The intensity of flowering of the holm oak trees is important for the annual phenological monitoring and as a predictive index of final acorn production. Their male flowers present in long catkins of intense yellow color and the estimation of their abundance in the field is a time-consuming task that becomes unfeasible at a large scale. In this work, a methodology to estimate the intensity of flowering of oak trees using RGB (Red Green Blue) images, provided by an unmanned aerial vehicle, was tested. During the spring of 2019, three aerial zenith images of 3 cm spatial resolution were taken in three selected dehesa sites, together with simultaneous ground digital photographs per tree (50 at each site). The intensity of flowering was visually estimated using the ground digital photographs in three categories, ranging from 1 (little or no flowering) to 3 (high flowering). A simple flowering intensity index, based on the closeness to pure yellow within a Cartesian RGB space, was developed to check the relationship between the drone images and the visually analyzed photographs. The results showed that those trees with lower flowering intensity were grouped in higher yellow distances and the high flowering intensity trees in the lower ones. As a result, it can be concluded that this index was able to identify qualitatively the flowering intensity of holm oaks at the farm level and could be useful for future phenological or productivity applications.
KEYWORDS: Vegetation, Remote sensing, Satellites, Biological research, Agriculture, Data modeling, Solar radiation models, Systems modeling, Solar radiation, Sensors
Cover crop in olive orchards is an increasingly applied soil and water conservation strategy, supported by European policies due to its multiple environmental benefits. To quantify these benefits, supervise and encourage the adoption of this practice, robust and affordable monitoring indicators of the cover crop dynamic and its biomass are required. This work represents the first attempt to estimate the biomass produced by olive grove cover crops based on remotely sensed data and an adaptation of the Monteith efficiencies approach. Ten olive tree fields were selected, distributed in three zones of Southern Spain. They comprised a high environmental variability and differed in the herbaceous layer management: cover crop in strips; non-tillage without strips (full coverage); and conventional tillage. An adaptation of the LUE (Light Use Efficiency)- model was applied to estimate Net Primary Production (NPP) using meteorological and Sentinel-2 data and subtracting the contribution of the wooded vegetation from the ground spectral response. The results showed an uneven adjustment in different fields. RMSD was equal to 650 kg ha-1, with an MBD of -17 kg ha-1, indicating a moderately high error (around 39%) but not too much bias. This error suggests that the model requires further refining, including the adjustment of model parameters to better represent this agrosystem. However, the evolution of biomass accumulation throughout the cover crop growing season and the behaviour of the daily biomass production provided interesting keys about the cover crops’ phenology and management, supporting the discrimination between management practices.
The regular monitoring of the evapotranspiration rates and their links with vegetation conditions and soil moisture may support management and hydrological planning leading to reduce the economic and environmental vulnerability of complex water-controlled Mediterranean ecosystems. In this work, the monitoring of water use over a basin with a predominant oak savanna (known in Spain as dehesa) was conducted for two years, 2013 and 2014, monitoring ET at both fine spatial and temporal resolution in different seasons.
A global 5 km daily ET product, developed with the ALEXI model and MODIS day-night temperature difference, was used as starting point. Flux estimations with higher spatial resolutions were obtained with the associated flux disaggregation scheme, DisALEXI, using surface temperature data from the polar orbiting satellites MODIS (1 Km, daily) and Landsat 7/8 (60-120m and sharpened to 30m, 16 days) and the previously estimated coarse resolution fluxes. The results achieved supported the ability of this scheme to accurately estimate daytime-integrated energy fluxes over this system, using input data with different spatio-temporal resolution and without the need for ground observations. Daily ET series at 30 m spatial resolution, generated using STARFM fusion technique, has provided a significant improvement in spatial heterogeneity assessment of the ET series, with RMSE values of 0.56 and 0.68 mm/day for each year, representing an enhancement with respect to interpolated Landsat series. In summary, this approach was demostrated to be robust and operative to map ET at watershed scale with a suitable spatial and temporal resolution for applications over the dehesa ecosystem.
