Retrieval Of Forest Fire Burn Severity With Coupled Radiative Transfer Model And Remotely Sensed Data

In recent years,with the global warming,the occurrence of forest fires in the world has increased.Once a forest fire occurs,how to quickly grasp the severity and distribution of the fire is of great significance for forest management and forest protection.Forest burn severity is generally defined as the impact of fire on vegetation and soil in a specific area.The traditional assessment of forest burn severity is mostly completed through artificial field investigation,which requires a lot of manpower,material and financial resources.Moreover,the artificial field survey can only cover a small range of landscape scale.Therefore,the traditional method has great limitations in research area and scope,and it is time-consuming and laborious.Satellite remote sensing technology relies on its advantages of high spatial-temporal resolution and large-scale monitoring,which provides a new way to evaluate forest burn severity.Fire will have a direct impact on the physical and chemical properties of surface vegetation and soil.Different state of vegetation,soil and ground ash have different echo signals to remote sensing sensor,so it is possible to estimate forest burn severity based on satellite remote sensing data.In view of this,this paper focuses on the spectral confusion among different forest burn severity caused by the change of tree coverage in remote sensing inversion of forest fire burn severity,and analyzes the interaction between forest fire burn severity and its environmental driving factors.Three kinds of work were conducted:(1)The influence of tree coverage on forest burn severity inversion was studied.The change of tree coverage was integrated into the modeling and inversion of multilayer vegetation radiative transfer model to improve the matching of burn severity and spectral reflectance of remote sensing image.The spectral reflectance characteristics under different measured burn severity extracted from remote sensing images show that tree coverage will have a great impact on the accurate estimation of burn severity.Specifically,it overestimates the low burn severity in the low tree coverage area and underestimates the medium and high burn severity in the high tree coverage area,resulting in the mismatch between the reflectivity curve and the burn severity.In this paper,Forest Reflection and Transmission(FRT)model was selected to retrieve the burn severity.In the modeling process,the canopy parameter "stand density" of the model was used to control the change of tree coverage parameters,and the spectral reflection curves of different burn severity under different tree coverage are simulated respectively;in the backward inversion process,the burning severity are inverted in the look-up table corresponding to the specific tree coverage interval.After confirming the tree coverage of the image element to be inverted by the auxiliary data,the inversion is carried out in the look-up table corresponding to the specific tree coverage interval,so as to effectively improve the correspondence between the burn severity and the canopy reflectance obtained from the remote sensing image,and then improve the precision of the forest fire burn severity inversion.(2)By coupling the multi-layer vegetation radiative transfer model,the ease of use and the scalability of the method are further improved.In view of the shortcomings of FRT model in burn severity inversion,this paper further explored a more simple and universal vegetation radiative transfer model for burn severity inversion.In the previous study,the canopy parameter "stand density" of FRT model was used to control the change of tree coverage to improve the matching of burn severity and spectral reflectance of remote sensing image.However,the tree coverage was not only determined by the stand density.In addition to the stand density,the canopy diameter was also an important parameter affecting the tree coverage.Furthermore,the correlation between stand density and tree coverage is only verified in a few areas,and the empirical relationship established is not universal,so it limits the scalability of FRT Based Inversion of forest burn severity.Moreover,FRT model has many input parameters,too many input parameters will introduce more uncertainty.In view of this,this paper further explores a more simple and universal vegetation radiative transfer model for burn severity inversion.The PROGeo Sail(PROSPECT + Geo Sail)model,which can simulate highly heterogeneous and discontinuous canopy reflectance,and the ACRM model,which can simulate onedimensional homogeneous distribution vegetation canopy reflectance,are coupled.The tree coverage parameter FCOV in Geo Sail model can directly control tree coverage without intermediate conversion parameters,which improves the usability and universality of the method.(3)The relationship between burn severity and environmental driving factors and the importance of various environmental driving factors to burn severity in different scenarios was analyzed.The temporal and spatial distribution characteristics of forest burn severity are closely related to its environmental driving factors before forest fire.It is very important to analyze the mechanism and importance of different environmental driving factors for forest burn severity early warning.Previous studies often focus on the analysis of the relationship between the field measured burn severity samples and environmental driving factors in a single scenario.Limited by the number of field sampling points,it is impossible to cover the whole scenarios,such as steep slopes and hilltops.Moreover,there is a lack of systematic analysis and summary for various typical scenarios.Therefore,based on the inversion of the spatial distribution map of burn severity,the relationship between environmental driving factors and different burn severity of Qinyuan forest fire was analyzed.Moreover,the importance of various environmental driving factors to burn severity in eight typical scenarios was analyzed and summarized.This study will help to further improve the accuracy of burn severity early warning,and provide basis for the accurate decision-making of forest management departments.