Jacarandas in Mexico City
The real reason is because they are beautiful and we actually have one of them in our colonial house in Coyoacan. But after so many years using GIS in my work, I was amazed of how much interest a visualization of Jacarandas in the Center of Mexico City brought to my Twitter account, so I guess this deserves some more explanation.
My interest in Jacarandas, or rather more generally, geocoding the tree cover of downtown Mexico City started with Acequias, the distinctive canals we can no longer see (a short lived reconstruction, now disappeared, was made in the street of Corregidora next to the Supreme Court), because the lake was drained as the City grew into the huge metropolis it is today. A tribute to the globalization of the 16th century is that the prehispanic apantli (canal) was replaced by a term from the Arabic (irrigation canal, sāqiyah), that the Spanish used to describe the ditches that allowed navigation into the City. Acequias were used throughout the colonial period to bring food and supplies to the City from the surrounding towns. Perhaps the most famous of them was the Acequia Real, that continued connecting fruit and flower growers in the chinampas of Xochimilco with the City, as can be seen in a beautiful biombo kept in the Museo Franz Mayer.
The canals are clearly visible in the first known map for Europeans of Tenochtitlan dating from 1524, a widely reprinted image accompanying the second letter from Hernan Cortez to the King of Spain. By far the most accurate depiction of the island city surrounded by its lakes, crisscrossed by acequias, is, however, the so called Uppsala Map, a detailed painting of the city after its reconstruction before the 1550s produced by indigenous artists, most likely from the Colegio de los Naturales in Tlatelolco. That map has sometimes been credited to Alfonso de Santa Cruz, who included in his Islario a simplified version of the map. The Santa Cruz Islario shows the two most magnificent city states surrounded by lakes in the world, Venice and Mexico City.
The Acequias are clearly visible in a later map of 1753 prepared by Jose Antonio de Villasenor y Sanchez, a creole intellectual who became the Contador de Azogues (Accountant of mercury, an essential ingredient for the separation of silver in the colonial period). The map by Villasenor was a complement to a monumental work called Theatro Americano, in which he depicts all the cities, towns and geographic features of the New Spain. This specific map shows all the main churches and points of interest, the neatly organized blocks of the city, and the Acequia Real with flowing waters, and smaller Acequias as well as the Aqueduct supplying water since prehispanic times from the springs in Chapultepec.
My initial assumption was that many trees today, especially any old growth trees surviving in downtown would be located either in the courtyards and surroundings of the first church edifications, or along the Acequias.
The oldest trees in Mexico City are Mexican Willows, called Ahuehuetes, which means precisely, old men of the water. I have seen very few surviving Ahuehuetes in the City proper, although it is still possible to see the remains of the Arbol de la Noche Triste (sad remembrance for the Spaniards, not for the Colhua of the Triple Alliance of the Empire of Mexico-Tenochtitlan, Tlacopan and Texcoco), where Cortez is said to have wept when he had to retreat in 1520, defeated despite his technological superiority at war making. And there are some scattered Ahuehuetes in Chapultepec park. I would want to eventually engage in a project where the surviving Old Trees in the City could be mapped as a testimony of the persistence of natural forces in a city that is all paved and built over.
But since I could not really map Ahuehuetes, and they were not easily distinguished with Google or other satellite images (although I am sure an algorithm could be programmed and machine learning process would be able to find them and catalogue them), large trees seemed to be a reasonable second best. When I started to examine satellite images, the truly striking feature I stumbled upon were the Jacarandas.
Jacarandas are not native to Mexico. They were brought from Brazil and clearly adapted well to the climate of the Basin of Mexico City. These magnificent trees turn purple in the spring and blanket the sidewalks and streets with their fallen flowers and the distinctive sap, and are familiar to anyone who has lived in the city. As far as trees are concerned they are young, just a few hundred years at most.
The images of the Jacaranda canopies are so distinctive in the satellite imagery that I am not concerned about false positives, namely placing a Jacaranda in a location that does not have one. There might be some errors in which Jacarandas are missed in dense canopies of various tree species, and it is also possible that smaller trees are missed from above.
The distinctiveness of Jacarandas (in the Spring, of course) contrasts with the difficulty at locating, for example, palm trees. Those elusive features can be difficult to geocode with satellite imagery because the branches and leaves may be less than the 1m resolution of the images. They were initially a feature of interest because old palm trees would probably only be found in neighborhoods where rich families lived in the 19th and early 20th centuries, when this type of ornament became fashionable. It is possible to know in cities like Mexico and Guadalajara where the old money families live by looking for their palm trees in their gardens.
Once I was coding Jacarandas, it would have made sense to include trees by species or size. At the beginning I was only planning to include large trees, but the canopy cover often prevents a clear definition of thresholds or even to identify individual trees. This is particularly clear in the Alameda park and other smaller parks and church atria (plural of atrium) in the downtown area. I am sure the geocoding of trees could be much improved, and as with the Ahuehuetes or the palm trees, perhaps a machine learning algorithm could distinguish these features with more accuracy than my human coding.
I should also mention something about the technical aspects of overlaying GIS features to produce the maps. In the visualizations I choose to enhance the presence of Jacarandas by making them much larger than the trees. There are around 8000 trees and only some 400 Jacarandas in the areas I coded.
The georectified map of Villasenor, and the 70 points of interest he includes, which are mostly churches, provided a reference where presumably some attraction to greater concentration of trees would be observed. And the Acequias would presumably provide a good location for an old tree to literally take root and grow. Those hypotheses were tested with some distance calculations and some spatial correlations, which suggest that the trees in Mexico City are not strongly associated with those historical legacies. There are more trees close to churches, and more Jacarandas in their atrium, but there are also many trees planted along the wide boulevards and in the newer urban compounds such as the apartments in Tlatelolco.
I am sure this geocoding can be improved, and nicer visualizations are possible. And a more automated strategy using machine learning methods could cover the whole city.
If you want to play with the data, here is a Github repository: https://github.com/adiazcayeros/jacarandas
And a map package is also (hopefully) available at my ArcGIS Online site: https://arcg.is/0a5uaq