Site Suitability: Vegetation-Based Dust Mitigation in Bakersfield, California

by Heather Robbins

Problem statement:

Bakersfield’s geography, situated at the southern end of California’s Central Valley, creates a natural basin that traps dust and particulate matter, contributing to some of the poorest air quality in the nation. Seasonal transitions from dry to wet conditions can intensify dust storms, increasing risks to respiratory health. Implementing dust mitigation planting strategies offers a proactive approach to reduce airborne particulates and improve environmental resilience in the face of climate variability.

Question: Where is the most suitable land for vegetation-based dust mitigation to reduce the risk of dust storms and associated health hazards in Bakersfield, CA?

Study area:

Bakersfield is a semi-arid city that receives little rainfall and is increasingly vulnerable to wildfires. Surrounded by mountains on three sides, the city experiences the “bowl effect,” where polluted air becomes trapped close to the ground. This, combined with high-emission industries and frequent drought conditions, consistently places the city among the top 10 most polluted in the United States (Yates, 2022).

Data:

Land Cover (California subset) is a raster data subset of the NLCD created by the USGS and provides up to date land cover classifications
Soil type data (polygons) from the Soil Survey Geographic Database (SSURGO) includes the physical, nutrient and moisture properties.
California roads (polylines) are provided by the TIGER Line shapefiles through the U.S Census Bureau.
U.S. Census populated place areas (polygons) is a subset of data provided by the U.S. Census Bureau and contains population and other demographic information.

Criteria:

Accessible via roads: potential planting of vegetation needs to be accessible via roads within 2,500 feet and will help to protect critical infrastructure in the event of a dust storm.
Proximity to urban centers: vegetation near urban centers will help to reduce particulate matter and will ideally be within 1,000 feet of the most populous area.
Vegetation ready land cover: barren or sparsely vegetated areas such as shrubs and developed open spaces are possible options for planting.
Loose soils: the soils that will work best for planting will be loose, sandy and loamy soils, which allow for quick root establishment and allow water to drain efficiently. Ideal plant species for dust and pollutant control are grasses and shrubs that will thrive in these types of soils.

Methods:

Projection: The StatePlane California V FIPS 0405 (Feet) projection is used to maintain accuracy in area and distance.

A study area surrounding the city is digitized by creating a new feature class. Each data set is then projected to the StatePlane California V projection and clipped to the study area.

The NLCD raster is clipped, reprojected and then converted to polygons using the Raster to Polygon tool.

Software: Analysis completed using ArcGIS Pro 3.4.2

Boolean analysis:

Roads are buffered by 2,500 feet
City of Bakersfield polygon is buffered by 1,000 feet.
Using the Select tool:
- land cover is identified as NLCD_Type is equal to Barren Land, Developed Open Space or Shrub
- Soil type is identified where soil name contains the text 'sandy' or 'loamy'

Each of the new selection and buffer layers are clipped to the study area, and then the Intersect tool is used to identify areas where all the criteria meets.

Weighted analysis:

Criteria is ranked from 1 - 4 (Highest Suitability - Not Suitable).

Multi-ring buffers (tool) are created on the roads layer at 1,000, 2,000 and 3,000 feet each, the Union tool is used to merge the buffers with the study area for ranking, output is then clipped to the study area and named roads_dist.

Rank A field is added to the roads_dist layer, and reclassified with the Calculate Field tool as:

def Reclass(dist):

if dist == 1000:

return 1

elif dist == 2000:

return 2

elif dist == 3000:

return 3

else:

return 4

Because the populated areas polygons do not cover the entire study area, a union is created so that non populated areas can be ranked not suitable, the output is called pop_union.

Rank B is added as a new field to the pop_union layer, and reclassified with the Calculate Field tool as:

def Reclass(pop):

if pop < 500:

return 4

elif pop >= 500 and pop < 2500:

return 3

elif pop >= 2500 and pop < 10000:

return 2

else:

return 1

Land cover polygons from the Boolean analysis are dissolved using the NLCD_Type field.

