| The traditional methods to monitor forest fire with remote sensing is mostly completed in local equipment.The storage and calculation capabilities of local equipment are directly related to the time of data processing,which also limits the current research on forest fires in large-scale areas.Google Earth Engine(GEE)is a comprehensive big data platform for scientific analysis of massive remote sensing data and geographic information visualization,on this count,data processing is not restricted by local conditions.This study is committed to fire monitoring based on GEE,and the city of Genhe of Inner Mongolia autonomous region is chosen as research area,proposing a new method for quickly identifying and extracting large-scale regional burned areas,establishing a municipal forest fire historical attribute database,analyzing forest fires intensity distribution pattern,exploring the method of judging fire time accurate to the time period of month,researching different conditions of vegetation restoration after fire evaluation method,so as to providing theoretical support for the local forestry authorities to monitor forest fires and formulate management decisions after fire restoration and data foundation.This study obtained interannual TM composite images of Genhe from 2001 to 2010 based on GEE.The near-infrared band of the image and d NBR were selected as the sensitive characteristics of the over-fire region.Using the OTSU algorithm to calculate the threshold and construct a decision tree classification model to extract the annual fire spots in Genhe.After verifying the accuracy,a city-level forest fire historical attribute database was established to analyze the spatial and temporal distribution of forest fires.Forest fire intensity was classified according to the threshold range of d NBR,and spatial distribution of forest fire intensity was analyzed based on DEM data.MOD09A1 data set was used to obtain the NDVI and NBR exponent time series data of the burned area and the patches in the adjacent vegetation area based on GEE,and the fire time in the burned area was determined by comparative analysis.Invoking Landsat EVI product based on GEE to extract EVI time series data of pixels of forest fire intensity regions in different fire regions and calculate the d EVI by comparing the mean of EVI in the three years before fire.Finally,the effects of different forest fire intensities on post-fire vegetation restoration were analyzed on the time scale,and the restoration years of post-fire vegetation were evaluated.The main conclusions are as follows:?The method of rapid extraction of large-scale fire regions based on GEE is feasible.The overall accuracy of extraction of burned area is 88.54%.The overall accuracy is relatively complete with clear boundaries and reliable results.?Forest fires in Genhe from 2001 to 2010 presented obvious differences in time and space.Burned areas were generally distributed in the northeast region,while the southwest region was less.Among all forest fires,the number of forest fire warnings was least,but it is easy to develop into a larger fire.General forest fires were the most frequent with the highest frequency.There were 14 major forest fires with a total area of 3969.3hm~2.Extraordinary forest fires were few in number,but the area was far more than other levels of forest fires.Of all the years,the total burned area was the largest in 2003 and the smallest in 2006.In 2004 and 2006,there were only 4 forest fires,and in 2008,there were at most 18,but most of them were general forest fires with burning areas ranging from 1 to 100 hm~2.In 2003,the area of the huge forest fire that occurred in the central region was as high as 84343.3hm~2,which was the largest forest fire in Genhe over the years.?The proportion of over-fire area of forest fire intensity at all levels in Genhe City was:Moderate>Severe>Mild.The proportion of Burned area in areas with an altitude of 500?1000 m was much higher than that in areas above 1000 m.The proportion of Burned area in different slopes was generally as follows:gentle slope>flat slope>flat slope>steep slope.The burned area in different slope directions from large to small was:East>West>North>South>Northwest>Southwest>Northeast>Southeast>Flat.In general,areas with an altitude of 500-1000 m,a slope of 6-15°,and the eastern slope were the areas where large-scale forest fires most likely to occur,where can be used as a key forest fire prevention area for the local forestry department.?Combining the timing changes of the NDVI and NBR to determine the fire time of the burned area can effectively avoid the impact of data noise and weather changes,and can determine the fire time without the support of other forest fire data and the determination accuracy can reach the time period of months.After the occurrence of forest fires,NDVI and NBR changed significantly.NDVI index usually shows a downward trend in the first year after fire,while NBR index showed a downward trend in the year of fire.?The overall restoration period of post-fire vegetation index in Genhe was about 10 years,and the restoration of post-fire vegetation was generally shown as follows:it was difficult to renew and restore vegetation after severe fire,and there were many influencing factors for restoration of vegetation after mild and moderate fire.There was no significant relationship between the vegetation restoration in different intensities and the burned area.After fire,EVI showed a significant upward trend accompanied by certain annual fluctuations,and EVI recovery was significantly different in areas with different forest fire intensity.The recovery rate of severe fire was significantly lower than that of mild and moderate fire.It takes at least 2 years and at least 8 years for the moderate burned area to recover to the pre-fire level.The maximum recovery time of the heavy burned area was 4 years,and the minimum recovery time was10 years. |
PDF Full Text Request