Summary

  • USGS described three parallel versions of its 1996 Northridge report: print, downloadable PDFs and an HTML tour with direct routes into chapters and topics.
  • Landslide maps and their underlying datasets also went online. Their usefulness began with fieldwork, aerial photographs, manual tracing and help using GIS.

A book has strong opinions about where you should begin. Scientists tend to arrive with questions of their own. In 1996, USGS described an arrangement that gave its readers more room to be awkward: a report on the Northridge earthquake with several ways in, and useful material to take away.

The printed volume had an electronic replica in several PDF files. Beside it sat an HTML tour. A reader could enter a chapter or topic directly from the earthquake home page, then follow links to supporting papers, reference lists and other material. The table of contents had acquired doors.

USGS had produced these versions in parallel. It also planned to keep updating the tour as new information arrived and obsolete material was removed. The arrangement appears on page 70 of the original 1996 report. A publication was becoming a continuing appointment with its readers. Someone would still have work to do after the copies went out.

The landslide inventory shows what this change made available. The earthquake struck Northridge, California, on 17 January 1994. Edwin L. Harp and Randall W. Jibson's subsequent inventory recorded more than 11,000 landslides. Their 1995 publication record describes field mapping, aerial photography and maps produced with a geographical information system, or GIS.

The files began with people looking hard at the ground. Aerial photographs taken within hours of the earthquake and field observations fed maps drawn on USGS topographic sheets at a scale of 1:24,000. The mapped landslides were then manually digitised in Arc/Info GIS as polygons. In the mapping account, the authors explain how sunlight and shadow affected what could be recognised. The clean shapes on a computer screen carried earlier decisions about where a slide began and ended. A download inherited all that work.

Making the files useful required some less photogenic provision too. The Computer Graphics Laboratory in Golden, Colorado, supplied hardware, GIS software, training and assistance. It helped investigators visualise their results, ran an FTP server for sharing spatial data among USGS scientists and maintained a web distribution site. Support extended to Menlo Park and Pasadena. The early web had a back room full of equipment and people who knew how to use it.

The 1996 report says the landslide inventory and the datasets used to make it were available through anonymous FTP. A hypertext version with graphics and photographs was on the web. Digital information could reach the wider connected public after review and release. Previously scattered group home pages were brought together through a common Northridge page. Readers got a place to start; the different research groups kept doing their particular jobs.

This combination let a map serve more than one appetite. A visitor could study the finished picture. An investigator could use the underlying material in another calculation. In a 1998 USGS study, Randall W. Jibson, Edwin L. Harp and John A. Michael developed a way to produce probabilistic landslide-hazard maps using Northridge data, demonstrated in the Oat Mountain quadrangle.

They brought the landslide inventory together with strong-motion records, geology, material strength and elevation models in GIS, organising inputs on a 10-metre grid. Comparing calculated displacement with mapped failures helped establish a probability relationship. The inventory had become an ingredient in another piece of research. Finished maps could describe the earlier investigation while their data went back to work.

That is the appealing ambition inside this modest episode of Internet history. Help one reader browse, another find a particular chapter and a third carry on calculating. The book remained useful. Its covers had simply become too small for the job.