Development and testing of geo-processing models for the automatic generation of remediation plan and navigation data to use in industrial disaster remediation

Abstract

This paper introduces research done on the automatic preparation of remediation plans and navigation data for the precise guidance of heavy machinery in clean-up work after an industrial disaster. The input test data consists of a pollution extent shapefile derived from the processing of hyperspectral aerial survey data from the Kolontar red mud disaster. Five algorithms were developed, the respective scripts were written in Python, and then tested. The first model aims at drawing a parcel clean-up plan. It tests four different parcel orientations (0, 90, 45 and 135°) and keeps the plan where clean-up parcels are less numerous. The second model uses the orientation of each contamination polygon feature to orientate the features of the clean-up plan accordingly. The third model tested if it is worth rotating the parcel features by 90° for some contamination feature. The fourth model drifts the clean-up parcel of a work plan following a grid pattern; here also with the belief to reduce the final number of parcel features. The last model aims at drawing a navigation line in the middle of each clean-up parcel. The best optimization results were achieved with the second model; the drift and 90° rotation models do not offer significant advantage. By comparison of the results between different orientations we demonstrated that the number of clean-up parcels generated varies in a range of 4 to 38 % from plan to plan. Such a significant variation with the resulting feature numbers shows that the optimal orientation identification can result in saving work, time and money in remediation.