•Application of Random forest approach for crop classification.•The use of EVI obtained from time-series Landsat 7 ETM+ imagery as predictor variables.•Effects of number of predictor variables, ...decision trees and training data on classification accuracy.•Random forest as an appropriate method to classify the upland field crops within a homogeneous region.
Crop classification of homogeneous landscapes and phenology is a common requirement to estimate land cover mapping, monitoring, and land use categories accurately. In recent missions, classification methods using medium or high spatial resolution data, which are multi-temporal with multiple frequencies, have become more attractive. A new mode of incorporating spatial and temporal dependence in a homogeneous region was tried using the Random Forest (RF) classifier for crop classification. A time-series of medium spatial resolution enhanced vegetation index (EVI) and its summary statistics obtained from Landsat 7 Enhanced Thematic Mapper Plus (Landsat 7 ETM+) were used to develop a new technique for crop type classification. Eight classes were studied: alfalfa, asparagus, avocado, cotton, grape, maize, mango, and tomato. Evaluation was based on several criteria: sensitivity to training dataset size, the number of variables, and mapping accuracy. Results showed that the training dataset size strongly affects the classifier accuracy, but if the training data increase, the rate of improvement decreases. The RF algorithm yielded overall accuracy of 81% and a Kappa statistic of 0.70, indicating high model performance. Additionally, the variable importance measures demonstrated that the mode and sum of EVI had extremely important variables for crop class separability. RF had computationally good performance. They can be enhanced by choosing an appropriate classifier for multiple statistics and time-series of Landsat imagery. It might be more economical to use no-cost imaging for crop classification using open-source software.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK
Soil salinization is widely recognized to be a major threat to worldwide agriculture. Despite decades of research in soil mapping, no reliable and up-to-date salinity maps are available for large ...geographical regions, especially for the salinity ranges that are most relevant to agricultural productivity (i.e., salinities less than 20dSm−1, when measured as the electrical conductivity of the soil saturation extract). This paper explores the potentials and limitations of assessing and mapping soil salinity via linear modeling of remote sensing vegetation indices. A case study is presented for western San Joaquin Valley, California, USA using multi-year Landsat 7 ETM+ canopy reflectance and the Canopy Response Salinity Index (CRSI). Highly detailed salinity maps for 22 fields comprising 542ha were used for ground-truthing. Re-gridded to 30×30m, the ground-truth data totaled over 5000pixels with salinity values in the range 0 to 35.2dSm−1. Multi-year maximum values of CRSI were used to model soil salinity. Soil type, meteorological data, and crop type were evaluated as covariates. All considered models were evaluated for their fit to the whole data set as well as their performance in a leave-one-field-out spatial cross-validation. The best performing model was a function of CRSI, crop type (i.e., cropped or fallow), rainfall, and average minimum temperature, with R2=0.728 when evaluated against all data and R2=0.611 for the cross-validation predictions. Broken out by salinity classes, the mean absolute errors (MAE) for the cross-validation predictions were (all units dSm−1): 2.94 for the 0–2 interval (non-saline), 2.12 for 2–4 (slightly saline), 2.35 for 4–8 (moderately saline), 3.23 for 8–16 (strongly saline), and 5.64 for >16 (extremely saline). On a per-field basis, the validation predictions had good agreement with the field average (R2=0.79, MAE=2.46dSm−1), minimum (R2=0.76, MAE=2.25dSm−1), and maximum (R2=0.76, MAE=3.09dSm−1) observed salinity. Overall, reasonably accurate and precise high resolution, regional-scale remote sensing of soil salinity is possible, even over the critical range of 0 to 20dSm−1, where researchers and policy makers must focus to prevent loss of agricultural productivity and ecosystem health.
Regional scale soil salinity assessment can successfully be carried out using multi-year Landsat ETM+ canopy reflectance and information on crop cover and meteorological settings. Display omitted
•Multi-year maxima of Landsat ETM+ vegetation indices correlates with soil salinity.•Linear regressions provide reliable salinity estimates at the regional scale.•Crop and meteorological covariates increase accuracy of soil salinity predictions.•Salinity assessment models are validated through a spatial cross-validation.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK
The Landsat satellites have been providing spectacular imagery of the Earth's surface for over 40years. However, they acquire images at view angles ±7.5° from nadir that cause small directional ...effects in the surface reflectance. There are also variations with solar zenith angle over the year that can cause apparent change in reflectance even if the surface properties remain constant. When Landsat data from adjoining paths, or from long time series are used, a model of the surface anisotropy is required to adjust all Landsat observations to a uniform nadir view (primarily for visual consistency, vegetation monitoring, or detection of subtle surface changes). Here a generalized approach is developed to provide consistent view angle corrections across the Landsat archive. While this approach is not applicable for generation of Landsat surface albedo, which requires a full characterization of the surface bidirectional reflectance distribution function (BRDF), or for correction to a constant solar illumination angle across a wide range of sun angles, it provides Landsat nadir BRDF-adjusted reflectance (NBAR) for a range of terrestrial monitoring applications.