The dehesa, the most widespread agroforest ecosystem in Europe (≈ 3 million ha), is recognized as an example of
sustainable land use and for its importance in rural economy. It is characterized by widely-spaced oak trees (mostly
Quercus Ilex L.), combined with crops, pasture and shrubs in the sub-canopy region. The estimation of the ecosystem
evapotranspiration (ET) using remote sensing may assist the monitoring of its state from local to regional scales,
improving the management and the conservation of the ecosystem. Thermal-based energy balance techniques which
distinguish soil/substrate and vegetation contributions to the radiative temperature and radiation/turbulent fluxes have
proven to be reliable in the estimation of the energy surface fluxes, and therefore in the estimation of ET. In particular,
the two-source energy balance (TSEB) model of Norman et al. and Kustas and Norman has shown to be robust for
semi-arid sparse canopy-cover landscapes. With the objective of evaluating the model over this environment, an energy
flux measurement system has been used. It was installed in a dehesa located in Southern Spain (38°12′ N; 4°17′ W, 736m a.s.l) with 1 km homogeneous fetch in wind direction. The quality of the measured data fluxes has been tested with the energy-balance closure criterion yielding an average closure of 86% which is within the error range found in similar studies. The TSEB model was evaluated in the area for 2012 summer season, using images from MODIS (Moderate
Resolution Imaging Spectroradiometer) sensor and ground measured meteorological data. The half-hourly estimates
were compared with the flux tower measurements, obtaining a RMSD between modeled and measured energy fluxes
within the closure balance error.
Spatio-temporal changes in vegetation at the basin scale are difficult to characterize, and remote sensing is a major
source of data for this purpose. These sensors may provide distributed series of spectral properties of the vegetation with
different spatial and temporal resolutions, but they do not always satisfy the requirements of some of the applications.
These limitations can be overcome with the use of image integration techniques, which allow for the combination of
sensors with different characteristics. This work presents the monitoring of the vegetation cover in the Guadalfeo River
Basin (Spain), with a view to its hydrological modeling, by using Landsat-TM and MODIS data, analyzing the
implications of the scale differences in an heterogeneous area. A preliminary study is carried out into the deviations of
NDVI and ground cover fraction (fv) between the concurrent data of both sensors. Thereafter, the STARFM integration
algorithm is applied and evaluated to obtain synthetic NDVI images at the spatial resolution of Landsat-TM data with
MODIS time steps. The comparison between Landsat-TM and MODIS parameters revealed deviations on average
between 2-5% for NDVI and 3-5% for fv. No direct relationship was found between these deviations and basin
topography. However, higher deviations corresponded with the vegetation types with higher ground cover fractions and
heterogeneous landuses (fv relative deviations of 10% and 6% for conifers and quercus-scrub, respectively) The
STARFM algorithm improved the NDVI estimations when compared to the previous Landsat-TM date, with reductions
in the average NDVI differences of around 0.02 on average for the six simulated dates, with the accuracy of the
predictions depending on data input for the model and vegetation cover types.
A two-source energy balance model that separates surface fluxes of the soil and canopy was applied to a drip-irrigated
vineyard in central Spain, using a series of nine Landsat-5 images acquired during the summer of 2007. The model
partitions the available energy, using surface radiometric temperatures to constrain the sensible heat flux, and computing ET as a residual of the energy balance. Flux estimations from the model are compared with half-hourly and daily values obtained by an eddy covariance flux tower installed on the site during the experiment. The performance of the twosource model to estimate ET under the low vegetation cover and semiarid conditions of the experiment, with RMSD between observed and model data equal to 49 W m-2 for half-hourly estimations and RMSD=0.5 mm day-1 at daily scale, is regarded as acceptable for irrigation management purposes. Model results in the separation of the beneficial (transpiration) and non-beneficial (evaporation from the soil) fractions, which is key information for the quest to improve water productivity, are also reported. However, the lack of measures of these components makes it difficult to draw conclusions about the final use of the water.
