Rice is the most important food crop in Taiwan. Early information on rice-growing conditions is thus vital for estimating rice production to guarantee national food security and grain exports. The rice-harvested area is conventionally inspected twice a year by costly interpretation of aerial photographs and intensive labor-field surveys. However, such methods of rice monitoring are inadequate for providing the government with timely information on rice-cultivated conditions. This study aims to use time series of Sentinel-1 synthetic aperture radar and Sentinel-2 multispectral data to develop a machine-learning approach for the early prediction of rice-growing fields in Taiwan. An object-based random forest (OBRF) was developed to process remotely sensed data for rice-cropping seasons from 2018 to 2021. The prediction results compared with the reference data showed that rice-growing fields could be accurately predicted before harvest, about three months for the first crop and two months for the second crop. The F-score and Kappa coefficient values achieved for the first crop were 0.87 and 0.85, and those for the second crop were 0.72 and 0.71, respectively. These findings were reaffirmed by close agreement between the official statistics and the rice area estimated from the satellite data, with the correlation coefficient of determination (R2) value greater than 0.96. A large portion of the first crop’s rice areas was abandoned or converted to upland crop cultivation in the second crop, which was confirmed by a visual interpretation of Landsat images and official statistics. Ultimately, this study proved the efficacy of using Sentinel-1/2 images and OBRF for the early prediction of rice-cultivated fields in Taiwan. Quantitative and geographical information produced from such methods was essential for the early estimation of rice production to nationally address food security concerns.
The tsunami that occurred on March 11, 2011, in Japan caused widespread devastation of infrastructure and rice cultivation areas. This study assessed the rice growing areas damaged due to this event and the restoration progress using moderate resolution imaging spectroradiometer data. The data were processed for 2010 to 2013, comprising four main steps: (1) data preprocessing to construct the smooth time-series enhanced vegetation index 2 data, (2) rice crop mapping using the dynamic time warping algorithm, (3) accuracy assessment, and (4) change analysis of rice cultivation areas. The mapping results validated using the ground reference data indicated overall accuracies and kappa coefficients generally >84.3% and 0.69, respectively. The results, also verified with the government’s rice area statistics, reaffirmed a close correlation between these two datasets (R2>0.77). When examining changes of rice cultivation between 2010 and 2013, the rice area damaged by the tsunami during 2010 to 2011 was ∼17,550 ha. The rice area restored after one year (2012) was ∼6663 ha, but this regressed slightly in 2013 (5656 ha). Such information could be used by officials to better understand the tsunami-damaged rice area and the restoration progress for crop management in the study region.
Flood is one of the most devastating and frequent disasters. Information on spatiotemporal flood dynamics is essential for planners to devise successful strategies for flood monitoring and mitigation of its negative effects. This study aimed to develop an approach for weekly monitoring of floods with the moderate resolution imaging spectroradiometer (MODIS) data in the Mekong River Delta, South Vietnam, using the water fraction model (WFM). We processed the data for 2009 and 2010 through three main steps: (1) data preprocessing to construct a smooth time series of the difference in the values between land surface water index and enhanced vegetation index, (2) flood derivation using WFM, and (3) accuracy assessment. The results compared to the ground reference data indicated satisfactory results with the overall accuracies and Kappa coefficients of 81.1% and 0.62 for 2009 and 80.3% and 0.61 for 2010, respectively. These results were reaffirmed by a close correlation between the MODIS-derived flood area and that of the ground reference map at the provincial level, with the correlation coefficients (R2) of 0.88 for 2009 and 0.83 for 2010. The results also confirmed the earlier arrival and greater intensity of floods in 2009 compared to 2010.
Rapid urbanization in Ho Chi Minh City (HCMC), Vietnam, is creating societal impacts on the environment attributed to the increasing population. Understanding spatio-temporal dimensions of land-use changes that shape the urbanization is thus critical to the process of urban planning. We explore the urban growth in HCMC through Landsat images for 1990, 2002, and 2010 using the linear mixture model (LMM). The data are processed through four steps: (1) data pre-processing, (2) image classification by LMM using endmembers extracted from the original image using minimum noise fraction, (3) accuracy assessment of the classification results using field verification data, and (4) urban growth analysis to understand the spatial changes of land cover. The results achieved by comparisons between the classification results and ground reference data indicate that the overall accuracy and Kappa coefficient obtained for 1990 were 87.1% and 0.83, respectively, while those for 2002 were 92.5% and 0.89, and those for 2010 were 89.6% and 0.86. The results of urban growth analysis indicate that high albedo class (i.e., built-up areas) expanded from 12.3% in 1990 to 27.2% in 2002 and to 31.1% in 2010. When investigating land-cover conversions to high albedo class from 1990 to 2002, the largest conversion is observed for soil class (9.2%), followed by vegetation class (7.2%), and low albedo class (2.2%). From 2002 to 2010, 4.5% area of soil class was converted to high albedo class, while conversions from vegetation and low albedo classes were 3.5% and 2.5%, respectively.
