Cropland Data Layer Research Articles

Validation and refinement of cropland data layer using a spatial-temporal decision tree algorithm

Space-based crop identification and acreage estimation have played a significant role in agricultural studies in recent years, due to the development of Remote Sensing technology. The Cropland Data Layer (CDL), which was developed by the U.S. Department of Agriculture (USDA), has been widely used in agricultural studies and achieved massive success in recent years. Although the CDL’s accuracy assessments report high overall accuracy on various crops classifications, misclassification is still common and easy to discern from visual inspection. This study is aimed to identify and resolve inaccurate crop classification in CDL. A decision tree method was employed to find questionable pixels and refine them with spatial and temporal crop information. The refined data was then evaluated with high-resolution satellite images and official acreage estimates from USDA. Two validation experiments were also developed to examine the data at both the pixel and county level. Data generated from this research was published online in two repositories, while both applications allow users to download the entire dataset at no cost.

Improved identification of cotton cultivated areas by applying instance-based transfer learning on the time series of MODIS NDVI

Cotton is an important cash crop and strategic material in the world, as the main source of natural and renewable fiber for textiles. Accurate and timely cotton distribution maps are crucial for monitoring and managing cotton cultivation system. Remotely sensed data have been widely used in cropland mapping, whereas relatively less attention has been paid specifically to cotton mapping partly due to the difficulty in obtaining the training samples over large regions. To resolve this issue, this study introduced the instance-based transfer learning to identify the cotton cultivated areas using the remotely sensed images. The annual time series of normalized difference vegetation index (NDVI) derived from Moderate Resolution Imaging Spectroradiometer (MODIS) images was adopted as input features. The random forest (RF) and long short-term memory (LSTM) algorithms integrated with the Transfer AdaBoost (TrAdaBoost) algorithm were adopted to generate the RF-based and LSTM-based TrAdaBoost approaches, respectively. The experiments were conducted in Arkansas State of the United States, where the cropland data layer (CDL) was available and utilized as a source of auxiliary data. The results showed that both the RF-based and LSTM-based TrAdaBoost performed better than the original RF and LSTM under the condition that the training samples in the target domain were limited. The advantages of transfer learning were much greater when the percentage of training samples from the source domain equals 80%. The two transfer learning approaches were then applied to identify the cotton cultivated areas in Uzbekistan. The cotton areas detected by MODIS images in 2018 were agreed well with the statistics at the sub-national level, with the R2 values of 0.57 and 0.64, respectively. These results demonstrate the potential of the RF-based and LSTM-based TrAdaBoost approaches in generating the cotton distribution maps when there are few samples in the target domain.

Rice Mapping in Training Sample Shortage Regions Using a Deep Semantic Segmentation Model Trained on Pseudo-Labels

A deep semantic segmentation model-based method can achieve state-of-the-art accuracy and high computational efficiency in large-scale crop mapping. However, the model cannot be widely used in actual large-scale crop mapping applications, mainly because the annotation of ground truth data for deep semantic segmentation model training is time-consuming. At the operational level, it is extremely difficult to obtain a large amount of ground reference data by photointerpretation for the model training. Consequently, in order to solve this problem, this study introduces a workflow that aims to extract rice distribution information in training sample shortage regions, using a deep semantic segmentation model (i.e., U-Net) trained on pseudo-labels. Based on the time series Sentinel-1 images, Cropland Data Layer (CDL) and U-Net model, the optimal multi-temporal datasets for rice mapping were summarized, using the global search method. Then, based on the optimal multi-temporal datasets, the proposed workflow (a combination of K-Means and random forest) was directly used to extract the rice-distribution information of Jiangsu (i.e., the K–RF pseudo-labels). For comparison, the optimal well-trained U-Net model acquired from Arkansas (i.e., the transfer model) was also transferred to Jiangsu to extract local rice-distribution information (i.e., the TF pseudo-labels). Finally, the pseudo-labels with high confidences generated from the two methods were further used to retrain the U-Net models, which were suitable for rice mapping in Jiangsu. For different rice planting pattern regions of Jiangsu, the final results showed that, compared with the U-Net model trained on the TF pseudo-labels, the rice area extraction errors of pseudo-labels could be further reduced by using the U-Net model trained on the K–RF pseudo-labels. In addition, compared with the existing rule-based rice mapping methods, he U-Net model trained on the K–RF pseudo-labels could robustly extract the spatial distribution information of rice. Generally, this study could provide new options for applying a deep semantic segmentation model to training sample shortage regions.

