- a/database-tables.md
+++ b/database-tables.md
@@ -412,6 +412,93 @@ Tables are organized into four categories:
 
 ---
 
+### Other Tables
+
+#### `opposition_cases_geocoded`
+**Rows**: 18  
+**Purpose**: Geocoded community-opposition cases against data center builds
+
+**Key Columns**:
+- `case_id` (TEXT) - Unique identifier
+- `developer` (TEXT) - Proposed developer/operator
+- `investment_billions` (DOUBLE PRECISION) - Investment amount in billions
+- `outcome` (TEXT) - Case outcome (approved, rejected, pending)
+- `governance_response` (TEXT) - Government response
+- `latitude`, `longitude`, `geom`
+
+**Source**: Compiled from news archives
+
+**Notes**: Loaded but currently unused - see research-ideas.md for proposed analyses
+
+#### `census_tract_huc8_link`
+**Rows**: 806  
+**Purpose**: Tract↔HUC8 spatial overlap table
+
+**Key Columns**:
+- `geoid` (TEXT) - Census tract GEOID
+- `huc8` (TEXT) - HUC8 watershed code
+- `overlap_pct` (DOUBLE PRECISION) - Percentage of tract overlapping watershed
+
+**Notes**: Useful for downstream tract-level water-stress joins; limited to tracts containing data centers
+
+#### `im3_state_projected_moderate_50`
+**Rows**: 328  
+**Purpose**: PNNL IM3 projected data center siting (moderate growth, gravity weight 0.50)
+
+**Key Columns**:
+- `facility_id` (TEXT)
+- `state` (TEXT)
+- `cost_millions` (DOUBLE PRECISION)
+- `it_mw` (DOUBLE PRECISION) - IT load in megawatts
+- `cooling_water_demand_gal_per_day` (DOUBLE PRECISION)
+- `latitude`, `longitude`, `geom`
+
+**Source**: PNNL Integrated Multisector Multiscale Modeling (IM3)
+
+**Notes**: Loaded but unused - potential for forward-projection analysis
+
+#### `im3_projected_state_demand_summary`
+**Rows**: 31  
+**Purpose**: State-level rollup of IM3 projected facility counts, IT MW, and cooling demand
+
+**Key Columns**:
+- `state` (TEXT)
+- `facility_count` (INTEGER)
+- `total_it_mw` (DOUBLE PRECISION)
+- `total_cooling_demand_mgd` (DOUBLE PRECISION) - Million gallons per day
+
+**Source**: IM3 model outputs
+
+#### `utility_rate_tracker_2025_2028`
+**Rows**: 374  
+**Purpose**: Utility rate-increase tracker by provider × state × service type
+
+**Key Columns**:
+- `provider` (TEXT) - Utility provider name
+- `state` (TEXT)
+- `service_type` (TEXT)
+- `effective_date` (DATE)
+- `monthly_increase_dollars` (DOUBLE PRECISION)
+- `percent_increase` (DOUBLE PRECISION)
+
+**Source**: Utility rate tracker database
+
+**Notes**: Loaded but unused in demographic/energy analysis
+
+#### `energy_atlas_layers_catalog`
+**Rows**: ~5  
+**Purpose**: Metadata catalog of EIA layers ingested
+
+**Key Columns**:
+- `table_name` (TEXT)
+- `source_url` (TEXT)
+- `import_timestamp` (TIMESTAMP)
+- `row_count` (INTEGER)
+
+**Notes**: Created by `ingest_eia_energy_layers.py`
+
+---
+
 ## Commonly Used Joins
 
 ### Data Center to Demographics
diff --git a/research-ideas.md b/research-ideas.md
index 0a82127..8e368aa 100644
--- a/research-ideas.md
+++ b/research-ideas.md
@@ -61,6 +61,8 @@ canonical_map = {
 ### 3. Water Stress Overlay
 **Status**: 257 HUC8 watersheds contain data centers; 15 watersheds hold 50% of facilities
 
+**Priority**: HIGH - Critical for environmental impact analysis
+
 **Approach**:
 - Join to USGS WaterWatch streamflow data
 - Add USGS Drought Watch indicators by HUC8
@@ -69,10 +71,18 @@ canonical_map = {
   - Surface water withdrawal permits
   - Drought frequency/severity (USDM historical data)
 
+**Key Watersheds for Focus**:
+- **Middle Potomac-Catoctin** (HUC8 02070008): 235 DCs (12.8% of US total) - Loudoun/Ashburn
+- **Middle Potomac-Anacostia-Occoquan** (02070010): 111 DCs - Fairfax/inner Loudoun
+- **Coyote** (18050003): 88 DCs - Silicon Valley
+- **Upper Scioto** (05060001): 73 DCs - Columbus OH
+- **Umatilla** (17070103): 29 DCs - AWS-exclusive watershed
+
 **Research Questions**:
 - Are data centers sited in water-stressed watersheds?
 - Do high-density clusters (Loudoun County, Columbus OH) face water constraints?
 - Compare water stress in hyperscaler non-metro sites (Columbia River corridor) vs. metro clusters
+- Does single-operator watershed capture (Umatilla = AWS only) correlate with water availability?
 
 **Tables to Create**:
 - `watershed_water_stress` - HUC8-level water stress indicators
@@ -83,27 +93,38 @@ canonical_map = {
 ---
 
 ### 4. Opposition Cases Overlay
-**Status**: Anecdotal evidence of community opposition to new data centers
+**Status**: 18 geocoded opposition cases in `opposition_cases_geocoded` table (loaded but unused)
 
 **Approach**:
-- Compile cases of rejected/delayed data center proposals (news archive scraping)
-- Geocode opposition cases, join to demographics/hazards
+- Expand dataset: Compile additional cases of rejected/delayed data center proposals from news archives
+- Geocode all opposition cases, join to demographics/hazards
 - Test hypotheses:
   - Do wealthier/more educated communities successfully block projects?
   - Are opposition cases more common in water-stressed or drought-prone areas?
   - Do smaller non-metro communities have less bargaining power?
+  - Does clustered vs. isolated location predict opposition likelihood?
 
 **Research Questions**:
 - What predicts opposition success?
 - Are opposition cases spatially clustered?
 - Do demographics differ between accepted vs. rejected sites?
+- Correlation with FEMA hazard exposure scores?
+
+**Analysis Plan**:
+```sql
+-- Join opposition cases to demographics
+SELECT o.*, ct.median_household_income, ct.bachelors_or_higher_pct
+FROM opposition_cases_geocoded o
+JOIN _dc_census_tract_acs_2024 ct 
+  ON ST_Contains(ct.geom, o.geom);
+```
 
 **Output**: `opposition_cases_analysis.md`
 
 ---
 
 ### 5. IM3 Forward Projection Integration
-**Status**: IM3 model includes projected data center demand growth
+**Status**: IM3 model data loaded in `im3_state_projected_moderate_50` (328 rows) and `im3_projected_state_demand_summary` (31 rows)
 
 **Approach**:
 - Load IM3 projected demand scenarios (2030, 2040, 2050)
@@ -113,10 +134,34 @@ canonical_map = {
   - Land availability (zoned industrial parcels)
 - Identify regions where projected demand may exceed infrastructure capacity
 
+**Grid Saturation Context** (from current analysis):
+- **New Jersey**: 83% of grid within 50 km of DC
+- **Nevada**: 75%
+- **Tennessee**: 70%
+- **Oregon**: 68%
+- **Arizona**: 56%
+- **Virginia**: 50%
+
 **Research Questions**:
 - Which states face grid saturation from data center growth?
 - Are projected sites in water-stressed watersheds?
 - Does IM3 assume spatial distribution patterns consistent with current siting?
+- Can states with >50% grid saturation accommodate projected demand?
+
+**Implementation**:
+```sql
+-- Compare current saturation to IM3 projected demand
+SELECT 
+  current.state,
+  current.dc_count,
+  current.pct_grid_saturated,
+  proj.facility_count AS projected_new_facilities,
+  proj.total_it_mw AS projected_new_mw
+FROM state_grid_saturation current
+JOIN im3_projected_state_demand_summary proj ON current.state = proj.state
+WHERE current.pct_grid_saturated > 50
+ORDER BY current.pct_grid_saturated DESC;
+```
 
 **Notebook**: `im3_projection_overlay.ipynb`