The Landsat NBAR is derived as the product of the observed Landsat reflectance and the ratio of the reflectances modeled using MODIS BRDF spectral model parameters for the observed Landsat and for a nadir view and fixed solar zenith geometry. In this study, a total of 567 conterminous United States (CONUS) January and July 2010 Landsat 5 Thematic Mapper (TM) and Landsat 7 Enhanced Thematic Mapper (ETM+) images that have swath edge overlapping paths sensed in alternating backscatter and forward scattering orientations were used. The average difference between Landsat 5 TM and Landsat 7 ETM+ surface reflectance in the forward and backward scatter directions at the overlapping Landsat scan edges was quantified. The CONUS July view zenith BRDF effects were about 0.02 in the Landsat visible bands, and about 0.03, 0.05 and 0.06, in the 2.1μm, 1.6μm and near-infrared bands respectively. Comparisons of Landsat 5 TM and Landsat 7 ETM+ NBAR derived using MODIS BRDF spectral model parameters defined with respect to different spatial and temporal scales, and defined with respect to different land cover types, were undertaken. The results suggest that, because the BRDF shapes of different terrestrial surfaces are sufficiently similar over the narrow 15° Landsat field of view, a fixed set of MODIS BRDF spectral model parameters may be adequate for Landsat NBAR derivation with little sensitivity to the land cover type, condition, or surface disturbance. A fixed set of BRDF spectral model parameters, derived from a global year of highest quality snow-free MODIS BRDF product values, are provided so users may implement the described Landsat NBAR generation method.
•Landsat NBAR derivation method developed using MODIS BRDF model parameters•Based on BRDF shapes of terrestrial surfaces similar over Landsat field of view•Fixed BRDF model parameters compared with local ones•Global fixed parameters provided so users may implement the method•Method insensitive to land cover and so is applicable to all Landsat record
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK, ZRSKP
The continued development of algorithms using multitemporal Landsat data creates opportunities to develop and adapt imputation algorithms to improve the quality of that data as part of preprocessing. ...One example is de-striping Enhanced Thematic Mapper Plus (ETM+, Landsat 7) images acquired after the Scan Line Corrector failure in 2003. In this study, we apply window regression, an algorithm that was originally designed to impute low-quality Moderate Resolution Imaging Spectroradiometer (MODIS) data, to Landsat Analysis Ready Data from 2014–2016. We mask Operational Land Imager (OLI; Landsat 8) image stacks from five study areas with corresponding ETM+ missing data layers, using these modified OLI stacks as inputs. We explored the algorithm’s parameter space, particularly window size in the spatial and temporal dimensions. Window regression yielded the best accuracy (and moderately long computation time) with a large spatial radius (a 7 × 7 pixel window) and a moderate temporal radius (here, five layers). In this case, root mean square error for deviations from the observed reflectance ranged from 3.7–7.6% over all study areas, depending on the band. Second-order response surface analysis suggested that a 15 × 15 pixel window, in conjunction with a 9-layer temporal window, may produce the best accuracy. Compared to the neighborhood similar pixel interpolator gap-filling algorithm, window regression yielded slightly better accuracy on average. Because it relies on no ancillary data, window regression may be used to conveniently preprocess stacks for other data-intensive algorithms.