Probabilistic co-occurrence assessment for suites of listed species.

Section 7 of the Endangered Species Act requires the US Environmental Protection Agency (US EPA) to consult with the Services (US Fish and Wildlife Service and National Marine Fisheries Service) over potential pesticide impacts on federally listed species. Consultation is complicated by the large number of pesticide products and listed species, as well as by lack of consensus on best practices for conducting co-occurrence analyses. Previous work demonstrates that probabilistic estimates of species' ranges and pesticide use patterns improve these analyses. Here we demonstrate that such estimates can be made for suites of sympatric listed species. Focusing on two watersheds, one in Iowa and the other in Mississippi, we obtained distribution records for 13 species of terrestrial and aquatic listed plants and animals occurring therein. We used maximum entropy modeling and bioclimatic, topographic, hydrographic, and land cover variables to predict species' ranges at high spatial resolution. We constructed probabilistic spatial models of use areas for two pesticides based on the US Department of Agriculture Cropland Data Layer and reduced classification errors by incorporating information on the relationships between individual pixels and their neighbors using object-based images analysis. We then combined species distribution and crop footprint models to derive overall probability of co-occurrence of listed species and pesticide use. For aquatic species, we also integrated an estimate of downstream residue transport. We report each separate species-by-use-area co-occurrence estimate and also combine these modeled co-occurrence probabilities across species within watersheds to produce an overall metric of potential pesticide exposure risk for these listed species at the watershed level. We propose that the consultation process between US EPA and the Services be based on such batched estimation of probabilistic co-occurrence for multiple listed species at a regional scale. Integr Environ Assess Manag 2022;18:1088-1100. © 2021 SETAC.

Early Classification Method for US Corn and Soybean by Incorporating MODIS-Estimated Phenological Data and Historical Classification Maps in Random-Forest Regression Algorithm

An early crop classification method is functionally required in a near-real-time crop-yield prediction system, especially for upland crops. This study proposes methods to estimate the mixed-pixel ratio of corn, soybean, and other classes within a low-resolution MODIS pixel by coupling MODIS-derived crop phenology information and the past Cropland Data Layer in a random-forest regression algorithm. Verification of the classification accuracy was conducted for the Midwestern United States. The following conclusions are drawn: The use of the random-forest algorithm is effective in estimating the mixed-pixel ratio, which leads to stable classification accuracy; the fusion of historical data and MODIS-derived crop phenology information provides much better crop classification accuracy than when these are used individually; and the input of a longer MODIS data period can improve classification accuracy, especially after day of year 279, because of improved estimation accuracy for the soybean emergence date.

A novel automatic phenology learning (APL) method of training sample selection using multiple datasets for time-series land cover mapping

