NOVEL APPLICATIONS OF GIS AND ARTIFICIAL INTELLIGENCE IN FOREST RESTORATION

Abstract

Forest restoration programmes take place globally and lay a pivotal role in addressing climate change and biodiversity loss. Often restoration programmes are based on simple plantation schemes, evenly planting trees that later on might contribute to economic activity. This, however, does not seem to be sufficient for supporting biodiversity. Recent research suggests that successful restorations should match original ecological patterns in any particular landscape, assuming that severe erosion and changing soil conditions have not taken place during disturbances. This means that understanding natural historic patterns is vital. However, achieving such understanding is often challenging, given the fact that historic satellite imagery is generally available only for relatively short time periods. It is therefore important, if possible, to model former landscape ecological patterns. Modelling might be based on different site-specific approaches and historical records. However, most powerful tools available today include deep learning and artificial intelligence. Construction and training of neural networks might allow simulation of historical forest patterns in cases when satellite imagery is not available for long time periods. Application of this technique is very likely to have important practical implications.