Search

Accurate crop-specific damage assessment immediately after flood events is crucial for grain pricing, food policy, and agricultural trade. The main goal of this research is to estimate the crop-specific damage that occurs immediately after flood events by using a newly developed Disaster Vegetation Damage Index (DVDI). By incorporating the DVDI along with information on crop types and flood inundation extents, this research assessed crop damage for three case-study events: Iowa Severe Storms and Flooding (DR 4386), Nebraska Severe Storms and Flooding (DR 4387), and Texas Severe Storms and Flooding (DR 4272). Crop damage is assessed on a qualitative scale and reported at the county level for the selected flood cases in Iowa, Nebraska, and Texas. More than half of flooded corn has experienced no damage, whereas 60% of affected soybean has a higher degree of loss in most of the selected counties in Iowa. Similarly, a total of 350 ha of soybean has moderate to severe damage whereas corn has a negligible impact in Cuming, which is the most affected county in Nebraska. A total of 454 ha of corn are severely damaged in Anderson County, Texas. More than 200 ha of alfalfa have moderate to severe damage in Navarro County, Texas. The results of damage assessment are validated through the NDVI profile and yield loss in percentage. A linear relation is found between DVDI values and crop yield loss. An R2 value of 0.54 indicates the potentiality of DVDI for rapid crop damage estimation. The results also indicate the association between DVDI class and crop yield loss.

Inland aquaculture in Bangladesh has been growing fast in the last decade. The underlying land use/land cover (LULC) change is an important indicator of socioeconomic and food structure change in Bangladesh, and fishpond mapping is essential to understand such LULC change. Previous research often used water indexes (WI), such as Normalized Difference Water Index (NDWI) and Modified Normalized Difference Water Index (MNDWI), to enhance water bodies and use shape-based metrics to assist classification of individual water features, such as coastal aquaculture ponds. However, inland fishponds in Bangladesh are generally extremely small, and little research has investigated mapping of such small water objects without high-resolution images. Thus, this research aimed to bridge the knowledge gap by developing and evaluating an automatic fishpond mapping workflow with Sentinel-2 images that is implemented on Google Earth Engine (GEE) platform. The workflow mainly includes two steps: (1) the spectral filtering phase that uses a pixel selection technique and an image segmentation method to automatically identify all-year-inundated water bodies and (2) spatial filtering phase to further classify all-year-inundated water bodies into fishponds and non-fishponds using object-based features (OBF). To evaluate the performance of the workflow, we conducted a case study in the Singra Upazila of Bangladesh, and our method can efficiently map inland fishponds with a precision score of 0.788. Our results also show that the pixel selection technique is essential in identifying inland fishponds that are generally small. As the workflow is implemented on GEE, it can be conveniently applied to other regions.

Cropland Data Layer (CDL) is an annual crop-specific land use map produced by the U.S. Department of Agricultural (USDA) National Agricultural Statistics Service (NASS). The CDL products are officially hosted on CropScape website which provides capabilities of geospatial data visualization, retrieval, processing, and statistics based on the open geospatial Web services. This study utilizes cloud computing technology to improve the performance of CropScape application and Web services. A cloud-based prototype of CropScape is implemented and tested. The experiment results show the performance of CropScape is significantly improved in the cloud environment. Comparing with the original system architecture of CropScape, the cloud-based architecture provides a more flexible and effective environment for the dissemination of CDL data.

Land use and land cover (LULC) classification using satellite images is an important approach to monitor changes on earth. To produce LULC maps, supervised classification methods are often used. For many supervised classification algorithms, independence of features is an implied assumption. However, this assumption is rarely tested. For LULC classification, using all bands as input features to models is the default approach. However, some of the bands may be highly correlated, which may cause model performances unstable. In this research, correlations and multicollinearity among multi-spectral bands are analyzed for four major LULC types, i.e. cropland, forest, developed area and water bodies. Guided by the correlation analysis, different band combinations were used to train Support Vector Machines (SVM) for four-class LULC classification and the results were compared. From our experiments, band 4, 5, 6 is the best three-band combination and band 1, 2, 5, 7 is the best four-band combination which achieved almost identical performance as using all bands for LULC classification.

WaterSmart project is an NSF funded projected seeks water consumption reduction using satellite observations. In order to fit the fine temporal resolution requirement, satellites are required to have a high revisit cycle. MODIS is an ideal platform for monitoring the ground thanks to its daily coverage while the spatial resolution is too coarse. Research has demonstrated the possibility to improve the spatial resolution of MODIS using the Landsat 8 images. This research is aimed to establish a workflow to adapt the data fusion algorithm to achieve automatically processing at real-time in order to support short-term decision making.

Bangladesh is one of the most vulnerable countries to sea level rise due to the climate change. Soil salinity is one of the potential threat to the coastal ecosystem and agriculture which might hinder the country's future food security. Conventional field-based soil salinity monitoring over vast region may not be cost and time efficient. Satellite remote sensing is offering an efficient way to monitor soil salinity via different indices. This study monitors soil salinity in five years interval from 1990 to 2015 using a regression equation which is developed and tested in the same geographical area to calculate soil salinity from Landsat images. The result shows that soil salinity increases significantly in most of the part of the study area in 25 years period. The percentage of very high saline soil is increased from less than 1% in 1990 to 33% in 2015 which is very alarming for the agriculture growth in this area.

Land use and land cover maps are essential to study how the earth surface change over time and how human activities interact with environments. The growing amount of available remote sensing images, especially the well archived Landsat images with 30 meters resolution, have been used to conduct supervised classification for land use and land cover maps. However, to achieve high classification accuracy, ground truth samples with fine quality and large quantity are required. Collecting ground truth samples is both time-consuming and expensive and sometimes even unviable when ground truth samples are needed for the past years. In this paper, we provided a way of using the GlobeLand30 (GLC30) 2000 and 2010 products as ground truth instead of manually labeling ground truth samples to produce land use and land cover maps for 2005 in our study area, Bangladesh country on Google Earth Engine (GEE) platform. The accuracy assessment is conducted on randomly generated samples from GlobeLand30 products, and the overall accuracy is around 84.8%.