Between Proxy and Instrument

In this dissertation, I produce novel scientific knowledge about the weather of the past using subjective human observations in the absence of instrumental records in the United States. Using a handwritten plantation journal from Shirley Plantation near Richmond, Virginia, I transcribe 34 years of sub--daily observations records (multiple observations throughout the day) spanning the period of 1816-1861. The resulting database contains 36,040 entries, averaging over 1000 entries for each year. I categorize and generate metadata for each of the entries. I use the resulting database to investigate the effects of volcanic forcing, develop a method for identifying tropical cyclones, evaluate the accuracy of written precipitation records, and test for spring advancement over the past two hundred years at my study site. The goal of this dissertation is to add to our knowledge of the local weather and climate for the study area in the southeastern United States during the antebellum period (1815-1860). My project draws on methods from across the field of geography. I begin with archival research utilized in historical geographic research. I bridge the gap between qualitative and quantitative methods by extracting the observations recorded in the historical documents and coding them into a dataset. These data are then used to answer hypothesis--driven questions about the weather and climate of the region during the study period. Plantation records from the southeastern United States have long been an important source for historical social and cultural narratives. However, they also represent an underutilized source for meteorological, environmental, and agricultural data from the antebellum period. While individual plantation records may have idiosyncratic limitations, plantation records, along with other types of detailed historical records, can still provide rich detail for specific locations or events. Plantation records are not limited to the southeastern US and include diverse geographic locations in less developed areas which were often the same areas where enslaved labor was exploited under the plantation system. Contextualizing historical data with present-day knowledge can be used to examine how society interacted with inter--annual climate variability and extremes. This study has three goals. Firstly, I advocate for a more systematic application of these records for quantitative analysis. Secondly, I show how these records can be mined for data on weather. These data lend themselves to quantitative analysis that can improve our understanding of the local weather and climate of that period, and how it has changed in the past two centuries. I introduce a method for extracting weather and climate data from a historical plantation document. I demonstrate the method on a document from Shirley Plantation in Virginia (USA) covering the period 1816--1861. I show how the resulting data are coded into a database that includes direct weather observations. Finally, I present four case studies using the Shirley Plantation database. Case study one investigates volcanic forcing following a large volcanic eruption in 1815, where I hypothesize that, on average, final spring freezes events occurred later in the spring for the earliest years in the record when compared to the later years in the record. Case study two examines indications of tropical cyclone activity in non-instrumental records, where I hypothesize that the impact of tropical cyclones along the mid Atlantic coast can be detected by specific weather changes at Shirley Plantation. Case study three evaluates the accuracy of written precipitation records compared to dendrochronological proxies, and I hypothesize that a precipitation index constructed from the database will correlate with precipitation estimates from tree rings. Case study four tests for spring advancement at Shirley Plantation over the past two centuries. I hypothesize that spring advancement caused by global warming can be identified by later last freeze dates in the Plantation data compared to last freeze dates from modern records at the nearby Richmond airport. I end with a discussion of the database and studies, the broader implications of the research I present, and a plan for future research.