Lingjun Kang

Implementation of Geospatial Data Provenance in a Web Service Workflow Environment With ISO 19115 and ISO 19115-2 Lineage Model

Data provenance, also called data lineage, records the derivation history of a data product. In the earth science domain, geospatial data provenance is important because it plays a significant role in data quality and usability evaluation, data trail audition, workflow replication, and product reproducibility. The generation of the geospatial provenance metadata is usually coupled with the execution of geo-processing workflow. Their symbiotic relationship makes them complementary to each other and promises great benefit once they are integrated. However, the heterogeneity of data and computing resources in the distributed environment constructed under the service-oriented architecture (SOA) brings a great challenge to resource integration. Specifically, the issues, such as the lack of interoperability and compatibility among provenance metadata models and between provenance and workflow, create obstacles for the integration of provenance, and geo-processing workflow. In order to tackle these issues, on one hand, this paper breaks the provenance heterogeneity through recording provenance information in a standard lineage model defined in ISO 19115:2003 and ISO 19115-2:2009 standards. On the other hand, this paper bridges the gap between provenance and geo-processing workflow through extending both workflow language and service interface, making it possible for the automatic capture of provenance information in the geospatial web service environment. The proposed method is implemented in the GeoBrain, a SOA-based geospatial web service system. The testing result from implementation shows that the geospatial provenance information is successfully captured throughout the life cycle of geo-processing workflows and properly recorded in the ISO standard lineage model.

Regression model to estimate flood impact on corn yield using MODIS NDVI and USDA cropland data layer

Abstract Flood events and their impact on crops are extremely significant scientific research issues; however, flood monitoring is an exceedingly complicated process. Flood damages on crops are directly related to yield change, which requires accurate assessment to quantify the damages. Various remote sensing products and indices have been used in the past for this purpose. This paper utilizes the moderate resolution imaging spectroradiometer (MODIS) weekly normalized difference vegetation index (NDVI) product to detect and further quantify flood damages on corn within the major corn producing states in the Midwest region of the US. County-level analyses were performed by taking weighted average of all pure corn pixels (>90%) masked by the United States Department of Agriculture (USDA) Cropland Data Layer (CDL). The NDVI-based time-series difference between flood years and normal year (median of years 2000–2014) was used to detect flood occurrences. To further measure the impact of the flood on corn yield, regression analysis between change in NDVI and change in corn yield as independent and dependent variables respectively was performed for 30 different flooding events within growing seasons of the corn. With the R2 value of 0.85, the model indicates statistically significant linear relation between the NDVI and corn yield. Testing the predictability of the model with 10 new cases, the average relative error of the model was only 4.47%. Furthermore, small error (4.8%) of leave-one-out cross validation (LOOCV) along with smaller statistical error indicators (root mean square error (RMSE), mean absolute error (MAE), and mean bias error (MBE)), further validated the accuracy of the model. Utilizing the linear regression approach, change in NDVI during the growing season of corn appeared to be a good indicator to quantify the yield loss due to flood. Additionally, with the 250 m MODIS-based NDVI, these yield losses can be estimated up to field level.

Investigating metrics of geospatial web services: The case of a CEOS federated catalog service for earth observation data

Abstract Geospatial Web Services (GWS) make geospatial information and computing resources discoverable and accessible over the Web. Among them, Open Geospatial Consortium (OGC) standards-compliant data, catalog and processing services are most popular, and have been widely adopted and leveraged in geospatial research and applications. The GWS metrics, such as visit count, average processing time, and user distribution, are important to evaluate their overall performance and impacts. However, these metrics, especially of federated catalog service, have not been systematically evaluated and reported to relevant stakeholders from the point of view of service providers. Taking an integrated catalog service for earth observation data as an example, this paper describes metrics information retrieval, organization, and representation of a catalog service federation. An extensible and efficient log file analyzer is implemented to retrieve a variety of service metrics from the log file and store analysis results in an easily programmable format. An Ajax powered Web portal is built to provide stakeholders, sponsors, developers, partners, and other types of users with specific and relevant insights into metrics information in an interactive and informative form. The deployed system has provided useful information for periodical reports, service delivery, and decision support. The proposed measurement strategy and analytics framework can be a guidance to help GWS providers evaluate their services.

Use of Geographically Weighted Regression Model for Exploring Spatial Patterns and Local Factors Behind NDVI-Precipitation Correlation

Normalized Difference Vegetation Index (NDVI)-precipitation correlation has long been studied. In previous studies, the correlation was usually based on global regression model, which assumed such correlation be constant across the space. However, NDVI-precipitation correlation is spatially dependent and affected by local factors (e.g., soil background). In this paper, geographically weighted regression model is utilized to analyze the NDVI-precipitation correlation on three land use types (i.e., 1) grassland, 2) fallow/idle land, and 3) winter wheat land) within U.S. central great plain area. Results suggest that geographically weighted regression model has better performances than global regression models. Specifically, higher average R2 (0.81) and lower proportion (9%) of residuals with spatial autocorrelation has been achieved under geographically weighted regression in comparison with lower average R2 (0.68) and higher proportion (38%) of residual with spatial autocorrelation under global regression models. In addition, the spatially dependent correlation between NDVI and precipitation has been revealed with geographically weighted regression model. From the north to south, the increasing unit rate of NDVIs change with precipitation has been found through spatially varying regression slopes. Moreover, local factors affecting NDVI-precipitation correlation, such as soil permeability and thickness, have been identified through analyzing the local goodness of fitting under geographically weighted regression model. In summary, unveiled spatial patterns of NDVI-precipitation correlation provide another perspective for studying correlations between NDVI and climatic factors. This work should also be helpful to better understand crop responses to precipitation in agricultural management.

Remote sensing based crop growth stage estimation model

Crop growth stages are important factors for segmenting the crop growing seasons and analyzing their growth conditions against normal conditions by periods. Time series of high temporal resolution, up to daily, satellite remotely sensed data are used in establishing crop growth estimation model and estimate the growth stages. The daily surface reflectance data from Moderate Resolution Imaging Spectroradiometer (MODIS) is used as the base data to calculate indices, form condition profiles, construct crop growth model, and estimate crop growth stage. Different crops have different condition profiles. To take into consideration of crop differences, models are built on each crop type. In the United States, ten major crops have been chosen to build crop growth stage estimation models using historical date tracing back to 2000 when MODIS launched. A kernel, double sigmoid model, is used to model the single mode crop growth season. The basic core model is double sigmoid model. The Best Index Slope Extraction (BISE) is applied to pre-filter the daily crop condition index. Estimated results have reasonably high accuracy, with root mean square error less than 10% on the state level evaluation.