Surface Cover Types Research Articles

A priori knowledge accumulation and its application to linear BRDF model inversion

A priori knowledge can significantly improve the retrieval of surface bidirectional reflectance and spectral albedo from satellite observations. Here a priori knowledge takes the form of field measurements of bidirectional reflectance factors for various surface cover types in red and near‐infrared bands. Bidirectional reflectance and albedo retrieval refers to inversion of a kernel‐driven bidirectional reflectance distribution function (BRDF) model using surface reflectance observations derived from orbiting spacecraft. A priori knowledge is applied when noise and poor angular sampling reduce the accuracy of model inversion given a limited number of observations. In such cases, a priori knowledge can indicate when retrieved kernel weights or albedos are outside expected bounds, leading to a closer examination of data. If data are noisy, a priori knowledge can be used to smooth the data. If the data exhibit poor angular sampling, a priori knowledge can be used according to Bayesian inference theory to yield a posteriori estimates of unknown kernel weights. In the latter application, Bayes theory is applied in data space rather than in parameter space. Extensive study and simulation using 73 sets of field observations and 395 spaceborne observation sets from the POLDER instrument validates the importance of a priori information in improving inversions and BRDF retrievals.

Local and regional albedo observations of arctic first‐year sea ice during melt ponding

The shortwave surface albedo is a critical climatological parameter for sea ice, especially during the spring melt period when the ice contains a mixture of highly reflective (snow) and absorptive (melt ponds) surfaces. Broadband and spectral albedo measurements were made over numerous surface types on first‐year sea ice during the melt period in Wellington Channel, Nunavut during the Collaborative‐Interdisciplinary Cryospheric Experiment 1997. Albedo measurements ranged from 0.75 (moist snow) to 0.21 (dark melt ponds) with many unique intermediate surfaces. Aircraft videography collected throughout Wellington Channel and Lancaster Sound was processed to reveal four main surface cover types (wet snow, mixed type, light ponds, and dark ponds). Surface albedo data were applied to the aircraft observations to upscale surface albedo measurements to regional scales. Aircraft video‐derived regional albedo (mean = 0.55±0.02) were comparable to helicopter and satellite‐derived (Advanced Very High Resolution Radiometer) albedo estimates.

Melt ponds on sea ice in the Canadian Archipelago: 1. Variability in morphological and radiative properties

The morphological and radiative properties of melt ponds on first‐year sea ice (FYI) were investigated during the summer of 1997 in the Canadian Archipelago as part of the Collaborative‐Interdisciplinary Cryospheric Experiment (C‐ICE) near Resolute Bay, Nunavut. In this paper we (1) describe a classification technique used to identify surface cover types from airborne videography during the melt pond season, (2) use the classification results to examine the fractional coverage of surface types and morphological characteristics of melt ponds, and (3) provide an estimate of the integrated shortwave albedo of this surface. Cluster analysis on the videography data identified four distinct surface cover types during the summer melt season: snow, saturated snow, light‐colored melt ponds, and dark‐colored melt ponds. Melt pond coverage was found to be highly variable over our study area. We found that pond sizes tended to be twice as large in areas of high melt pond density compared to ponds in areas of low pond density owing to their interconnective nature. Video data also identified an elongated melt pond morphology pattern over the smoothest FYI within our study region. A derived estimate of the integrated shortwave albedo was strongly related to the fractional cover of snow on the surface (R2 = 0.86). An analysis of combined fractional coverage of light and dark melt ponds from two aerial survey dates (Julian Days 181 and 184) revealed an aerial increase in melt ponds of ∼10.3%.

A two-site correlation model for wind speed, direction and energy estimates

A two-site wind correlation model was first modified and then tested on long-term data from five pairs of Canadian weather stations. The stations were chosen to cover a variety of topographic situations and surface cover types. A preliminary analysis of data from each of the 10 stations concluded that an absolute minimum of 12 months of monitoring is needed to estimate the long-term wind speed distribution. Increasing this short-term monitoring period to 24 months causes noticeable improvements; thereafter, the improvement is more gradual. It was also found that a short-term monitoring period between 12 and 24 months might produce worse estimates than for a 12-month period. For the two-site tests, one station from each of the five pairs was designated as a reference station; the other was designated as a target station. It was found that the model results derived from 1 yr of short-term simultaneous monitoring at the two stations and long-term data at the reference station outperform the estimates based solely on 2 yr of monitoring at the target station. The model results are further improved by using 2 yr of short-term monitoring. The conclusions from this study have implications for wind energy and air pollution studies where long-term estimates of the wind speed and direction distribution are needed and must be based on a relatively short-term period of monitoring.

Local advection of sensible heat in the snowmelt landscape of Arctic tundra

The spring landscape of the Arctic tundra is dominated by a snow cover which is highly variable in depth owing to redistribution by wind. Because of different energy dynamics, this heterogeneous land cover produces a horizontal transfer of energy at a small scale, a process termed local advection. An advection efficiency term (F S ), which represents the fraction of the sensible heat from snow-free patches which is advected to snow patches, was determined from field studies and published model results. Energy balance calculations demonstrated the strong contrast between the two surface cover types that drive advective processes, and F S was found to decrease exponentially with decreasing snow cover fraction. The field results suggest higher values of F S compared with the model results for single snow patches of varying size, but similar in magnitude to F S for multiple small snow patches. Utilizing exponential best-fit relationships between F S and fractional snow cover shows an increase in sensible heat flux of over 100% for low snow cover fractions. When considering the average flux over a composite snow and snow-free surface, the average sensible heat flux obtained from weighting the fluxes for each surface by their respective areas underestimates the composite flux when compared with when advection is considered. This work provides a simple method to estimate the effect of local advection on sensible heat to snow patches and the average flux from a composite surface during the snowmelt period, using only fluxes calculated independently for 0% snow cover and 100% snow cover and an estimate of F S . It demonstrates a good first estimate of the role of advection, but for future study the influence of wind speed, patch distribution patterns and fetch lengths needs to be considered more explicitly. This has important implications in studies of areal energy fluctuations over melting, patchy snow covers, basin water balance studies and regional and global climate modelling.

A distributed physical approach for surface-subsurface water transport modeling in agricultural watersheds

Abstract Surface cover and soil type have a major influence upon groundwater recharge and groundwater quality in agricultural watersheds. However, several hydrological models focus on simulating groundwater recharge without including the influence of agricultural practices and soil characteristics. In this study, ANSWERS, a distributed parameters surface nonpoint source model has been modified to include the simulation of water transport in the vadose and saturated zones. This model takes into account the spatial and temporal variability of crop cover and management practices, and the spatial variability of soil type and rainfall distribution. It is physically based and uses parameters that can be easily determined from readily available soil and plant information. It has been validated at multiple scales: local scale, field scale and watershed scale. At the local and field scale, it predicts accurately drainage below the root zone and evapotranspiration on different type of soil cover. At the watershed scale, it reproduces well the piezometric levels and trends of variation.

Corn Residue Age and Placement Effects on Evaporation and Soil Thermal Regime

AbstractDifferentiation between tillage‐ and residue‐induced effects on soil moisture and thermal regimes is necessary to improve residue management practices. The objective of this research was to quantify the influence of corn (Zea mays L.) residue age and placement on evaporation from and heat flow in two soils. Soil temperature, thermal conductivity, heat flux, and moisture content were monitored in duplicate monoliths (0.45 [length] by 0.35 [width] by 0.5 [height] m) of Nicollet loam (fine‐loamy, mixed, mesic Aquic Hapludoll) and Monona silt loam (fine‐loamy, mixed, mesic Typic Hapludoll) in a controlled environment chamber. Mass changes of one monolith of each soil were measured with load cells to estimate evaporation. Surface cover types consisted of a bare soil surface (B), fresh residue (F), weathered residue (W), and weathered residue with a 0.15‐m‐wide bare strip (S). Bare‐soil evaporation averaged 9.97 ± 0.66 mm for both soils during a 96‐h period and was reduced to an average of 5.76 ± 0.70 mm with residue cover. Nicollet soil with weathered residue had an average evaporation rate of 1.33 to 1.4 mm d−1. By comparison, evaporation rates measured in a field having Nicollet and associated soils with weathered corn residue averaged 1.37 and 1.11 mm d−1. Residue cover had no significant effect on thermal conductivity measured at 0.15 m that averaged 1.22 and 1.35 W m−1 K−1 for the Nicollet and Monona soils, respectively. Phase shift in soil temperature and average daily conductive soil heat flux were similar for field and chamber measurements but the amplitude of the soil temperature wave and conductive soil heat flux were much greater in the field.

Application of spectral mixture modelling to the regional assessment of land degradation: A case study from basilicata, Italy

AbstractMapping and monitoring of land degradation processes such as soil erosion has become an important task for both agricultural and environmental planners. The potential of using SPOT‐HRV data for assessing the types of soil erosion and land degradation is obvious. The spectral information and the spatial resolution of the multispectral data allow a high accuracy in local mapping and rapid regional assessment.In Basilicata, southern Italy, spectral mixture modelling (end‐member techniques) has been applied to extract relevant information for assessing soil erosion. This method allows an estimate of the proportion of surface cover types (end‐members) within the pixel. From the data sets average spectral responses were extracted for growing vegetation, non‐green sclerophylous vegetation mixed with dry grasses and bare soil. Using these spectra as end‐members in the model, it was possible to decompose the spectral information of all pixels into the three surface components, giving the percentage cover of the three types within every pixel. Classification of the pixels according to the percentage of surface cover or bare soil allows an assessment of erosion or erosion risk.

A simple hydrologically based model of land surface water and energy fluxes for general circulation models

A generalization of the single soil layer variable infiltration capacity (VIC) land surface hydrological model previously implemented in the Geophysical Fluid Dynamics Laboratory general circulation model (GCM) is described. The new model is comprised of a two‐layer characterization of the soil column, and uses an aerodynamic representation of the latent and sensible heat fluxes at the land surface. The infiltration algorithm for the upper layer is essentially the same as for the single layer VIC model, while the lower layer drainage formulation is of the form previously implemented in the Max‐Planck‐Institut GCM. The model partitions the area of interest (e.g., grid cell) into multiple land surface cover types; for each land cover type the fraction of roots in the upper and lower zone is specified. Evapotranspiration consists of three components: canopy evaporation, evaporation from bare soils, and transpiration, which is represented using a canopy and architectural resistance formulation. Once the latent heat flux has been computed, the surface energy balance is iterated to solve for the land surface temperature at each time step. The model was tested using long‐term hydrologic and climatological data for Kings Creek, Kansas to estimate and validate the hydrological parameters, and surface flux data from three First International Satellite Land Surface Climatology Project Field Experiment intensive field campaigns in the summer‐fall of 1987 to validate the surface energy fluxes.

The importance of instrument pointing accuracy for surface bidirectional reflectance distribution function mapping

Abstract An investigation into the magnitude of error introduced into bidirectional reflectance distribution functions (BRDFs) due to unqualified attitude perturbations and uncertainties in the pointing accuracy of the sensor is described. For a given perturbation in the sensing geometry the magnitude of the error in the measured BRDF depends on the shape of the BRDF, i.e. on the Sun-target-sensor geometry and on the surface cover type. Using typical BRDF data sets measured in the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE) it is shown that an error of 1° in the sensing geometry could result in approximately 1-5 per cent error in the BRDF. It is suggested that these errors could be reduced by sensing multi-angle data from stabilized platforms and/or by incorporating facilities for recording the sensor viewing geometry.

A distributed snowmelt prediction model in mountain areas based on an energy balance method

A distributed snowmelt prediction model was developed for a mountain area. Topography of the study area was represented by a digital map. Cells On the map were divided into three surface-cover types; deciduous forest, evergreen forest and deforested area. Snowmelt rates for each cell were calculated by an energy balance method. Meteorological elements were estimated separately in each cell according to topographical characteristics and surface-cover type. Distributions of water equivalent of snow cover were estimated by the model. Snowmelt runoff in the watershed was also simulated by snowmelt rates calculated by the model. The model showed thai the snowmelt period and snowmelt runoff after timber harvests would be about two weeks earlier than under the forest-covered condition.

A distributed snowmelt prediction model in mountain areas based on an energy balance method

A distributed snowmelt prediction model was developed for a mountain area. Topography of the study area was represented by a digital map. Cells On the map were divided into three surface-cover types; deciduous forest, evergreen forest and deforested area. Snowmelt rates for each cell were calculated by an energy balance method. Meteorological elements were estimated separately in each cell according to topographical characteristics and surface-cover type. Distributions of water equivalent of snow cover were estimated by the model. Snowmelt runoff in the watershed was also simulated by snowmelt rates calculated by the model. The model showed thai the snowmelt period and snowmelt runoff after timber harvests would be about two weeks earlier than under the forest-covered condition.

Soil Loss as Affected by Different Combinations of Surface Litter and Rock

AbstractThe effects of various surface cover types, their combinations, and bare ground on soil loss from a clay loam topsoil were studied on small plots in the laboratory using a rotating boom rainfall simulator. A total of 105 plots with various types and amounts of surface cover were exposed to simulated rainfall at 100 mm/h for 30 min. Cover treatments included herbaceous litter of sedges (Carex spp.) and bluegrass (Poa spp.), shrub litter of willows (Salix spp.), small rocks, and mixtures of these cover types at several levels of cover (0, 25, 50, 75, 100%) and bare ground (control). Experimental data for high intensity storm were used to develop an equation to quantify soil loss and surface cover relationships for a clay loam soil. Regression analysis indicated high correlation (R2 = 0.86) between soil loss and types and amounts of surface cover. Soil loss from bare plots was about 13 000 kg/ha. A surface cover proportion of about 0.60 for vegetation litter, and 0.85 for rock, would reduce soil loss to about 5000 kg/ha. Rocks were not as effective as vegetation litter in our experiment in protecting the soil from erosion. However, surface coverage of 100% with combinations of vegetation litter and rocks offered the best protection of the soil against erosion. This combination was better than either 100% vegetation litter cover or 100% rock cover alone. Rock in combination with vegetation litter may have increased roughness and allowed deposition of sediment.

Texture-based classification of cloud and ice-cap surface features

It is well known that the interpretation of high resolution (<100 m) visible and near infrared (e.g. Landsat) imagery of large ice masses is hindered by the uniform reflectivity of snow, ice and cloud surfaces. Such interpretation is at present largely performed manually, but there is a good prospect that it could be automated by the incorporation of image texture. This paper describes preliminary work towards the identification of the most appropriate texture technique, or combination of techniques, and assesses the likely performance of such methods.Different textures are identified with different types of surface cover, and the use of these differences to classify images is investigated. Specifically, we compare a traditional texture measure, the Grey Level Co-occurrence Matrix (GLCM), with a modification of a relatively new technique, fractional Brownian motion (FBM). These two methods are applied to three Landsat MSS images of the Nordaustlandet ice cap, Svalbard. The classification accuracy, computation time and memory required, advantages and limitations of the two methods are compared. The GLCM technique appears to be able to distinguish three groups of image classes, namely dry snow, wet snow, and melt features, ablation areas or cloud cover. The FBM technique is computationally more efficient, and though it performs in general less well than the GLCM technique it gives better discrimination of cloud cover.

Texture-based classification of cloud and ice-cap surface features

It is well known that the interpretation of high resolution (<100 m) visible and near infrared (e.g. Landsat) imagery of large ice masses is hindered by the uniform reflectivity of snow, ice and cloud surfaces. Such interpretation is at present largely performed manually, but there is a good prospect that it could be automated by the incorporation of image texture. This paper describes preliminary work towards the identification of the most appropriate texture technique, or combination of techniques, and assesses the likely performance of such methods. Different textures are identified with different types of surface cover, and the use of these differences to classify images is investigated. Specifically, we compare a traditional texture measure, the Grey Level Co-occurrence Matrix (GLCM), with a modification of a relatively new technique, fractional Brownian motion (FBM). These two methods are applied to three Landsat MSS images of the Nordaustlandet ice cap, Svalbard. The classification accuracy, computation time and memory required, advantages and limitations of the two methods are compared. The GLCM technique appears to be able to distinguish three groups of image classes, namely dry snow, wet snow, and melt features, ablation areas or cloud cover. The FBM technique is computationally more efficient, and though it performs in general less well than the GLCM technique it gives better discrimination of cloud cover.

Change detection with synthetic aperture radar

Abstract The purpose of the studv was to explore the information available in a multi-temporal Synthetic Aperture Radar (SAR) data set covering a full seasonal cycle, and to identify issues that must be considered in conducting further research in this area. The data set was compiled from images obtained using the C-band SAR of the Canada Centre for Remote Sensing over a test flight line near Ottawa. All SAR images were geometrically corrected, registered to a UTM projection and resamplcd to 6-25 m pixels. A fall Landsat Thematic Mapper image was also resampled and co-registered with the SAR images. A procedure was developed to display differences between SAR images taken at successive dates. Both individual and difference SAR images were analysed with reference to ground and hydrome-teorological data acquired from various sources. Results of the analysis indicate that SAR images obtained over the annual cycle contain information on the dynamic behaviour of various land surface cover types. Variations were most pervasive in the case of vegetation and secondly in the case of surface water, including vegetation-water interactions. The results also highlighted the requirement of precise geometrical registration, calibrated (relatively or better) SAR and-accurate hydrometeorological data (including within site variability) within the site to detect and analyse changes in SAR images.

The seasonality of AVHRR data of temperate coniferous forests: Relationship with leaf area index

Advanced Very High Resolution Radiometer (AVHRR) data from the National Oceanic and Atmospheric Administration (NOAA)-9 satellite were acquired of the western United States from March 1986 to November 1987. Monthly maximum value composites of AVHRR normalized difference vegetation index (NDVI) [(near infrared — visible)/(near infrared + visible)] were calculated for 19 coniferous forest stands in Oregon, Washington, Montana, and California. The leaf area index (LAI) of the conifer forests explained 70% and 79% of the variation of the summer maximum AVHRR NDVI in July 1986 and July 1987, respectively. The seasonal variation of NDVI was related to phenological changes in LAI, as well as the proportion of surface cover types contributing to the overall reflectance. The varying solar zenith angles in the summer and winter months complicated analyses of the seasonal differences in LAI of the forest stands by reducing NDVI values in the winter months. It is concluded that AVHRR NDVI data from July were related to the seasonal maximum leaf area index of coniferous forests of the western United States, and that seasonal differences in the AVHRR NDVI were related to: a) phenological changes in LAI caused by climate, b) the proportions of surface cover types contributing to the overall reflectance, and c) large variations in the solar zenith angle.

Soil erosion processes and nutrient loss. 1. The interpretation of enrichment ratio and nitrogen loss in runoff sediment

The ratio of nutrient concentration in eroded sediment to that in the original soil (the enrichment ratio, ER) commonly varies with the accumulated soil loss. The objective of this study was to investigate possible factors contributing to this change in ER when erosion was accompanied by a significant depth of water. The enrichment ratio was directly measured on sediment from a sandy clay loam soil. ER was followed as a function of time for eight erosion experiments in which the mix of erosion processes and the fractional surface cover was varied. By using a simulated rainfall tilting flume facility, experiments covered low slope (0.1%), when rainfall detachment was the only erosion process, and 3% soil surface slope, where the processes of rainfall detachment and entrainment occurred. The type and extent of fractional surface cover was varied for the experiments with the 3% slope. In all cases, the rainfall rate was 100 mm h-1and the drop size was 2.2 mm. A new analytical framework is described, showing that ER can be interpreted from the product of two component distributions. The first component distribution is the concentration of sediment as a function of sediment size (a distribution found to vary with time and mix of erosion processes). The second distribution is nitrogen concentration (largely organic) as a function of size (found to be much less time-variable than sediment size). The conclusions reached, after analysis of these experimental data by using this framework, were: (i) time variation in ER was largely due to time variation in the first component distribution; (ii) values of ER different from unity require some variation with sediment size (or settling velocity) in the concentration of the nutrient sorbed to the soil or closely associated with the soil organic matter; (iii) the more that rainfall detachment dominates runoff erosion as the major erosion process, the more likely it is that ER is greater than unity.

Lowest order correction for solar zenith angle to Global Vegetation Index (GVI) data

Abstract NOAA produces vegetation indices as part of a project to develop the uses of meteorological satellite data for global agricultural monitoring (Henderson-Sellers et al. 1986). However, no consideration is given to the variability of vegetation indices with the solar zenith angle. This paper focuses on this particular issue. A brief summary of an inversion technique is presented in which raw values of the normalized difference vegetation indices (NDVIs) for a variety of surface-cover types are simulated as a function of solar zenith angle. A relationship between a change in NDVI and solar zenith angle is presented. This relation is used to correct global vegetation index (GVI) data. The results show that for NOAA-7 and NOAA-9 data there is little correction in the neighbourhood of the equator (± 10) but the amount of correction increases with increasing latitude. Such corrections are also shown to be important in data comparison and integration. For example, in comparing the NOAA-6- and NOAA-8-der...

The effect of atmospheric correction on the interpretation of multitemporal AVHRR-derived vegetation index dynamics

The actual evapotranspiration (AE) and potential evapotranspiration (PE) data for the 1984 vegetation season over the United Kingdom supplied by the U.K. Meteorological Office Rainfall and Evapotranspiration Calculation System (MORECS) are used to indicate moisture stress variation on the vegetation canopy. Raw and atmospherically corrected normalized difference vegetation indices (NDVIs) for the same period are calculated from 10 images obtained from the Advanced Very High Resolution Radiometer (AVHRR) in order to test the effect of the atmosphere on the NDVI dynamics interpretation.. Comparison of the NDVI seasonal cycle (for various surface-cover types) with the moisture stress variation shows that i) there is a marked improvement in our understanding of the vegetation dynamics if remotely sensed data are corrected for atmospheric effects and ii) the relationship between the NDVI and moisture stress seems to depend on the type of vegetation present in the area in question.