The long record of Landsat imagery, which is the cornerstone of Earth observation, provides an opportunity to monitor land use and land cover (LULC) change and understand the interactions between the climate and earth system through time. A few change detection algorithms such as Continuous Change Detection and Classification (CCDC) have been developed to utilize all available Landsat images for change detection and characterization at local or global scales. However, the reliable, rapid, and reproducible collection of training samples have become a challenge for time series land cover classification at a large scale. To meet the challenge, we proposed an automatic phenology learning (APL) method with the assumption that the temporal profiles of samples within the same land cover type are the same or similar at a local scale to generate evenly distributed training samples automatically. We designed the method to build land cover patterns for each category based on consensus samples derived from multiple existing scientific datasets including LANDFIRE's (LF) Existing Vegetation Type (EVT), USGS National Land Cover Database (NLCD), National Agricultural Statistics Service (NASS) Cropland Data Layer (CDL), and National Wetlands Inventory (NWI). Then we calculated the Time-Weighted Dynamic Time Warping (twDTW) distance between any undefined samples and land cover patterns in the same geographical region as prior knowledge. Finally, we selected the optimal land cover category for each undefined sample from the land cover products based on the designed criteria iteratively using the twDTW distance as an indicator. The method was applied in the footprint of 10 selected Landsat Analysis Ready Data (ARD) tiles in the eastern and western conterminous United States (CONUS) to produce annual land cover maps from 1985 to 2017. The accuracy assessment and visual comparison revealed that the APL method can generate reliable training samples without any manual interpretation, producing better land cover results especially for the grass/shrub and wetland land cover classes. Applying the APL method, the overall accuracy of the annual land cover maps was improved by 2% over the accuracy of Land Change Monitoring, Assessment, and Projection (LCMAP) Collection 1.0 Science Products in the research regions. Our results also indicate that the APL method provides an approach for best use of different land cover products and meets the requirement of intensive sampling for training data collection.

Examining patterns and drivers of variability in playa water status on the High Plains of western Kansas, 2016–2019

Playa wetlands are widely distributed across the High Plains of the central United States, providing a range of ecosystem services, such as groundwater recharge, surface water storage, and wetland habitat. Although playas are essential resources, few studies have examined the variability and controls on playa water storage. The purpose of this project is to determine how playa and watershed morphology, watershed land cover, and precipitation patterns affect timing and duration of water storage in playas. This project focuses on 92 playas distributed throughout a 10-county region in western Kansas.
 Playa and watershed morphology were calculated in a GIS environment and classified into quartiles based on playa and watershed surface area. Watershed tilled index (i.e., percent cropland versus grassland) was determined using 2016, 2017, 2018, and 2019 Cropland Data Layers available from the National Agricultural Statistics Service and classified as either cropland (more than 75% cropland), grassland (more than 75% grassland), or mixed. Monthly precipitation data for 2016–2019 were compiled from the Oakley 22S High Plains Regional Climate Center weather station. Playa water status for 2016–2019 was classified monthly as either standing water or dry (i.e., no visible standing water) by visually examining four-band satellite imagery with 3.7 m resolution available from Planet Explorer (www.planet.com).
 Playa water status is influenced by a combination of factors, including playa and watershed morphology, watershed land cover, and precipitation patterns. Larger playas have larger watersheds and standing water more frequently and for longer periods than smaller playas. Playas in cropland watersheds store water more frequently and for longer periods than playas in grassland watersheds, though differences are not statistically significant. Standing water within playas is positively correlated with monthly precipitation and reflects a short-term response to precipitation patterns, regardless of playa size or watershed land cover. The strongest controls on playa water status are playa area, monthly precipitation, and watershed area.
 Playas are critical resources for the High Plains, providing a range of ecosystem services that are dependent upon the playa’s ability to store water. Playa functions are under continued threat from cropland expansion, climate change, and playa and watershed modifications. To sustain playa functions in Kansas, efforts should focus on conserving larger grassland playas and reducing sediment inputs to playas in cropland watersheds.

Transferable deep learning model based on the phenological matching principle for mapping crop extent

Accurate and timely crop mapping is essential for global food security assessments; however, conventional crop mapping models are usually applicable to specific spatial or temporal scales, i.e. “one-time, one-place” model. Moreover, the extensive application of a trained model to other regions is challenging when sufficient ground-truth samples used for training process are unavailable. This study exploited Cropland Data Layer (CDL) and Landsat data for Arkansas, United States (US), to train a U-Net model and then extensively tested the generalization ability of the model in the Corn Belt and California in the US, and even further tested in Liaoning, China, on a transcontinental scale. Two temporal images were generated by compositing the median values of images obtained during two crop growing time windows representing the sowing (from March to May) and vigorous growth periods (from June to August). In order to ensure the consistency of the data distribution between the target areas (testing areas) and the training area, we shifted the time windows of the target areas to match that of the training area following the phenological matching principle. Then we can composite the target data (testing data) according to the time windows matched in the target areas. The results showed a satisfactory accuracy. The average optimal Overall Accuracy of crop mapping in all the target areas exceeded 87%. The average optimal F1-score of corn and rice was 0.79. Finally, we compared the generalization performance of U-Net and Random Forest (RF) classifiers. The results showed that U-Net performed better in all the target areas in the US while RF performed better in target areas where the plots were smaller. The procedure and strategy developed will facilitate the realization of high-performance and automated global model transfer.

