Warning: This post may seem off-topic or even scattered, but stay with me!
In last week’s post, we presented some preliminary analyses to show how temperature data can be extended both spatially and temporally using select NCDC-Daily weather stations. These NCDC-Daily weather stations were well correlated to the CBNERRS sites of Jug Bay, Maryland and Taskinas Creek, Virginia (Tables 1 & 2).
We now have an additional 12 meteorological datasets for Maryland (Northern Chesapeake Bay) and 14 for Virginia (Southern Chesapeake Bay) in addition to the CBNERRS site data. However, temperature can vary by location depending on many factors
Here is a way to think about how important land-use can be: think about how hot black asphalt gets in the summer compared to a field of green grass. A simple change in color, moisture, and texture can alter how much heat a surface can absorb or reflect. This is why you are likely to wear a light-colored t-shirt on a hot day. The word used to describe this process is albedo, which is the measure of a surface’s reflectivity. A pure white object reflects 100% of incoming sunlight while a pure black object would absorb 100% of that sunlight.
This same idea happens on land, but of course is more complex than a black or white t-shirt! As we change the local and regional landscape, we may unintentionally change the ability of that landscape to absorb and/or reflect incident sunlight. (And, I must add, we change many other conditions such as the permeability of the ground type that can also affect surface temperature!)
With that in mind, it may be a good idea to “classify” the land-use surrounding each of the selected weather stations before the extreme temperature indices are calculated. To make things easy, I was able to use Google Earth to map the population density and cropland percentage by county for both Maryland and Virginia.
The United States Geological Survey created a map of the percentage of total cropland and the 2000 United States Census Database per county. Additionally, the United States Census Bureau created a map of the 2010 total population by state, county, and town. I also acquired a map from Juice Analytics of the population density for Maryland and Virgina from the 2000 Census Bureau by county. The population density was adjusted using the population change by county from 2000 to 2010.
These maps, as you can tell from their titles, give the percentage of land that has agricultural crops (and specifies the dominant crop type) as well as the population density in number of people per square mile. The Chesapeake Bay region has varied amounts of cropland dominated by corn and soybeans. As one can guess, there is an inverse relationship between population density and croplands (Figure 3). In other words, we can loosely say that the more people per square mile, the more urban the area; conversely, the less people per square mile, the more rural the area.
So what could this mean for our weather stations? Well, this allows us to qualitatively classify stations as rural, suburban, or urban depending on the amount of cropland versus population density. Why is this important? The answer is to identify possible regions with urban heat island effects. Urbanization can influence the local temperature in many ways, but mostly will cause the immediate area to be warmer than an open field in the same region.
Think about all the pavement that an urbanized city has. That blacktop is impervious, preventing soil moisture from escaping into the air. This makes the air drier in cities. Sunlight, with the drier air, can be absorbed by dark surfaces (like rooftops and pavement), making the area warmer. That heat is then radiated back into the air at night, again causing the temperature to stay warmer than if it had been a lush forest. Anyone who has ever lived in a city can tell you that you get no relief from a heat wave at night!
In a nutshell, weather stations located in more urbanized areas may have this anthropogenic influence of being warmer than the regional climate would expect. We believe that this is an important factor for of climate assessment. As the land-use in Chesapeake Bay changes, habitats in that area could experience those changes. For example, an increase in urbanization in one area could cause the temperatures to become elevated more than expected. Thus, we want to classify each station to predict how the extreme climate indices may change as the landscape changes.
So, can we classify these stations as urban, suburban, or rural? This task is more difficult than one would think. The qualitative classifications presented below are a preliminary, work-in-progress, and could very likely change! My classification preliminary guide is demonstrated in Figure 6. I defined the “urban” boundary as a weather station in a county with <30% cropland and a population density >4000 people per square mile. “Rural” was defined as a weather station in a county with >70% cropland and <1000 people per square mile. “Suburban” was defined as a weather station in a county between 30-70% cropland and 1000-4000 people per square mile. Weather stations that fell in “mixed” zones were classified as a mix of urban, rural, and suburban.
The take-away: land-use is important to consider for temperature studies, but it is difficult to classify. And, Google Earth is a fun, and completely free, technique to look at land-use. Here is a nice website that gives some information on using Google Earth to look at a watershed.