Rank C is added to the land cover layer, and reclassified with the Calculate Field tool as:

four_lands = ["Cultivated Crops", "Developed, High Intensity", "Developed, Medium Intensity", "Open Water"]

three_lands = ["Deciduous Forest", "Emergent Herbaceous Wetlands", "Evergreen Forest", "Hay/Pasture", "Woody Wetlands"]

two_lands = ["Developed, Low Intensity", "Developed, Open Space", "Herbaceous", "Mixed Forest"]

one_lands = ["Barren Land", "Shrub/Scrub"]

def Reclass(land):

if land in four_lands:

return 4

elif land in three_lands:

return 3

elif land in two_lands:

return 2

elif land in one_lands:

return 1

Soil type polygons are also dissolved using the taxonomic order field.

Rank D is added as a new field to the soils type layer, and reclassified using the Calculate Field tool as:

three_type = ["Entisols", "Vertisols"]

two_type = ["Alfisols", "Inceptisols"]

def Reclass(type):

if type == "Aridisols":

return 4

elif type in three_type:

return 3

elif type in two_type:

return 2

elif type == "Mollisols":

return 1

else:

return 4

Weighted Criteria Maps

Calculating Suitability Scores

Using the Identity tool on the roads distance layer, with identifying population layer (Rank B) to combine attribute tables, the output is called rank1.

The Identity tool is used again on rank1, to identify land cover types with all attributes except feature ID, and the output is rank2.

The Identity tool is used a final time on rank2, to identify features in the soils layer, all attributes except feature ID, and the final output is called rank3.

All fields except for rank (A - D) are removed from the rank3 layer using the Delete Field tool

Using the following weights:

Distance to Roads (15%)
Population (15%)
Land Cover (40%)
Soil Type (30%)

A new field is then added to the rank3 layer, 'suitability' and is calculated using the Field Calculator tool as:

(0.15 * !RankA!) + (0.15 * !RankB!) + (0.40 * !RankC!) + (0.30 * !RankD!)

Model Assumptions

The surrounding regions of Kern County were included in this analysis to capture potential planting suitability at the urban fringes of Bakersfield. While some areas identified in the weighted suitability model are located far from major roads or population centers, and therefore may be less practical for immediate implementation, they are still relevant in a long-term planning context. The Boolean analysis provides a more restrictive perspective, identifying only those areas where all criteria are fully met. Future model iterations could expand or refine these parameters to include additional or alternative criteria.

Assumptions for the weighted overlay model are based on the idea that soil and land characteristics play a greater role in determining the success of vegetation planting than proximity to roads or population density. For example, highly developed urban areas and actively cultivated agricultural lands were ranked not suitable due to limited accessibility or land use conflicts. Similarly, Aridisols—characteristic of very dry, low-organic-matter soils—were considered unsuitable for effective plant growth.

To reflect these priorities, land cover and soil type were assigned the highest weights in the model:

Land cover: 40%
Soil type: 30%
Proximity to roads: 15%
Distance from population centers: 15%

While proximity to roads and urban areas is advantageous for planting logistics and maintenance, it was not considered essential to identifying high-priority zones for dust mitigation through vegetation. The final model reflects this balance between biophysical feasibility and accessibility, acknowledging that some trade-offs may exist between ideal planting conditions and ease of implementation.

Results

The first analysis (Boolean) identifies several areas along the Old River (Highway 119) in southwest Bakersfield, Westside Parkway along the Kern River leading into downtown, the Rexland Acres community and along Highway 178 between the Bakersfield Country Club and Rio Bravo Golf Club. A general trend of 'patches' is noticeable along roads where these areas meet criteria of sandy, loamy soils and are available for planting.

The second analysis (weighted and ranked) finds the most suitable locations are along the Highway 119, between Panama and Greenfield communities, and the hillsides along east Bakersfield near the Rio Bravo Golf Club. While many areas are deemed suitable (calculated scores 1 - 2) throughout the study area, they are not within the city limits nor within a reasonable distance to roads.

Weighted Suitability Results

Conclusion

While both analyses identified overlapping areas of suitability between the Bakersfield Country Club and Rio Bravo Golf Club, the Boolean method produced a much broader surface area of potential sites. In contrast, the weighted analysis—by incorporating ranked factors such as proximity to roads and population centers—significantly reduced the selection, with each of those criteria contributing just 15% to the total suitability score. One of the most notable differences is that the weighted model did not identify any highly suitable locations within the city limits of Bakersfield. Instead, suitable zones were concentrated along the city’s periphery.

Several areas deemed ideal for vegetation planting in the weighted analysis are located far from major roads, presenting a practical limitation when considering access for implementation. Additionally, the broad extent of the study area may have contributed to the identification of suitable zones outside a reasonable operational range from Bakersfield, which could also be seen as a constraint of the analysis.

Ongoing research continues to explore best practices for reducing dust concentrations and mitigating dust storms—particularly in arid and drought-prone environments. Recent findings suggest that bare soils exposed to prolonged drought and strong winds are key contributors to dust storms, while vegetation such as trees may provide either mitigation or unintended consequences, depending on regional atmospheric and climate conditions (Jiao, 2025).

Future research could incorporate wind speed and directional modeling to refine site selection and maximize the effectiveness of dust mitigation strategies. As climate change intensifies extreme weather events—ranging from wildfires and hurricanes to severe droughts—air quality concerns and dust pollution are often overlooked. With more robust planning tools and targeted analysis like this study, communities can be better equipped to prioritize dust mitigation in both environmental and urban planning efforts.

References

Calculate Field Python Examples—ArcGIS Pro | Documentation. (n.d.). https://pro.arcgis.com/en/pro-app/latest/tool-reference/data-management/calculate-field-examples.htm

Jiao, H. (2025, April 10). Research & Conservation: Reviewing the Illinois dust Storm of 2023. Farmdoc Daily. https://farmdocdaily.illinois.edu/2025/04/research-conservation-reviewing-the-illinois-dust-storm-of-2023.html

Models warn of dustier summers in U.S. Southwest and Southern Plains, but Northern Plains may see fewer dust storms year round. (2018, June 25). NOAA Climate.gov. https://www.climate.gov/news-features/featured-images/models-warn-dustier-summers-us-southwest-and-southern-plains-northern

National Land Cover Database Class Legend and Description | Multi-Resolution Land Characteristics (MRLC) Consortium. (2024, August 22). https://www.mrlc.gov/data/legends/national-land-cover-database-class-legend-and-description

Official Soil Series Descriptions (OSD) | Natural Resources Conservation Service. (n.d.). Natural Resources Conservation Service. https://www.nrcs.usda.gov/resources/data-and-reports/official-soil-series-descriptions-osd

Soil Survey Geographic Database (SSURGO) | Natural Resources Conservation Service. (n.d.). Natural Resources Conservation Service. https://www.nrcs.usda.gov/resources/data-and-reports/soil-survey-geographic-database-ssurgo#:~:text=The%20%EE%80%80SSURGO%EE%80%81%20database

U.S. Census Bureau. (n.d.). California primary and secondary roads [Dataset; Feature Service]. https://www.arcgis.com/home/item.html?id=edcc4c8d875644b18557e4d42bb78ca5

U.S. Census Bureau, U.S. Department of Commerce, & National Geodetic Survey. (2013). USA census populated place areas [Dataset; Feature Service]. Esri. https://www.arcgis.com/home/item.html?id=d8e6e822e6b44d80b4d3b5fe7538576d

U.S. Geological Survey. (n.d.). NLCD 2021 Land Cover California Subset (Version 2021) [Dataset; Image Service]. https://www.arcgis.com/home/item.html?id=8e336f9a4a644db3b8627f07a7e19d97

USDA Natural Resources Conservation Service. (2019). USA soils map units [Dataset; Feature Service]. Esri. https://www.arcgis.com/home/item.html?id=06e5fd61bdb6453fb16534c676e1c9b9

Yates, B. (2022, August 15). Why does Bakersfield have bad air? PartyShopMaine. https://partyshopmaine.com/bakersfield/why-does-bakersfield-have-bad-air/

Page updated

Google Sites

Report abuse