Full text
Available for:
IZUM, KILJ, NUK, PILJ, PNG, SAZU, UL, UM, UPUK
Landsat 8, originally known as the Landsat Data Continuity Mission (LDCM), is a National Aeronautics and Space Administration (NASA)-U.S. Geological Survey (USGS) partnership that continues the ...legacy of continuous moderate resolution observations started in 1972. The conception of LDCM to the reality of Landsat 8 followed an arduous path extending over nearly 13years, but the successful launch on February 11, 2013 ensures the continuity of the unparalleled Landsat record. The USGS took over mission operations on May 30, 2013 and renamed LCDM to Landsat 8. Access to Landsat 8 data was opened to users worldwide. Three years following launch we evaluate the science and applications impact of Landsat 8. With a mission objective to enable the detection and characterization of global land changes at a scale where differentiation between natural and human-induced causes of change is possible, LDCM promised incremental technical improvements in capabilities needed for Landsat scientific and applications investigations. Results show that with Landsat 8, we are acquiring more data than ever before, the radiometric and geometric quality of data are generally technically superior to data acquired by past Landsat missions, and the new measurements, e.g., the coastal aerosol and cirrus bands, are opening new opportunities. Collectively, these improvements are sparking the growth of science and applications opportunities. Equally important, with Landsat 7 still operational, we have returned to global imaging on an 8-daycycle, a capability that ended when Landsat 5 ceased operational Earth imaging in November 2011. As a result, the Landsat program is on secure footings and planning is underway to extend the record for another 20 or more years.
•Landsat Data Continuity Mission (Landsat 8) was launched February 11, 2013.•22 manuscripts highlight the science impacts of Landsat 8.•Landsat 8 is acquiring more data than ever before.•Radiometric and geometric quality are superior to previous Landsat data.•New bands, e.g., coastal aerosol and cirrus, create new applications opportunities.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK
With the advent of the free U.S. Landsat data policy it is now feasible to consider the generation of global coverage 30m Landsat data sets with temporal reporting frequency similar to that provided ...by the monthly Web Enabled Landsat (WELD) products. A statistical Landsat metadata analysis is reported considering more than 800,000 Landsat 5 TM and Landsat 7 ETM+ acquisitions obtained from the U.S. Geological Survey (USGS) Earth Resources Observation and Science (EROS) Center archive. The global monthly probabilities of acquiring a cloud-free land surface observation for December 1998 to November 2001 (2000 epoch) and from December 2008 to November 2011 (2010 epoch) are reported to assess the availability of the Landsat data in the USGS Landsat archive for global multi-temporal land remote sensing applications. The global probabilities of acquiring a cloud-free land surface observation in each of three different seasons with the highest seasonal probabilities of cloud-free land surface observation are reported, considering one, two and three years of Landsat data, to assess the availability of Landsat data for global land cover mapping. The probabilities are derived considering Landsat 5 TM only, Landsat 7 ETM+ only, and both sensors combined, to examine the relative benefits of using one or both Landsat sensors. The results demonstrate the utility of combing both Landsat 5 TM and Landsat 7 ETM+ data streams to take advantage of their different acquisition patterns and to mitigate the deleterious impact of the Landsat 7 ETM+ 2003 scan line failure. Sensor combination provided a greater global acquisition coverage with a 1.7% to 14.4% higher percentage of land locations acquired monthly compared to considering Landsat 7 ETM+ data alone. The mean global monthly probability of a cloud-free land surface observation for the combined sensors was up to nearly 1.4 and 6.7 times greater than for ETM+ and TM alone respectively. The probability of acquiring a cloud-free Landsat land surface observation in different seasons was greater when more years of data were considered and when both Landsat sensor data were combined. Considering combined sensors and 36months of data, 86.4% and 84.2% of the global land locations had probabilities ≥0.95 for the 2000 and 2010 epochs respectively, with a global mean probability of 0.92 (σ 0.24) for the 2000 epoch and 0.90 (σ 0.28) for the 2010 epoch. These results indicate that 36months of combined Landsat sensor data will provide sufficient land surface observations for 30m global land cover mapping using a multi-temporal supervised classification scheme.
► The impact of clouds and SLC_OFF on availability of clear Landsat land observations. ► Combing Landsat 5 TM and Landsat 7 ETM+ data streams is advantages. ► Compare to ETM+ alone both sensors provide up to 14.4% higher monthly land coverage. ► 36 months of combined Landsat 5 and 7 data can support 30m global land cover mapping.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK
Aboveground biomass estimation is critical in understanding forest contribution to regional carbon cycles. Despite the successful application of high spatial and spectral resolution sensors in ...aboveground biomass (AGB) estimation, there are challenges related to high acquisition costs, small area coverage, multicollinearity and limited availability. These challenges hamper the successful regional scale AGB quantification. The aim of this study was to assess the utility of the newly-launched medium-resolution multispectral Landsat 8 Operational Land Imager (OLI) dataset with a large swath width, in quantifying AGB in a forest plantation. We applied different sets of spectral analysis (test I: spectral bands; test II: spectral vegetation indices and test III: spectral bands+spectral vegetation indices) in testing the utility of Landsat 8 OLI using two non-parametric algorithms: stochastic gradient boosting and the random forest ensembles. The results of the study show that the medium-resolution multispectral Landsat 8 OLI dataset provides better AGB estimates for Eucalyptus dunii, Eucalyptus grandis and Pinus taeda especially when using the extracted spectral information together with the derived spectral vegetation indices. We also noted that incorporating the optimal subset of the most important selected medium-resolution multispectral Landsat 8 OLI bands improved AGB accuracies. We compared medium-resolution multispectral Landsat 8 OLI AGB estimates with Landsat 7 ETM+estimates and the latter yielded lower estimation accuracies. Overall, this study demonstrates the invaluable potential and strength of applying the relatively affordable and readily available newly-launched medium-resolution Landsat 8 OLI dataset, with a large swath width (185-km) in precisely estimating AGB. This strength of the Landsat OLI dataset is crucial especially in sub-Saharan Africa where high-resolution remote sensing data availability remains a challenge.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK
The potential strengths and limitations of the Landsat systems for water clarity and colored dissolved organic matter (CDOM) measurement were evaluated in Minnesota in the summers of 2013 and 2014. ...Optical water quality characteristics, including chlorophyll a, total suspended solids (TSS), dissolved organic carbon (DOC), and CDOM were collected along with imagery from Landsats 7 and 8. Sites represented a wide range of concentrations of CDOM, chlorophyll, and mineral suspended solids (MSS), the primary factors that affect reflectance. Clear images from September 24, 2013 (Landsat 7) and September 16, 2013 (Landsat 8) acquired for northern Minnesota eight days apart allowed comparison of the respective ETM+ and OLI sensors for CDOM measurements. We examined a wide variety of potential band and band ratio models and found some two-variable models that included the NIR band worked well for Landsat 8 (R2=0.82) and reasonably well for Landsat 7 (R2=0.74). The commonly used green/red model had a poor fit for both sensors (R2=0.24, 0.25), and five sites with high MSS were clear outliers. Exclusion of these sites and other sites not included with the Landsat 7 dataset yielded a less optically complex subset of 20 coincident lakes. For this subset strong models were found for many band and band ratio models, including the commonly used green/red model with R2=0.79 for Landsat 7 and R2=0.81 for Landsat 8. The less optically complex subset may explain why the green/red model has worked well in other areas. For optically complex waters CDOM models that used the new Landsat 8 ultra-blue and narrower NIR band worked best for the full dataset indicating that the new bands and other Landsat 8 characteristics, such as higher radiometric sensitivity and improved signal-to-noise ratios, improve CDOM measurements.
For water clarity measured as Secchi depth (SD), we compared September 1, 2008 Landsat 7 and August 22, 2013 Landsat 8 images from path 28 using stepwise regression to identify the best model using all bands and band ratios including the new blue and narrower NIR band. The best water clarity model for Landsat 8 used the OLI 2/4 band ratio plus OLI band 1 and was nearly identical with a model using the OLI 2/4 band ratio plus OLI band 2. The latter model is similar to the model used for previous Landsat water clarity assessments, which used the ETM+ 1/3 band ratio plus ETM+ band 1. For SD measurements we found strong relationships with both sensors, with only slight improvements for the OLI sensor for the lakes in our datasets. In contrast to some previous reports that indicated Landsat 7's ETM+ lacked sufficient sensitivity for reliable retrieval of CDOM, we found that overall both sensors worked well for water clarity and CDOM measurements. This will allow their continued use for current and historical measurements of important water characteristics on a regional scale.
•Landsats 7 and 8 data were compared for mapping CDOM and clarity of inland lakes.•Landsat 8 was better for estimating CDOM than Landsat 7 in optically complex waters.•Landsat 8 was only a slight improvement over Landsat 7 for measuring water clarity.•Landsats 7 and 8 will continue and enhance remote sensing of regional water quality.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK, ZRSKP
Landsat time-series multi-spectral data, GLAS (Geoscience Laser Altimeter System) height data and a regression tree model were used to estimate tree height for a transect in Sub-Saharan Africa ...ranging from the Sahara Desert through the Congo Basin to the Kalahari Desert (+22 to −22° latitude and 23 to 24° longitude). Objectives included comparing the performance of Landsat 7- and 8-derived inputs separately and combined in mapping tree height at a regional scale, assessing the relative value of good observation counts and different Landsat spectral inputs for tree height estimation across a range of environments, and describing tree height distributions and discontinuities in Sub-Saharan Africa. A total of 5371 images were processed and per pixel quality assessed to create a set of multi-temporal metrics for the 2013 and 2014 calendar years for Landsat 7 only, Landsat 8 only and both Landsat 7 and 8 combined. Differences in performance were slight between different sensor inputs. However, performance generally improved with increasing numbers of good observations. Metrics derived from red reflectance data contributed most in estimating tree height. The regression tree algorithm accurately reproduced the LiDAR-derived height training data with an overall mean absolute error (MAE) for tree height estimation of 2.45m using integrated Landsat 7 and 8 data. Significant underestimations were quantified for tall tree cover (MAE of 4.65m for >20m heights) and overestimations for low/no tree cover (MAE 1.61 for <5m heights). Resulting tree distributions were found to be discontinuous with a primary dry seasonal woodlands cluster of 5–10m in height, a second cluster of primarily dry evergreen forest tree cover from 11–17m, and a third cluster of humid evergreen forest tree cover ≥18m. The integration of Landsat 7 and 8 and forthcoming Sentinel 2 time-series optical data to extend the value of LiDAR forest structure measurements is recommended.
•Employed Landsat 7 and 8 in mapping tree height in Sub-Saharan Africa•Best performance found using integrated Landsat 7 and 8 (MAE=2.45m)•Differences in performance were slight between different Landsat 7 and 8 inputs.•Performance generally improved with increasing numbers of good observations.•Identified discrete height clusters of woodlands 5–10m and 11–17m, and forests ≥18m
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK
The newly launched Landsat 8 satellite continues the long and extremely important record of Earth observation from the Landsat program. We analyzed differences between Landsat 7 and Landsat 8 surface ...reflectances and cirrus cloud characterization to address how substitutable Landsat 8 observations are within this long archive. Comparison of surface reflectance estimates acquired near simultaneously during Landsat 8's underflight orbital placement shows that Landsat 8 surface reflectance is consistently darker in the blue, green, and red bands and brighter in the near infrared than in Landsat 7. Vegetation indices that rely on the visible and near infrared bands should be used with caution as individual biases in index components can be amplified to create large biases in vegetation indices. We also analyzed time series datasets from the Landsat Climate Data Record (CDR) surface reflectance product across four scenes that contained only Landsat 7 data, Landsat 7 data and only Landsat 8 data post-launch, and Landsat 7 data and data from both sensors post-launch to investigate how sensor differences propagate in time series analysis. If left uncorrected or unexplained, the difference in reflectance between Landsat 7 and Landsat 8 creates spurious time trends in visible wavelengths and in the Normalized Difference Vegetation Index (NDVI). The introduction of Landsat 8 into time series of Landsat 7 data also biases the mean reflectance or vegetation index value as measured by a time series model intercept while increasing the Root Mean Squared Error of such models. We characterized the spectral reflectance of cirrus clouds in the underflight data that were omitted from Landsat 7 cloud masks but were detected in Landsat 8's cloud mask due to the use of the newly added cirrus band. While these cirrus cloud observations missed in Landsat 7's cloud mask are only slightly brighter in the visible bands, a simulation of time series containing Landsat 8 data that does not use the cirrus band shows that omission of cirrus clouds can result in anomalously brighter time series intercepts and positive time trends. Our results indicate that while Landsat 8 has improved on the legacy of previous sensors through increased radiometric resolution, better cloud identification, and better geometric accuracy, difference in reflectance between sensors in the current Landsat CDR product must be corrected or explained within time series analysis to avoid deleterious consequences. Future efforts should identify the contributions of target specific effects versus differences in atmospheric correction methods to better inform approaches to synthesize the two sensors.
•We compare the Landsat Climate Data Record surface reflectance from Landsats 7 and 8.•Landsat 8 is consistently darker in the blue, green, and red bands than Landsat 7.•Sensor specific differences in reflectance bias time series model estimates.•Landsat 8's cirrus band can detect subtle contamination by cirrus clouds.•Cirrus cloud contamination can produce spurious results in time series analysis.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK