Identification and Classification of Complex Agricultural Croplands Using Multi-Temporal ALOS-2/PALSAR-2 Data: A Case Study in Central Java, Indonesia
Identification and Classification of Complex Agricultural Croplands Using Multi-Temporal ALOS-2/PALSAR-2 Data: A Case Study in Central Java, Indonesia
复制标题
DOI：
10.5539/jas.v10n2p58
复制
发表时间：
2018-01
期刊：
影响因子：
--
通讯作者：
P. R. Mirelva;R. Nagasawa
中科院分区：
文献类型：
--
作者：
P. R. Mirelva;R. Nagasawa
农业部门对印度尼西亚经济作出重大贡献，并已成为国民收入的来源之一。因此，精确的农业制图对国家和地区管理部门非常重要。卫星遥感为查明大片农田提供了最有效的工具。然而，热带地区的云层覆盖限制了光学遥感的使用。SAR是一种主动遥感技术，它提供完全无云的观测数据。在这项研究中使用了多时相ALOS-2/PALSAR-2数据，并辅之以多时相光学遥感数据，即Landsat 8 OLI，用于对复杂的农田进行分类。研究区，位于Klaten县，中爪哇省，112平方公里的覆盖范围，被选中，因为它的动态种植模式和复杂的农业土地利用类型。在这项研究中，RGB复合HH，HV和HV-HH，来自ALOS-2/PALSAR-2偏振，被发现是有效的分离两种类型的稻田种植模式：全年水稻（水稻-I）和水稻高地（水稻-II）。还发现多时相Landsat 8数据对观测种植模式也很有用。此外，分类精度，这是高达85.02%的整体精度，kappa系数为0.824，从多时相ALOS-2/PALSAR-2数据，得到。这些结果表明，多时相ALOS-2/PALSAR-2数据能够区分两种不同的稻田种植类型，以及有利于区分的种植阶段和种植模式的信息，为其他几种土地利用。
The agriculture sector makes a significant contribution to the Indonesian economy and has become one of the sources of national income. Therefore, precise agricultural mapping is very important to national and regional administrations. Satellite remote sensing provides the most effective tool for identifying a wide expanse of agriculture croplands. However, cloud coverage in tropical regions limits the use of optical remote sensing. SAR is an active remote sensing technique, which offers completely cloud-free observation data. The multi-temporal ALOS-2/PALSAR-2 data were used in this study, complemented by optical multi-temporal remote sensing data, that is, Landsat 8 OLI for classifying complex agricultural croplands. The study area, located in the Klaten Regency, Central Java Province, with 112 km 2 coverage, was selected because of its dynamic cropping pattern and complex agricultural land use types. In this study, the RGB composite of HH, HV and HV-HH, derived from ALOS-2/PALSAR-2 polarizations, was found to be effective at separating two types of paddy field cropping pattern: all-year paddy (paddy-I) and paddy upland fields (paddy-II). The multi-temporal Landsat 8 data were also found to be useful for observing the cropping pattern. Moreover, the classification accuracy, which was as high as 85.02% in terms of overall accuracy, with a kappa coefficient of 0.824, from multi-temporal ALOS-2/PALSAR-2 data, was obtained. These results show that multi-temporal ALOS-2/PALSAR-2 data are capable of discriminating between two different paddy field cropping types, as well as beneficial for discriminating between the cropping stage and cropping pattern information for several other land uses.