Evapotranspiration (ET) is a critical variable in hydrological processes and an accurate estimation of the rate of
evapotranspiration is required if we wish to apply integrated management procedures to water resources. This study
offers new insights into remote sensing-based models that estimate ET at basin scale, evaluating the combination of a
surface energy balance based on thermal remote sensing and the use of the crop coefficient (Kc), a simple operational method that is widely used in irrigated agriculture. The study area is the Guadalfeo river basin in southern Spain, a large watershed with major topographical and landscape contrasts. Reference evapotranspiration (ETo) surfaces were generated by applying the FAO56-PM [1] equation, and real ET surfaces were estimated following a two-source energy balance model [2] [3]. Crop and vegetation coefficients were obtained as the ratio between ET and ETo. Kc maps were analysed in terms of vegetation type and development. The resulting coefficients generally ranged between 0.1 and 1.5, and could be directly related to vegetation ground cover for the main vegetation types, including natural vegetation and crops, with the determination coefficient (r2) lying between 0.77 and 0.97 in both humid and dry seasons. Relationships based on these coefficients are proposed as a simple proxy to monitor the water use of the basin on a regular basis by means of optical remote sensors alone, providing data with higher frequency and spatial resolution than can be obtained by thermal measurements; data that could complement thermal sensors whenever these were available.
The integrated water resource management required to face the water scarcity situation in semiarid regions relies on the
ability to obtain accurate information about the use of water by crops and natural vegetation. Thermal remote sensing
provides key data about the vegetation water status. The integration of this remotely sensed data into water and energy
balance models help to better estimate evapotranspiration under heterogeneous cropping and natural vegetation patterns,
extending the field of application of these models from point to basin and regional scales.
In this work, we present an approach to estimate spatially distributed surface energy fluxes using a series of Landsat TM
satellite images combined with simulation modeling and ground-based measurements. A physically-based method for the
energy budget partitioning following the Two Source Model [1, 2] has been applied over an heterogeneous agricultural
area located in southern Spain. The study was performed during 2009 crop growing season and the results were validated
with field data collected with an eddy covariance system installed over a corn field during the season. The instantaneous
and daily estimations were compared to the measured data, obtaining a general good adjustment at both scales and
setting the basis for a larger scale application that may assist a decision - making tool for water resources planning in the
region.
Rainfall interception by the vegetation may constitute a significant fraction in the water budget at local and watershed
scales, especially in Mediterranean areas. Different approaches can be found to model locally the interception fraction,
but a distributed analysis requires time series of vegetation along the watershed for the study period, which includes both
type of vegetation and ground cover fraction. In heterogeneous watersheds, remote sensing is usually the only viable
alternative to characterize medium to large size areas, but the high number of scenes necessary to capture the temporal
variability during long periods, together with the sometimes extreme scarcity of data during the wet season, make it
necessary to deal with a limited number of images and interpolate vegetation maps between consecutive dates.
This work presents an interception model for heterogeneous watersheds which combines an interception continuous
simulation derived from Gash model and their derivations, and a time series of vegetation cover fraction and type from
Landsat TM data and vegetation inventories. A mountainous watershed in Southern Spain where a physical hydrological
modelling had been previously calibrated was selected for this study. The dominant species distribution and their
relevant characteristics regarding the interception process were analyzed from literature and digital cartography; the
evolution of the vegetation cover fraction along the watershed during the study period (2002-2005) was produced by the
application of a NDVI analysis on the available scenes of Landsat TM images. This model was further calibrated by field
data collected in selected areas in the watershed.
The integration of time series of high-resolution remote sensing images in the FAO crop evapotranspiration (ET) model
is receiving growing interest in the last years, specially for operational applications in irrigated areas. In this study, a
simplified methodology to estimate actual ET for these areas in large watersheds was developed. Then it was applied to
the Guadalquivir river watershed (Southern Spain) in the 2007 and 2008 irrigation seasons. The evolution of vegetation
indices, obtained from 10 Landsat and IRS images per season, was used for two purposes. Firstly, it was used for
identifying crop types based on a classification algorithm. This algorithm used training data from a screened subset of
the information declared by farmers for EU agriculture subsidies purposes. Secondly, the vegetation indices were used to
obtain basal crop coefficients (Kcb, the component of the crop coefficient that represents transpiration). The last step was
the parameterization of the influence of evaporation from the soil surface, considering the averaged effect of a given rain
distribution and irrigation schedule. The results showed only small discrepancies between the crop coefficients
calculated using the simplified model and those calculated based on a soil water balance and the dual approach proposed
by FAO. Therefore, it was concluded that the simplified method can be applied to large irrigation areas where detailed
information about soils and/or water applied by farmers lacks..
