| In recent years,people pay more and more attention to the utilization of farmland,especially in China where there are many people and little land.With the urbanization,industrialization and the adjustment of the rural economic structure,a large amount of farmland has been occupied,especially in the southeast coastal areas of China with a high degree of economic development.Multiple cropping index is a key index to measure the utilization degree of cultivated land.It refers to the ratio of the planting area or the harvesting area to the cultivated land area in the region,and reflects the average situation of the macro region.At present,there have been a lot of studies on the reconstruction of NDVI,EVI and LAI and the extraction of multiple cropping index of cultivated land,but the reconstruction effect of each method on remote sensing data still needs to be further discussed,and the accuracy of multiple cropping index extracted based on data reconstruction still needs to be verified.Guangdong province is the key area for the distribution of high quality cultivated land in China,especially paddy fields.Monitoring the change of paddy field and planting situation in Guangdong province has important academic and practical value for regional food security,adjustment of farmland utilization structure,farmland protection and ecological environment.Based on the EVI data of MOD13Q1 and its quality control data set,this study used timesat software and IDL programming language to realize 8 reconstruction methods,including DL,DLWeight,AG,AGWeight,SG,SG2,SGWeight and STSG.The reconstruction effects of different algorithms were compared qualitatively and quantitatively to establish the reconstruction data set of MODIS-EVI.The reconstruction method was used to extract the multiple cropping index of paddy field and verify its accuracy.The method with the best reconstruction effect and the highest verification accuracy was selected to construct the spatial and temporal database of the multiple cropping index of paddy field in Guangdong province.Then the spatial and temporal differences of multiple cropping index of paddy fields in Guangdong province were analyzed from the scales of province,county and pixel.The main conclusions are as follows:(1)The data quality of MOD13Q1 has obvious seasonal characteristics under the influence of meteorological conditions.The data quality of spring and summer is poor,while that of autumn and winter is relatively good.MODIS-EVI low quality pixel frequency is mainly distributed in the Guangdong-Hong Kong-Macao greater bay area,Chaoshan area,Shaoguan city,Heyuan city and its surrounding areas.The region with better data quality is located in eastern and central regions of western Guangdong.(2)Characteristics of 8 time series data reconstruction methodsIn terms of qualitative evaluation,the curves of the reconstruction effect of EVI by DL and AG are basically similar,the curves are relatively smooth and have a better effect on vegetation reconstruction with only one phenological period,and the reconstruction curves of paddy fields with increased weights by DLWeight and AGWeight are also similar,and the results of AG are slightly better than that of DL.SG retained a large number of original data features,because SG2 repeated SG filtering twice,so the difference between SG2 and SG is small,but SG2 curve is smoother than SG,the peak value is slightly smaller,the trough value is slightly larger.SGWeight adds quality control data as weights,reduces the impact of low quality pixels,retains key details.After removing a large number of noise values,the STSG is the closest to the original EVI curve,and compared with the other 7 methods,it can better deal with the low value of local trough.STSG had better effects on paddy field and dry land reconstruction,DLWeight had better effects on forest land reconstruction,and AGWeight had better effects on grassland reconstruction.In terms of quantitative evaluation,there is no significant difference in the spatial distribution of fidelity evaluation between AG and DL,SG is better than the previous two,and STSG is significantly different in the spatial distribution of fidelity evaluation.RMSE on artificial surface is the largest among all methods,and low quality pixels in continuous period have a great impact on STSG method.Reduced fidelity of AGWeight,DLWeight and SGWeight.SG,SG2 and SGWeight had better performance evaluation results,followed by AG,AGWeight,DL and DLWeight.Due to the quality of the original data,the evaluation result of STSG is relatively poor,but it is optimal for the reconstruction of cultivated land.(3)Combined with the spatial and temporal information of the data,the STSG truly restored the EVI value of paddy fields.Therefore,this study used the STSG method to reconstruct the EVI time series data to extract the multiple cropping index of paddy fields in Guangdong province.In the past 19 years,the average annual multiple cropping index of paddy field in Guangdong province showed a downward trend,and the interannual variation was large.The areas with an average annual multiple cropping index of paddy fields less than 100%are mainly distributed in the vicinity of the Guangdong-Hong KongMacao greater bay area,the Chaoshan area and the surrounding areas of each city.The 100%-150%area are mainly distributed in the north of Qingyuan city and Shaoguan city.This area is a mountainous area with poor topographic conditions and slightly low utilization efficiency of paddy fields.In addition,there are a small number of 100%-150%and less than 100%pixels in Zhanjiang city,which may be due to the frequent drought,typhoons and other meteorological disasters affecting the rice planting in this area.The multiple cropping index of paddy field in contiguous area is 150%-200%.For example,Nanxiong city,Maoming city,Leizhou city,Xingning city and so on.The areas with annual average multiple cropping index of paddy fields greater than 200%are mainly located in jieyang city,chaozhou city,lianjiang city and wengyuan county. |
PDF Full Text Request