Assessing Marginal Land Availability Based on Land Use Change Information in the Contiguous United States.

Utilization of marginal land for growing dedicated bioenergy crops for second-generation biofuels is appealing to avoid conflicts with food production. This study develops a novel framework to quantify marginal land for the Contiguous United States (CONUS) based on a history of satellite-observed land use change (LUC) over the 2008-2015 period. Frequent LUC between crop and noncrop is assumed to be an indicator of economically marginal land; this land is also likely to have a lower opportunity cost of conversion from food crop to bioenergy crop production. We first present an approach to identify cropland in transition using the time series of Cropland Data Layer (CDL) land cover product and determine the amount of land that can be considered marginal with a high degree of confidence vs with uncertainty across the CONUS. We find that the biophysical characteristics of this land and its productivity and environmental vulnerability vary across the land and lie in between that of permanent cropland and permanent natural vegetation/bare areas; this land also has relatively low intrinsic value and agricultural profit but a high financial burden and economic risk. We find that the total area of marginal land with confidence vs with uncertainty is 10.2 and 58.4 million hectares, respectively, and mainly located along the 100th meridian. Only a portion of this marginal land (1.4-2.2 million hectares with confidence and 14.8-19.4 million hectares with uncertainty) is in the rainfed region and not in crop production and, thus, suitable for producing energy crops without diverting land from food crops in 2016. These estimates are much smaller than the estimates obtained by previous studies, which consider all biophysically low-quality land to be marginal without considering economical marginality. The estimate of marginal land for bioenergy crops obtained in this study is an indicator of the availability of economically marginal land that is suitable for bioenergy crop production; whether this land is actually converted to bioenergy crops will depend on the market conditions. We note the inability to conduct field-level validation of cropland in transition and leave it to future advances in technology to ground-truth land use change and its relationship to economically marginal land.

Towards interpreting multi-temporal deep learning models in crop mapping

Multi-temporal deep learning approaches have exhibited excellent classification performance in large-scale crop mapping. These approaches efficiently and automatically transform remote sensing time series into high-dimensional feature representations to identify crop types. The lack of interpretation, however, is regarded as a major drawback of these high-performance approaches. Interpreting deep learning approaches in multi-temporal crop mapping is critical for verifying their reliability. This study aims to quantify the impact of multi-temporal information in input time series on classification performance and develop a multi-perspective interpretation pipeline for deep learning models. The pipeline involves three interpretation approaches: evaluating input feature importance, analyzing hidden features, and monitoring temporal changes in model's soft output. An experiment is conducted to classify corn and soybean in the U.S corn belt in 2018. The study area consists of three sites each encompassing millions of pixel-level samples at 30 m resolution. The Landsat Analysis Ready Data are used as the input remote sensing time series and Cropland Data Layer is used as the ground reference. Attention-based Long Short-Term Memory (AtLSTM) and Transformer models are built as multi-temporal deep learning models, and compared to Random Forest (RF). Complete time series input in the correct order achieves a higher overall accuracy of 97.8% than using single-window or out-of-order inputs, indicating multi-temporal information facilitates crop classification. An assessment of the input feature importance demonstrates that the AtLSTM, Transformer, and RF models all consider the period from weeks 11 to 20 (early-July to late-August) as a key growth period and the shortwave infrared band as the critical band for corn and soybean discrimination. Hidden feature analysis suggests that the AtLSTM model accumulates the useful information over the growth period, while the Transformer model extracts the temporal dependencies that contribute important information to high-level feature learning. The learned features contain more effective and refined information than the raw input features and thus are better suited for crop classification. The soft output analysis in the in-season classification scenario demonstrates that increased length of input time series improves the model's confidence in the classification results. The further comparison of input feature importance in different sites and years demonstrates the applicability of the interpretation approach at larger spatiotemporal extents with heterogeneous landscapes and interannual variability. This study provides a multi-perspective evaluation to identify key features in multi-spectral and multi-temporal remote sensing data, and yields a practical approach to integrate agronomy knowledge in deep learning-based crop mapping.

