100+ degrees and nary a cloud in the sky. These are typical peak summer conditions. The shade under a tree during a Texas summer can certainly take the edge off the heat. During most of the day, you might tend to avoid going outside altogether. But your house, and your neighborhood are going to feel very different if they are under a canopy of leaves compared to just sitting out there baking under the direct sun.
Excess heat means more energy costs
Areas without shade tend to accumulate a lot more heat during the day and radiate it back out at night, keeping an area much hotter than it is supposed to be. This means more energy costs. Not to mention wear and tear on your foundation from the famously "expansive soils" of Texas that can crack the foundation slab supporting your house.
A recent series of New York Times articles described big differences in shade and vegetation within a city and linked those differences to the history of those neighborhoods, and by extension, their income levels.
First, the rankings
By using data from satellites, it is possible to measure the amount of greenery in a location.
Which cities in the D-FW region had the most shade? The tree-lined enclaves of the Park Cities rank highly but don't make the top 10. Interestingly, the south side of the Metroplex has some very dense foliage: Cedar Hill and Duncanville, thanks to the green hills adjacent to Cedar Hill State Park.
The "Cross Timbers" cities of Colleyville, Flower Mound, Keller, Denton and Southlake rank highly thanks to the narrow band of forest that led to plentiful trees in these cities. On the other extreme we have the cities in more of a prairie setting, and newer neighborhoods with fewer mature trees: Allen, Frisco, Plano, Carrollton, and The Colony. To make this comparison fair, we focus on neighborhoods only (with single family homes), excluding all other land uses, and excluding flood plains.
Let's get detailed. Showing the greenery index by Census block, we can see a lot of intricate differences between neighborhoods at a micro level.
Source: NDVI (Normalized Difference Vegetation Index), captured through Copernicus Open Access Hub of the European Space Agency Note: To make a fair comparison, flood plains (including creeks) which are not habitable were excluded. The darkest greens correspond to an NDVI of 0.84 or higher, where the highest level is 1.0.
Relationships to factors like income
We set out to see what location factors influence the level of shade in Dallas-Fort Worth city blocks. We looked at relationships between greenery and the age of the community, as well average home prices.
We went into this thinking that greenery should simply be proportional to the age of the neighborhood and proximity to a creek or flood plain. After all, decades-old neighborhoods more often have mature trees, and homes closer to a creek have clusters of trees taking advantage of the year-round water source that is not reliable on higher ground. Also, certain areas of the Metroplex have more trees due to soil differences. The Eastern Cross Timbers are a 15-mile-wide belt of woodland that extend through areas like Flower Mound, Southlake, Colleyville, and Arlington. That belt of woodland tends to be greener, on average.
The averages do suggest that older neighborhoods tend to be leafier. It also suggests that the Blackland Prairie neighborhoods (i.e. Dallas) don't reach peak green for about 60 years after they are built, meaning neighborhoods built 1960 or earlier. Meanwhile, those in the Cross Timbers and Grand Prairie (not to be confused with the city of Grand Prairie) eco-regions, might reach peak green in about half that time.
More greenery = $$$
Also, higher priced homes tend to have more greenery, especially in the Blackland Prairies and Grand Prairie eco-regions, suggesting greenery commands a premium. This makes sense when you consider the mature, close-in neighborhoods of Dallas tend to be greener. This seems to be less true in the Cross Timbers belt, where a higher price does not clearly equate to more greenery. This makes sense as homes in Arlington, parts of which are in the Cross Timbers, are not too expensive. We put all of the houses on a 1-10 price scale to account for the long tail of expensive homes that run into the millions. Those get condensed into the 9-10 category.
Is there clear cause-and-effect?
What if we test for true cause-and effect? We looked not just at age of the neighborhood but a a whole host of potential greenery predictors: population density, household income, home prices, demographics, gentrification, age of the homes, retail development, and growth levels. Even with all of these factors considered, taken together, they only explained about 35% of the variance in greenery. Age and housing price were the most powerful predictors out of this group, but even these were imperfect. Therefore, while things like price and age do matter, there are many exceptions.
Highly granular voting data for the 2016 election was compiled a few years ago and made public online via websites operated by Harvard and MIT. For the first time, this made precinct-level data across the country relatively accessible. "Voter precincts" are fairly small, at about a third of a square mile to one square mile in a typical urban area. But the people living within a given precinct can vary widely. From one neighborhood or subdivision to the next, there are different types of housing for different income levels and family sizes. These differences contribute to different political preferences.
What if we wanted to estimate what is happening at the city block level? You may want to do this when screening blocks that you want to live in, if it is important to you to find the prefect niche for yourself to live where you feel most comfortable. Or, it is valuable for other purposes, like campaign planning. Referring to our favorite diagram, we can see just how much more detail we can get by taking it down to the block level.
Let's get detailed!
We can get an idea of the block level patterns by using the datasets we developed for the Locate Alpha product. We use location factors and block-level demographics to create a predictive model. See some example results below:
About Spatial Laser
We are building a software-as-a-service offering to help real estate investors in single-family homes form a micro-level investment strategy and avoid making mistakes. We use machine learning to model sales prices, rents, investment returns, risk, and market conditions at the hyper-local level using ranking and scoring. We make it easy and actionable with maps and color codes.
As we build our product, we are looking for early adopters and test users. Be the first to know about product updates by signing up here: https://mailchi.mp/spatiallaser/prelaunch