Carole Delenne

Textural approaches for vineyard detection and characterization using very high spatial resolution remote sensing data

Vine‐plot mapping and monitoring are crucial issues in land management, particularly for areas where vineyards are dominant like in some French regions. In this context, the availability of an automatic tool for vineyard detection and characterization would be very useful. The objective of the study is to compare two different approaches to meet this need. The first one uses directional variations of the contrast feature computed from Haralicks co‐occurrence matrices and the second one is based on a local Fourier transform. For each pixel, a ‘vine index’ is computed on a sliding window. To foster large‐scale applications, test and validation were carried out on standard very high spatial resolution remote sensing data. 70.8% and 86% of the 271 plots of the study area were correctly classified using the co‐occurrence and the frequency method, respectively. Moreover, the latter enabled an accurate determination (less than 3% error) of inter‐row width and row orientation.

An Automatized Frequency Analysis for Vine Plot Detection and Delineation in Remote Sensing

The availability of an automatic tool for vine plot detection, delineation, and characterization would be very useful for management purposes. An automatic and recursive process using frequency analysis (with Fourier transform and Gabor filters) has been developed to meet this need. This results in the determination of vine plot boundary determination and accurate estimation of interrow width and row orientation. To foster large-scale applications, tests and validation have been carried out on standard very high spatial resolution remotely sensed data. About 89% of vine plots are detected corresponding to more than 84% of vineyard area, and 64% of them have correct boundaries. Compared with precise on-screen measurements, vine row orientation and interrow width are estimated with an accuracy of 1deg and 3.3 cm, respectively.

Uncertainty analysis of river flooding and dam failure risks using local sensitivity computations

The potential of Local Sensitivity Analysis (LSA) for analysis of uncertainty with respect to two major risks in river hydrodynamics -flash flood and dam failure -is assessed. LSA, implemented as an equation-based method, is compared to a Global Uncertainty Analysis (GUA) consisting in running Monte Carlo simulations with a hydrodynamic model. For a given statistical distribution of the model input parameters, the mean and standard devi-ation of the output variables are estimated with the two methods. In all single or multiple parameter cases investigated, including as much as ±80% relative variation, LSA provides similar results to GUA, while requiring only one simulation instead of several hundreds or thousands. Only within a few meters of the shock (flow discontinuity) generated by the breaking of a dam do the two methods depart. This paper shows that despite the non-linearity of river flow processes, the first order, local approach remains generally valid for uncertainty analysis of hydrodynamic risks, even in the case of large parameter uncertainty. The contrast in importance of the various parameters on both sides of a shock is also high-lighted.

Finite-volume solution of one-dimensional shallow-water sensitivity equations

Solving the Shallow-Water-Sensitivity Equations for discontinuous flows involves the discretization of a Dirac source term accounting for discontinuities. Failing to account for this source term usually results in solution instability, with empirical sensitivity solutions exhibiting artificial peaks in the neighbourhood of shocks. An extension of the Harten-Lax-van Leer approximate Riemann solver is presented that allows the one-dimensional shallow-water-sensitivity equations to be discretized more accurately than in previously published versions. A discretization of boundary conditions and source terms is also provided. The proposed discretization allows for discontinuities in both the hydraulic and sensitivity boundary conditions. Numerical experiments indicate the superiority of the proposed approach for sensitivity analysis over the classical, empirical approach if the flow solution is discontinuous. A numerical convergence analysis demonstrates that the numerical and analytical solutions converge.

Detection of manhole covers in high-resolution aerial images of urban areas by combining two methods

The detection of small objects from aerial images is a difficult signal processing task. To localise small objects in an image, low-complexity geometry-based approaches can be used, but their efficiency is often low. Another option is to use appearance-based approaches that give better results but require a costly learning step. In this paper, we treat the specific case of manhole covers. Currently many manholes are not listed or are badly positioned on maps. We implement two conventional previously published methods to detect manhole covers in images. The first one searches for circular patterns in the image while the second uses machine learning to build a model of manhole covers. The results show non optimal performances for each method. The two approaches are combined to overcome this limit, thus increasing the overall performance by about forty percent.

Testing the Relevance of Daily MODIS Data to Monitor Mediterranean Shrubland Canopy Water Content with Temporal Cross-Correlation Analyses

Drought alert systems for forest fire prevention often rely on vegetation water content (VWC) monitoring which is a key parameter in forest fire hazard. In southern France, VWC is up to now monitored through regular field surveys. Thanks to the theoretical sensitivity of shortwave infrared reflectance to VWC, MODIS satellite data are potentially able to monitor VWC depending on plant species VWC magnitude. In this paper, a specific statistical approach based on temporal cross-correlations is developed in order to test the correlation between two MODIS water indices and VWC measurements coming from field surveys. This test aims in assessing the ability of daily MODIS data to monitor Mediterranean shrubland canopy water content and detect any delay effect between MODIS and field survey temporal series. Statistical tests are carried out for 29 sites containing 18 dominant shrubland Mediterranean species. 67 % and 54 % of significant correlation were found using respectively the NDII and NDWI indices from MODIS data. Correlation were found low (0.5 maximum) with a dominant negative delay effect, i.e. with a MODIS signal that reacts a few days after the field VWC. Test results show that, even if deeper pre-processing of MODIS data may be required, site soil, site vegetation cover and heterogeneity at MODIS pixel scale, as well as species VWC sensitivity make correlation between field VWC and MODIS water indices non univoque and highly variable. It stems from this study that many obstacles are still to overcome, for an accurate monitoring of Mediterranean shrubland canopy water content using MODIS daily data.

Detection of Manhole Covers in High-Resolution Aerial Images of Urban Areas by Combining Two Methods

Mispositioning of buried utilities is an increasingly important problem both in industrialized and developing countries because of urban sprawl and technological advances. However, some of these networks have surface access traps, which may be visible on high-resolution airborne or satellite images and could serve as presence indicators. We put forward a methodology to detect manhole covers and grates on very high-resolution aerial and satellite images. Two methods are tested: the first is based on a geometrical circular filter, whereas the second one uses machine learning to retrieve some patterns. The results are compared and combined to benefit from the two approaches.

Frequency analysis for vineyard detection and characterization from airborne images

Fourier analysis has a great amount of applications. For example, it appears to be a very suited tool for vineyard detection in aerial images, due to the periodic patterns induced by this culture. Vine‐plot mapping and monitoring are crucial issues in land management, particularly for areas where vineyards are dominant, like in some French regions. In this context, the availability of an automatic tool for vineyard detection and characterization would be very useful. A method based on a local Fourier Transform has been developed to meet this need. Fast Fourier Transform is computed on a sliding window, providing information concerning vineyard presence in the window as well as an estimation of interrow width and row orientation. These characteristics are used to compute a homogeneous ‘Vine index’, which allows a good discrimination of both class ‘vine’ and ‘non‐vine’. To foster large‐scale applications, tests and validation were carried out on standard very high spatial resolution remote‐sensing data. Usi...