Advanced algorithms reduce land cover classification errors in landslide analysis.

Land use and land cover (LULC) analysis is becoming increasingly important in environmental research as it has a direct impact on the environment. Changes in LULC not only increase the susceptibility of landscapes to hazardous phenomena, but also affect ecological and climate balance.

However, one of the key challenges when analyzing LULC time series is the presence of classification errors that can cause invalid transitions. Invalid transitions represent the improbability or impossibility of land cover change over a particular period of time, especially in highly sensitive areas such as mountainous areas, which may cause or result in hazardous events. may lead to misunderstandings.

To improve the accuracy of identifying these changes, a team of researchers from Brazil, Ecuador, and China developed a technique called RF, which integrates the temporal approaches of the Random Forest (RF) algorithm and the composite a posteriori maximum (CMAP) algorithm. announced. -CMAP.

Unlike traditional methods that treat each year independently, the CMAP algorithm takes into account the temporal dynamics of LULC changes. We assess the probability of transition over time and confirm that the reported changes are consistent with observed natural processes. By integrating the advantages of RF with CMAP, the new RF-CMAP method reduces invalid transitions and improves LULC classification.

Improving accuracy of land cover change analysis

The classification process integrated the probability of each LULC class being classified within each image pixel determined by the RF algorithm with the temporal approach provided by the CMAP algorithm. For this purpose, Landsat images from three years (2000, 2008, 2016) were used for analysis and the results were compared with those obtained using traditional RF methods.

Although both methods showed high performance with an overall accuracy value above 0.87, the RF-CMAP method outperformed RF in all years analyzed, with a total of 99.92 km2 (total 12% of the area).

The study is published in the journal Remote Sensing Applications: Society and Environment.

This study also focuses on validation and performance analysis of the classifications produced by each model. For example, the overall accuracy of the LULC change region from 2000 to 2008 is 0.622 for RF and 0.703 for RF-CMAP, with RF-CMAP accounting for 78% of the errors related to invalid transitions during this period. Fixed. The error correction rate with RF-CMAP increased to 81% from 2008 to 2016.

Furthermore, RF-CMAP significantly reduced the salt-and-pepper effect, increased the homogeneity of the classified regions, and eliminated the errors observed in RF classification. This includes significant improvements in the classification of areas where LULC does not change, such as forests, bare land, and water bodies. Between 2000 and 2008, RF-CMAP corrected 50% more errors in these areas than traditional RF methods.

The important role of technology in preventing disasters caused by natural disasters

In 2017, heavy rains occurred in the L’Orante River basin, the study area, and more than 300 landslides occurred. When extreme rainfall events are combined with changes in LULC, soil instability can increase. In this context, accurately identifying changes in LULC is essential to understand the factors contributing to the occurrence of dangerous phenomena and to prevent future disasters.

The study found that 35% of landslides could be associated with invalid transitions between 2000 and 2008, and 16% between 2008 and 2016. There is a gender. These invalid transitions can misrepresent environmental conditions that cause landslides.

For example, 35% of these landslides may be related to reforestation, even though there is no evidence that reforestation has occurred in this area over eight years. The RF-CMAP method effectively avoided these invalid transitions and accurately classified 66% of the landslide-affected areas. In contrast, traditional RF models had only 21%.

In conclusion, the integration of advanced technologies such as RF and CMAP represents an important advance in the temporal analysis of LULC changes and provides valuable insights for improving disaster risk management. This model has the potential to significantly enhance disaster prevention and protect vulnerable communities by addressing ineffective migration in landslide-prone areas.

As remote sensing technologies and predictive algorithms continue to improve, their widespread adoption has the potential to revolutionize the sustainable management of natural resources and support disaster risk management.