Rice is a major economic crop in the Lower Mekong Subregion (LMS). Information on rice growing areas is thus vital for
agricultural planning for the sake of food security. This study aimed to map rice cropping systems in LMS from timeseries
MODIS NDVI data for 2010. We processed the time-series NDVI data using wavelet transform and crosscorrelation.
The classification results were assessed using ground verification data. The results indicated that smooth
NDVI profiles derived from wavelet transform reflected the temporal characteristics of rice crop phelonogy under different
rice cropping systems, enabling us to select proper training patterns used in cross-correlation based classifier. The
comparison between the classification map and ground verification data revealed that the results achieved from this
classification approach was promising for regional mapping of rice cropping patterns. The overall accuracy and Kappa
coefficient were 79.5 and 0.72 respectively.
Agro-drought usually refers to the shortage of water for crop irrigation in a short period, creating serve impacts on crop
production due to insufficient soil moisture. This phenomenon has been considered as a challenge for rice farmers in the
Lower Mekong Basin (LMB), especially in the dry season (from November to April). Thus, information on agro-drought
is important for water scheduling to mitigate adverse impacts on rice production. The main objective of this study is to
investigate the applicability of the monthly MODIS normalized difference vegetation index (NDVI) and land surface
temperature (LST) data for drought monitoring from 2008 to 2010. The data was processed for the dry season because this
period is usually suffered from droughts. A simple temperature vegetation difference index (TVDI) was used to estimate
the surface soil moisture content. We investigated the sensitivity between the preliminary TVDI results and TRMM
(Tropical Rainfall Measuring Mission) precipitation. The results revealed good agreement between the two datasets. TVDI
was declined during or after rain events indicating greater soil moisture content, but increased again later indicating less
soil moisture content. The results by analysis of TVDI showed that the moderate and serve droughts were spatially
scattered over the region from November to March and returned to normal condition by the end of the dry season (April)
with the onset of rainy season. The drought was found more serve and extensive in plains of Thailand and Cambodia. The
larger area of serve drought was especially observed for the 2008-2009 dry seasons compared to 2010. The results
achieved from this study could be useful for crop irrigation scheduling.
Rice is the most important economic crop in Vietnam's Mekong Delta (MD). It is the main source of employment and income for rural people in this region. Yearly estimates of rice growing areas and delineation of spatial distribution of rice crops are needed to devise agricultural economic plans and to ensure security of the food supply. The main objective of this study is to map rice cropping systems with respect to monitoring agricultural practices in the MD using time-series moderate resolution imaging spectroradiometer (MODIS) normalized difference vegetation index (NDVI) 250-m data. These time-series NDVI data were derived from the 8-day MODIS 250-m data acquired in 2008. Various spatial and nonspatial data were also used for accuracy verification. The method used in this study consists of the following three main steps: 1. filtering noise from the time-series NDVI data using wavelet transformation (Coiflet 4); 2. classification of rice cropping systems using parametric and nonparametric classification algorithms: the maximum likelihood classifier (MLC) and support vector machines (SVMs); and 3. verification of classification results using ground truth data and government rice statistics. Good results can be found using wavelet transformation for cleaning rice signatures. The results of classification accuracy assessment showed that the SVMs outperformed the MLC. The overall accuracy and Kappa coefficient achieved by the SVMs were 89.7% and 0.86, respectively, while those achieved by the MLC were 76.2% and 0.68, respectively. Comparison of the MODIS-derived areas obtained by the SVMs with the government rice statistics at the provincial level also demonstrated that the results achieved by the SVMs (R2 = 0.95) were better than the MLC (R2 = 0.91). This study demonstrates the effectiveness of using a nonparametric classification algorithm (SVMs) and time-series MODIS NVDI data for rice crop mapping in the Vietnamese MD.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.