Pre- and within-season crop type classification trained with archival land cover information

Crop type maps were created without the traditional need for in-season training data across the Corn Belt and Great Plains regions of the United States. This was accomplished through machine learning of historical land cover information, paired with a time-series of multi-spectral satellite imagery composites spanning the growing season, to develop rulesets, which were used for real-time prediction in the current year. Specifically, a decade's worth of annual 30 m resolution crop specific maps, known as the Cropland Data Layer (CDL), provided the foundation, and prior and current year's satellite imagery from Landsat 7 and 8 and Sentinel-2a and -2b built upon it. Four modeling scenarios, all using random forests, were performed to understand the crop mapping abilities of the datasets independently and combined. They were 1) use of CDLs only (i.e. prediction based solely on crop rotation history) 2) use of Landsat 7 and 8 bottom-of-atmosphere surface reflectance imagery only, 3) use of Sentinel-2a and -2b top-of-atmosphere imagery only, and 4) integration of the CDL, Landsat, and Sentinel-2 information together in a unified effort. Furthermore, the model runs were generated monthly, beginning in April, through the growing season to provide understanding of classification performance as a function of time. The 2020 crop year, relatively normal in terms of planting and weather, was used for the test. Accuracy statistics were generated by randomly sampling 50 counties and comparing those classification outputs to the actual 2020 CDL. Pixel-level results showed that prediction by midsummer using only the CDL information provided a crop type map with corn and soybean consumer and producer agreement above 70% and winter wheat just below 50%. The early season imagery-based classifications were markedly worse. However, as Landsat or Sentinel-2 imagery accumulated through July, those classifications became significantly better than those reliant on the use of the CDL information only. Ultimately, the very best crop maps resulted from integrating the CDLs with a full season's worth of Landsat and Sentinel-2 imagery. At that point in late September, the corn and soybean agreements were around 85% and winter wheat near 70%. All analysis was performed within Google Earth Engine cloud-based public imagery repository and high-performance computing system. The classification outputs provide practitioners with US crop type maps in near real-time.

Mapping the Location and Extent of 2019 Prevent Planting Acres in South Dakota Using Remote Sensing Techniques

The inability of a farmer to plant an insured crop by the policy’s final planting date can pose financial challenges for the grower and cause reduced production for a widely impacted region. Prevented planting is primarily caused by excess moisture or rainfall such as the catastrophic flooding and widespread conditions that prevented active field work in the midwestern region of United States in 2019. While the Farm Service Agency reports the number of such “prevent plant” acres each year at the county scale, field-scale maps of prevent plant fields—which would enable analyses related to assessing and mitigating the impact of climate on agriculture—are not currently available. The aim of this study is to demonstrate a method for mapping likely prevent plant fields based on flood mapping and historical cropland maps. We focused on a study region in eastern South Dakota and created flood maps using Landsat 8 and Sentinel 1 images from 2018 and 2019. We used automatic threshold-based change detection using NDVI and NDWI to accentuate changes likely caused by flooding. The NDVI change detection map showed vegetation loss in the eastern parts of the study area while NDWI values showed increased water content, both indicating possible flooding events. The VH polarization of Sentinel 1 was also particularly useful in identifying potential flooded areas as the VH values for 2019 were substantially lower than those of 2018, especially in the northern part of the study area, likely indicating standing water or reduced biomass. We combined the flood maps from Landsat 8 and Sentinel 1 to form a complete flood likelihood map over the entire study area. We intersected this flood map with a map of fallow pixels extracted from the Cropland Data Layer to produce a map of predicted prevent plant acres across several counties in South Dakota. The predicted figures were within 10% error of Farm Service Agency reports, with low errors in the most affected counties in the state such as Beadle, Hanson, and Hand.

The RapeseedMap10 database: annual maps of rapeseed at a spatial resolution of 10 m based on multi-source data

Abstract. Large-scale, high-resolution maps of rapeseed (Brassica napus L.), a major oilseed crop, are critical for predicting annual production and ensuring global energy security, but such maps are still not freely available for many areas. In this study, we developed a new pixel- and phenology-based algorithm and produced a new data product for rapeseed planting areas (2017–2019) in 33 countries at 10 m spatial resolution based on multiple data. Our product is strongly consistent at the national level with official statistics of the Food and Agricultural Organization of the United Nations. Our rapeseed maps achieved F1 spatial consistency scores of at least 0.81 when compared with the Cropland Data Layer in the United States, the Annual Crop Inventory in Canada, the Crop Map of England, and the Land Cover Map of France. Moreover, F1 scores based on independent validation samples ranged from 0.84 to 0.91, implying a good consistency with ground truth. In almost all countries covered in this study, the rapeseed crop rotation interval was at least 2 years. Our derived maps suggest, with reasonable accuracy, the robustness of the algorithm in identifying rapeseed over large regions with various climates and landscapes. Scientists and local growers can use the freely downloadable derived rapeseed planting areas to help predict rapeseed production and optimize planting structures. The product is publicly available at https://doi.org/10.17632/ydf3m7pd4j.3 (Han et al., 2021).

Rapid in-season mapping of corn and soybeans using machine-learned trusted pixels from Cropland Data Layer

A timely and detailed crop-specific land cover map can support many agricultural applications and decision makings. However, in-season crop mapping over a large area is still challenging due to the insufficiency of ground truth in the early stage of a growing season. To address this issue, this paper presents an efficient machine-learning workflow for the rapid in-season mapping of corn and soybeans fields without ground truth data for the current year. We use trusted pixels, a set of pixels that are predicted from the historical Cropland Data Layer (CDL) data with high confidence in the current year’s crop type, to label training samples on multi-temporal satellite images for crop type classification. The entire mapping process only involves a limited number of satellite images acquired within the growing season (normally 3–4 images per scene) and no field data needs to be collected. According to the investigation on 12 states of the U.S. Corn Belt, it is found that a considerable number of trusted pixels can be identified from the historical CDL data by the trusted pixel prediction model based on artificial neural network. According to the experiment on 49 Landsat-8 scenes and 31 Sentinel-2 tiles, the in-season maps of corn and soybeans are expected to reach 85%–95% agreement with CDL as well as field data by mid-July. Once the in-season satellite imagery becomes available, the crop cover map can be rapidly created even with limited computational resources. This study provides a new perspective and detailed guidance for rapid in-season mapping of corn and soybeans, which can be potentially applied to identify more diverse crop types and scaled up to the entire United States.

Mapping fine-scale human disturbances in a working landscape with Landsat time series on Google Earth Engine

Large fractions of human-altered lands are working landscapes where people and nature interact to balance social, economic, and ecological needs. Achieving these sustainability goals requires tracking human footprints and landscape disturbance at fine scales over time—an effort facilitated by remote sensing but still under development. Here, we report a satellite time-series analysis approach to detecting fine-scale human disturbances in an Ohio watershed dominated by forests and pastures but with diverse small-scale industrial activities such as hydraulic fracturing (HF) and surface mining. We leveraged Google Earth Engine to stack decades of Landsat images and explored the effectiveness of a fuzzy change detection algorithm called the Bayesian Estimator of Abrupt change, Seasonality, and Trend (BEAST) to capture fine-scale disturbances. BEAST is an ensemble method, capable of estimating changepoints probabilistically and identifying sub-pixel disturbances. We found the algorithm can successfully capture the patterns and timings of small-scale disturbances, such as grazing, agriculture management, coal mining, HF, and right-of-ways for gas and power lines, many of which were not captured in the annual land cover maps from Cropland Data Layers—one of the most widely used classification-based land dynamics products in the US. For example, BEAST could detect the initial HF wellpad construction within 60 days of the registered drilling dates on 88.2% of the sites. The wellpad footprints were small, disturbing only 0.24% of the watershed in area, which was dwarfed by other activities (e.g., right-of-ways of utility transmission lines). Together, these known activities have disturbed 9.7% of the watershed from the year 2000 to 2017 with evergeen forests being the most affected land cover. This study provides empirical evidence on the effectiveness and reliability of BEAST for changepoint detection as well as its capability to detect disturbances from satellite images at sub-pixel levels and also documents the value of Google Earth Engine and satellite time-series imaging for monitoring human activities in complex working landscapes.

Multi-scale fallow land dynamics in a water-scarce basin of the U.S. Southwest

ABSTRACT Improving our understanding of land-use change is critical for water management in semi-arid areas, due to its effects on the hydrological cycle. In the U.S. Southwest, fallowing farmland has become one strategy to reduce water use. Previous to this study, the magnitude and location of changes in fallowing have not been studied in depth. Using the 30-meter Cropland Data Layer, this study assessed the spatial and temporal patterns of fallowing in the U.S. portion of the Rio Grande/Río Bravo Basin (RGB) at three spatial scales (basin, state, and ecoregion) between 2008 and 2018. Our results do not show evidence of an increasing trend in fallowing at the basin level. However, the spatio-temporal patterns differed considerably among states and ecoregions, revealing hotspots of fallowing. By showing that land fallowing is not a widespread practice across the basin, our findings indicate that the potential of this strategy to save water has been underused.

Examining the Characteristics of the Cropland Data Layer in the Context of Estimating Land Cover Change

The United States Department of Agriculture (USDA) Cropland Data Layer (CDL) provides spatially explicit information about crop production area and has served as a prevalent data source for characterizing cropland change in the U.S. in the last decade. Understanding the accuracy of the CDL is paramount because of the reliance on it for management and policy making. This study examined the characteristics of the CDL from 2007 to 2017 using comparisons to other USDA datasets. The results showed when examining the cropland area for the same year, the CDL produced comparable trends with other datasets (R2 > 0.95), but absolute area differed. The estimated area of cropland changes from 2007 to 2012, 2008 to 2012 and 2012 to 2017 varied from weak to moderate correlation between the CDL and the tabular data (R2 = 0.005~0.63). Differences in area of cropland change varied widely between data sources with the CDL estimating much larger change area. A series of image processing techniques designed to improve the confidence in cropland change estimated using the CDL reduced the area of estimated cropland change. The techniques also, unexpectedly, lowered the correlation in change estimated between the CDL and the tabular datasets. Estimated land cover change area varied widely based on analyses applied and could reverse from increasing to declining area in cropland. Further analyses showed unlikely change scenarios when comparing different year combinations. The authors recommend the CDL only be used for land cover change analysis if the error can be estimated and is within change estimates.

Spatiotemporal Changes of Winter Wheat Planted and Harvested Areas, Photosynthesis and Grain Production in the Contiguous United States from 2008–2018

Winter wheat is a main cereal crop grown in the United States of America (USA), and the USA is the third largest wheat exporter globally. Timely and reliable in-season forecast and year-end estimation of winter wheat grain production in the USA are needed for regional and global food security. In this study, we assessed the consistency between the agricultural statistical reports and satellite-based data for winter wheat over the contiguous US (CONUS) at both the county and national scales. First, we compared the planted area estimates from the National Agricultural Statistics Service (NASS) and the Cropland Data Layer (CDL) from 2008–2018. Second, we investigated the relationship between gross primary production (GPP) estimated by the vegetation photosynthesis model (VPM) and grain production from the NASS. Lastly, we explored the in-season utility of GPPVPM in monitoring seasonal production. Strong spatiotemporal consistency of planted areas was found between the NASS and CDL datasets. However, in the Southern Great Plains, both the CDL and NASS planted acreage were noticeable larger (>20%) than the NASS harvested area, where some winter wheat fields were used as forage for cattle grazing. County-level GPPVPM was linearly related with grain production of winter wheat, with an R2 value of 0.68 across the CONUS. The relationships between grain production and GPPVPM in those counties without a substantial difference (<20%) between planted and harvested area were much stronger and their harvest index (HIGPP) values ranged from 0.2–0.3. GPPVPM in May could explain about 70–90% of the variance of winter wheat grain production. Our findings highlight the potential of GPPVPM in winter wheat monitoring, especially for those high harvested/planted ratio, which could provide useful data to guide planning and marketing for decision makers, stakeholders, and the public.

A two-step mapping of irrigated corn with multi-temporal MODIS and Landsat analysis ready data

Timely and reliable information about irrigated croplands is important for crop water stress analysis and studies of water, energy, and food security. This study mapped irrigated and non-irrigated corn at 30 m resolution for the state of Nebraska using a two-step multi-temporal image classification of MODIS and Landsat Analysis Ready Data (ARD). Starting from the drought year of 2012, when there was a high contrast between irrigated and non-irrigated fields, we first conducted image classification using the 250 m MODIS multi-temporal NDVI data. Training pixels were automatically derived, based on counties with predominant irrigated and non-irrigated cornfields. The MODIS-derived irrigated vs. non-irrigated map was further spatially filtered to generate training data covering the entire Nebraska to support automated Landsat ARD classification, footprint-by-footprint. Three classification algorithms of multi-layer perceptron (MLP) neural network, Random Forest (RF), and Support Vector Machine (SVM) were implemented to classify all available Landsat ARD images within the growing season (i.e. May to November). Given the issues of scanline corrector (SLC) error and cloud contamination, the provisional Landsat-based classifications were finally gap-filled to generate a seamless state-wise irrigation map guided by decreasing cross-validation accuracy. Pixel-wise accuracy assessments showed similar overall accuracies of 89.6%, 89.3%, and 90.0% for MLP, RF, and SVM, respectively. They are 3-6% higher than a commonly used gap-filing procedure based on valid (cloud free) pixel count for growing season images. The estimated areas of irrigated corn from Landsat-based mapping were consistent with the 2012 USDA county level census data (R2 = 0.97 and RMSE = 37.70 km2). Using the 2012 Landsat-derived irrigation map and the USDA’s annual Cropland Data Layer as inputs, we further developed training data for annual irrigation mapping between 2013 and 2018. Pixel-wise assessment of the 2016 map showed reasonable overall accuracies of 78.4–79.6% for three classification algorithms. The annual maps yielded R2 of 0.94–0.98 and RMSE values of 37.70–57.62 km2 for various mapping years compared with USDA county statistics. These results suggest that our proposed two-step analytical method has a high potential for automated annual irrigation mapping at 30 m spatial resolution (especially for the arid and semi-arid western U.S.), providing clear field boundaries and irrigation frequency information that are vitally important for accurate agricultural water use analysis.

RETRACTED: Economic value of agricultural water use for Arizona

This paper aims to estimate the economic value of agricultural water use in Arizona based on an improved economic value estimation model (EVIW model). In this research, EVIW method is optimized for the economic value estimation of agricultural water use from the perspective of crops. Crop Price per unit for crop production, crop yield per acre and irrigation area are used to calculate gross profit. Crop variable costs per acre (labor costs, chemical costs, machinery fuel and repair costs, etc.) and corresponding irrigation areas are used to calculate variable costs. Gross revenues and variable costs are used to calculate the net returns generated by irrigation water. Arizona agricultural water use for crops are estimated based on meteorological data and Cropland Data Layers. Finally, economic values of agricultural water are determined by the net returns of irrigated crops and agricultural water use. The results show that the economic values of agricultural water are positive and negative with great variations. The models applied in the research bring forward practical conclusions for further water allocations, transfer, and policies.