Many marsh areas in Southern Spain were dewatered during the 1950's for agricultural purposes. These actions were not
successful due to the high salinity trend exhibited in such soils, especially notable during the long dry summers in these
locations. Recently, many attempts to restore the marshes have been made to try to return the original flooding cycles to
the dewatered areas, and promote the development of spontaneous vegetation suitable to salty environments. This work
deals with the monitoring of the increase of the flooding area in the San Pedro River marshes (Cádiz) in Spain after the
demolition of a dam near the mouth, from the analysis of Landsat TM images with a linear mixture spectral model.
Three different components (vegetation, dry soil and wet soil) were quantified in the area over the two years following
the destruction of the dam and the increase in tidal entry to the marsh and compared to the results from a previous date.
The results were calibrated with field data measured directly on the terrain surface. The model used was capable of
discriminating such components with satisfactory accuracy, providing data on the evolution of the flooding area
throughout the year and the increase in vegetation distribution one year after the dam break. Differences in the tidal
advance along tidal creeks in the main reach of the river before and after the demolition were successfully identified. The
impact of the dam action on the development of vegetation was also quantified; the results showed the potential to
restore this degraded marsh land.
Two common approaches for estimating crop evapotranspiration (ET) using satellite imagery are the reflectance-based
crop coefficient method and the energy balance method. The reflectance-based crop coefficient method
relates a reflectance-based vegetation index such as the soil adjusted vegetation index (SAVI) to ET basal crop
coefficients such as those described by Wright (1982) [1] and the FAO 56 manual [2]. A time-series of remotely
sensed inputs is then used to build the crop coefficient curve in each field being monitored. In order to obtain actual
ET, a water balance must be maintained in the root zone of the crop in order to make the appropriate adjustments due
to soil moisture deficits and wet soil surface from irrigation and/or rain. Ground meteorological data must be
provided by a weather station located in the modeled area for the estimation of reference ET. In the energy balance
approach, surface temperatures are used in the estimation of sensible heat fluxes and depending on the complexity of
the model, different methods are used to either handle the aerodynamic temperature term or deal with sparse
canopies (empirical approaches, two-source model, SEBAL model). Remotely sensed inputs are also used for the
estimation of net radiation and soil heat flux, with latent heat flux (ET) obtained as a residual from the energy
balance equation. The energy balance approach results in the actual ET being estimated directly. Instantaneous
values of ET must be extrapolated to the entire day and over time in between satellite overpass inputs. This paper
describes a hybrid approach that uses both methods in combination to monitor actual ET over a growing season for
irrigated and non-irrigated crops. The model has been coded in an ArcGIS environment, using visual basic for the
calculations. This paper describes the modeling environment and coded ET models within and presents some
application results.
One of the approaches of estimating crop evapotranspiration over large areas using remote sensing is the use of canopy
reflectance (vegetation indices) derived from multi-temporal satellite imagery to estimate and update evapotranspiration
crop coefficients. When this method is applied after the irrigation season is over, a spectral crop classification using one
or more of the images can be conducted to produce a crop type map of the entire area, allowing the application of the
appropriate crop coefficients on a field-by-field basis. However, if the application is to be run in real-time during an
irrigation season using satellite images as they become available, a different classification scheme is required as early
season images might not be optimally suited for a traditional spectral classification. This paper presents a real-time
method of classification based on a combination of spectral classification and logic using the prior knowledge of the crop
types and growth curves in the region. The method is applied to images acquired every two weeks over the 2004
irrigation season at the Lebrija Irrigation District on the Guadalquivir River in Southern Spain. Ground truth information
was provided by the local irrigation district.
Remote sensing of evapotranspiration has become more common during the last decade. Two of the approaches
being used are the reflectance-based crop coefficient method and the energy balance method. In the energy balance
approach, surface temperature is used to calculate sensible heat flux and long wave radiation and depending on the
complexity of the model, different methods are used to handle the aerodynamic temperature term. This paper
compares two energy balance approaches with different methodology for estimating sensible heat flux (H): (i) one
layer energy balance model and (ii) two-layer energy balance model called the Two Source Model. The results from
the two approaches are compared using a set of comprehensive field and remote sensing measurements of model
input data and actual evapotranspiration measured with a dense network of eddy covariance stations during the
SMACEX campaign in central Iowa during the 2002 growing season. Results show that both methods of estimating
H perform well using calibrated Landsat Thematic Mapper imagery as remotely sensed inputs.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.