Collection of Historical Topographic Maps
The study of historical topographic maps reveals fascinating insights, and in this regard, a meticulous collection of maps from Kenya and the UK serves as the foundation of our research. These maps were predominantly published between the 1950s and the early 1980s, offering a glimpse into the geographical evolution of Kenya during a significant period in its history.
Sources and Acquisition
The majority of the topographic maps for this study were sourced from the Survey of Kenya, the official authority responsible for cartography in the region. This institution has a long history, dating back to 1903 when it was first established in Nairobi. It aimed primarily at registering land claims of settlers and facilitating transactions for various commercial enterprises. By the 1950s, it had expanded its operations significantly, focusing on mapping rural land, which was an essential part of the colonial governance structure.
In total, our collection comprised 449 maps at a 1:50,000 scale and 71 maps at a 1:100,000 scale. However, a notable portion of the mapped data—166 of the 1:50,000 and 94 of the 1:100,000—were only available as data frame clips without clear publication info. While many maps conformed to established topographic standards, some were merely technical drawings, yet they captured essential road networks, vital for our study.
Given the constraints in procuring comprehensive coverage from the Survey of Kenya, we also purchased a selection of maps from UK archives. Notably, nine maps were acquired from Cambridge University Library and an additional four from the Bodleian Libraries of Oxford. These maps were produced by the Directorate of Overseas Surveys (DOS) and the War Office, showcasing collaborative international efforts in mapping. Despite our efforts, there remained an area of 31,400 km² for which we could not obtain suitable maps for our research.
Provenance of the Historical Maps
A deep dive into the provenance highlights the role of the Survey Department in Kenya’s colonial history. Originally tasked with registering land claims and issuing deeds, the department transformed over time into the Survey of Kenya. It became recognized for its technical capabilities, particularly in aerial surveying starting from the late 1930s.
The Directorate of Overseas Surveys, established post-World War II, played a pivotal role in global mapping of British dependencies, notably in Africa. The DOS’s systematic approach to aerial surveys significantly advanced the quality of maps produced during this era and laid the groundwork for detailed topographic information. The collaboration between DOS and the Survey of Kenya continued even after Kenya’s independence in 1963, contributing to a rich cartographic legacy.
Scanning & Georeferencing
Once we acquired these precious maps, a significant step was their digitization. We scanned all analogue maps into high-resolution raster data to facilitate further processing. The digital maps were then processed using ArcGIS Pro 3.2, where we meticulously georeferenced them to the Universal Transverse Mercator (UTM) projection, essential for ensuring consistency and accuracy in spatial analysis.
Given Kenya’s geographical position straddling the equator, the maps were projected into appropriate UTM zones based on their location. A Topographical Index Grid shapefile, developed by the Survey of Kenya, served as a reference grid during the georeferencing process, allowing us to maintain consistency across the datasets.
Training & Classification
The subsequent phase revolved around road extraction, essential for understanding connectivity across historical landscapes. The maps featured various road types—Main Road, Secondary Road, and Dry Weather Road—with different symbology. To prepare for extraction, maps were grouped based on style and appearance, given their varied representations of similar features.
A careful training process was implemented, where individual road symbols were classified through a systematic approach of twenty classifications to encompass the diverse representations discernible across different maps. Ground truth datasets were established in ArcGIS Pro, ensuring representative coverage of road types for accurate model training.
Postprocessing
As the road networks were extracted, the need for conversion from raster datasets to vector data became apparent. This transformation permitted extensive dataset processing, including simplifying polygons and converting them into line shapefiles. Through various tools in ArcGIS, we streamlined our datasets, ensuring accuracy and continuity.
Further enhancements in the classification accuracy followed by manually rectifying any remaining misclassifications strengthened the dataset. Topological checks ensured the integration of divided segments while eliminating overlaps and undershooting lines, yielding a seamless, cohesive road network dataset.
Validation
To assess the extraction accuracy rigorously, we adopted comprehensive evaluation metrics. Confusion matrices calculated the model’s overall accuracy, alongside precision and recall scores. The model’s efficacy was encapsulated in the F1 score, balancing precision and recall to quantify performance.
We also utilized the Intersection over Union (IoU) score to evaluate overlapping segments between classified and ground truth datasets. These scores ranged from 0 to 1, with higher values indicating better accuracy, offering a reliable basis for assessing our classification’s performance.
In tandem with these metrics, positional accuracy assessments were vital to understand how effectively the extracted highways corresponded to their actual locations. By comparing the historical dataset intersections against a contemporary road network provided by the Kenyan Roads Board, we calculated mean deviations to establish spatial accuracy.
Conclusion
Delving into the intricacies of historical topographic maps not only enriches our understanding of Kenya’s developmental trajectory but also exemplifies the robust methodologies applied in mapping historical landscapes. Through meticulous acquisition, processing, and validation of topographic data, we can unveil connectivity patterns crucial to understanding Kenya’s geographical and historical narratives.