Dataset Information

Comparison of Hybrid Classifiers for Crop Classification Using Normalized Difference Vegetation Index Time Series: A Case Study for Major Crops in North Xinjiang, China.

ABSTRACT: A range of single classifiers have been proposed to classify crop types using time series vegetation indices, and hybrid classifiers are used to improve discriminatory power. Traditional fusion rules use the product of multi-single classifiers, but that strategy cannot integrate the classification output of machine learning classifiers. In this research, the performance of two hybrid strategies, multiple voting (M-voting) and probabilistic fusion (P-fusion), for crop classification using NDVI time series were tested with different training sample sizes at both pixel and object levels, and two representative counties in north Xinjiang were selected as study area. The single classifiers employed in this research included Random Forest (RF), Support Vector Machine (SVM), and See 5 (C 5.0). The results indicated that classification performance improved (increased the mean overall accuracy by 5%~10%, and reduced standard deviation of overall accuracy by around 1%) substantially with the training sample number, and when the training sample size was small (50 or 100 training samples), hybrid classifiers substantially outperformed single classifiers with higher mean overall accuracy (1%~2%). However, when abundant training samples (4,000) were employed, single classifiers could achieve good classification accuracy, and all classifiers obtained similar performances. Additionally, although object-based classification did not improve accuracy, it resulted in greater visual appeal, especially in study areas with a heterogeneous cropping pattern.

SUBMITTER: Hao P

PROVIDER: S-EPMC4567051 | biostudies-literature | 2015

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Comparison of Hybrid Classifiers for Crop Classification Using Normalized Difference Vegetation Index Time Series: A Case Study for Major Crops in North Xinjiang, China.

Hao Pengyu P Wang Li L Niu Zheng Z

PloS one 20150911 9

A range of single classifiers have been proposed to classify crop types using time series vegetation indices, and hybrid classifiers are used to improve discriminatory power. Traditional fusion rules use the product of multi-single classifiers, but that strategy cannot integrate the classification output of machine learning classifiers. In this research, the performance of two hybrid strategies, multiple voting (M-voting) and probabilistic fusion (P-fusion), for crop classification using NDVI ti ...[more]

PMID: 26360597

Similar Datasets

Project description:The estimation of long-term groundwater recharge rate ([Formula: see text]) is a pre-requisite for efficient management of groundwater resources, especially for arid and semi-arid regions. Precise estimation of [Formula: see text] is probably the most difficult factor of all measurements in the evaluation of GW resources, particularly in semi-arid regions in which the recharge rate is typically small and/or regions with scarce hydrogeological data. The main objective of this study is to find and assess the predicting factors of [Formula: see text] at an aquifer scale. For this purpose, 325 Iran's phreatic aquifers (61% of Iran's aquifers) were selected based on the data availability and the effect of eight predicting factors were assessed on [Formula: see text] estimation. The predicting factors considered include Normalized Difference Vegetation Index (NDVI), mean annual temperature ([Formula: see text]), the ratio of precipitation to potential evapotranspiration ([Formula: see text]), drainage density ([Formula: see text]), mean annual specific discharge ([Formula: see text]), Mean Slope ([Formula: see text]), Soil Moisture ([Formula: see text]), and population density ([Formula: see text]). The local and global Moran's I index, geographically weighted regression (GWR), and two-step cluster analysis served to support the spatial analysis of the results. The eight predicting factors considered are positively correlated to [Formula: see text] and the NDVI has the greatest influence followed by the [Formula: see text] and [Formula: see text]. In the regression model, NDVI solely explained 71% of the variation in [Formula: see text], while other drivers have only a minor modification (3.6%). The results of this study provide new insight into the complex interrelationship between [Formula: see text] and vegetation density indicated by the NDVI. The findings of this study can help in better estimation of [Formula: see text] especially for the phreatic aquifers that the hydrogeological ground-data requisite for establishing models are scarce.

Project description:Phenology of plants is important for ecological interactions. The timing and development of green leaves, plant maturity, and senescence affects biophysical interactions of plants with the environment. In this study we explored the agreement between land-based camera and satellite-based phenology metrics to quantify plant phenology and phenophases dates in five plant community types characteristic of the semi-arid cold desert region of the Great Basin. Three years of data were analyzed. We calculated the Normalized Difference Vegetation Index (NDVI) for both land-based cameras (i.e., phenocams) and Landsat imagery. NDVI from camera images was calculated by taking a standard RGB (red, green, and blue) image and then a near infrared (NIR) plus RGB image. Phenocam NDVI was calculated by extracting the red digital number (DN) and the NIR DN from images taken a few seconds apart. Landsat has a spatial resolution of 30 m², while phenocam spatial resolution can be analyzed at the single pixel level at the scale of cm² or area averaged regions can be analyzed with scales up to 1 km². For this study, phenocam regions of interest were used that approximated the scale of at least one Landsat pixel. In the tall-statured pinyon and juniper woodland sites, there was a lack of agreement in NDVI between phenocam and Landsat NDVI, even after using National Agricultural Imagery Program (NAIP) imagery to account for fractional coverage of pinyon and juniper versus interspace in the phenocam data. Landsat NDVI appeared to be dominated by the signal from the interspace and was insensitive to subtle changes in the pinyon and juniper tree canopy. However, for short-statured sagebrush shrub and meadow communities, there was good agreement between the phenocam and Landsat NDVI as reflected in high Pearson's correlation coefficients (r > 0.75). Due to greater temporal resolution of the phenocams with images taken daily, versus the 16-day return interval of Landsat, phenocam data provided more utility in determining important phenophase dates: start of season, peak of season, and end of season. More specific species-level information can be obtained with the high temporal resolution of phenocams, but only for a limited number of sites, while Landsat can provide the multi-decadal history and spatial coverage that is unmatched by other platforms. The agreement between Landsat and phenocam NDVI for short-statured plant communities of the Great Basin, shows promise for monitoring landscape and regional-level plant phenology across large areas and time periods, with phenocams providing a more comprehensive understanding of plant phenology at finer spatial scales, and Landsat extending the historical record of observations.

Dataset Information

Comparison of Hybrid Classifiers for Crop Classification Using Normalized Difference Vegetation Index Time Series: A Case Study for Major Crops in North Xinjiang, China.

Publications

Comparison of Hybrid Classifiers for Crop Classification Using Normalized Difference Vegetation Index Time Series: A Case Study for Major Crops in North Xinjiang, China.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets