Roberta Anniballe

Downscaling Landsat Land Surface Temperature over the urban area of Florence

Abstract A new downscaling algorithm for land surface temperature (LST) images retrieved from Landsat Thematic Mapper (TM) was developed over the city of Florence and the results assessed against a high-resolution aerial image. The Landsat TM thermal band has a spatial resolution of 120 m, resampled at 30 m by the US Geological Survey (USGS) agency, whilst the airborne ground spatial resolution was 1 m. Substantial differences between Landsat USGS and airborne thermal data were observed on a 30 m grid: therefore a new statistical downscaling method at 30 m was developed. The overall root mean square error with respect to aircraft data improved from 3.3 °C (USGS) to 3.0 °C with the new method, that also showed better results with respect to other regressive downscaling techniques frequently used in literature. Such improvements can be ascribed to the selection of independent variables capable of representing the heterogeneous urban landscape.

Comparison of Different Neural Network Approaches for the Tropospheric Profiling over the Inter-tropical lands Using GPS Radio Occultation Data

In this study different approaches based on multilayer perceptron neural networks are proposed and evaluated with the aim to retrieve tropospheric profiles by using GPS radio occultation data. We employed a data set of 445 occultations covering the land surface within the Tropics, split into desert and vegetation zone. The neural networks were trained with refractivity profiles as input computed from geometrical occultation parameters provided by the FORMOSAT-3/COSMIC satellites, while the targets were the dry and wet refractivity profiles and the dry pressure profiles obtained from the contemporary European Centre for Medium-Range Weather Forecast data. Such a new retrieval algorithm was chosen to solve the atmospheric profiling problem without the constraint of an independent knowledge of one atmospheric parameter at each GPS occultation.

Earthquake damage mapping by using remotely sensed data: the Haiti case study

Abstract. This work proposes methodologies aimed at evaluating the sensitivity of optical and synthetic aperture radar (SAR) change features obtained from satellite images with respect to the damage grade due to an earthquake. The test case is the Mw 7.0 earthquake that hit Haiti on January 12, 2010, located 25 km west–south–west of the city of Port-au-Prince. The disastrous shock caused the collapse of a huge number of buildings and widespread damage. The objective is to investigate possible parameters that can affect the robustness and sensitivity of the proposed methods derived from the literature. It is worth noting how the proposed analysis concerns the estimation of derived features at object scale. For this purpose, a segmentation of the study area into several regions has been done by considering a set of polygons, over the city of Port-au-Prince, extracted from the open source open street map geo-database. The analysis of change detection indicators is based on ground truth information collected during a postearthquake survey and is available from a Joint Research Centre database. The resulting damage map is expressed in terms of collapse ratio, thus indicating the areas with a greater number of collapsed buildings. The available satellite dataset is composed of optical and SAR images, collected before and after the seismic event. In particular, we used two GeoEye-1 optical images (one preseismic and one postseismic) and three TerraSAR-X SAR images (two preseismic and one postseismic). Previous studies allowed us to identify some features having a good sensitivity with damage at the object scale. Regarding the optical data, we selected the normalized difference index and two quantities coming from the information theory, namely the Kullback–Libler divergence (KLD) and the mutual information (MI). In addition, for the SAR data, we picked out the intensity correlation difference and the KLD parameter. In order to analyze the capability of these parameters to correctly detect damaged areas, two different classifiers were used: the Naive Bayes and the support vector machine classifiers. The classification results demonstrate that the simultaneous use of several change features from Earth observations can improve the damage estimation at object scale.

Sinergistic use of radar and optical data for agricultural data products assimilation: A case study in Central Italy

The paper describes the preliminary results of the January-August 2015 multi-frequency EO data acquisition campaign conducted over the Maccarese (Central Italy) farm. From January to May radar Cosmo SkyMed Ping-Pong (HH-VV), RapidEye and ZY-3 multispectral VHR optical images, as well as in situ data, have been acquired to retrieve biophysical and/or bio-chemical characteristics of soil and crops. LAI trend has been analyzed and compared by using both polarimetric and optical retrieval algorithms while soil moisture measurements have been compared with the radar backscattering.

Identification of building double-bounces feature in very high resoultion SAR data for earthquake damage mapping

Nowadays very high resolution (VHR) Synthetic Aperture Radar (SAR) systems can provide near real time earthquake damage maps with an high degree of details to stakeholders in charge of managing the emergency phase. However, the increased resolution introduces new challenges to interpret and detect changes in urban areas caused by seismic events. In metric resolution SAR sensors a building appears as a complex of image structures associated to different scattering mechanisms, preventing the use of pixel-based algorithms. In this paper we propose an object oriented approach, focusing the attention on the double-bounce return from buildings, trying to detect damages looking at changes of these particular image patterns. The identification of double-bounce regions is performed using open and close morphological filters and assuming linear structuring elements with different orientation and length. The change detection analysis based on a pre- and a post-event image is carried out using four change detection indicators, such as: intensity ratio, interferometric coherence, intensity correlation and Kullback-Leibler divergence. All change features are extracted using all pixels within each identified object, i.e., double-bounce regions. The test case is the earthquake that hit LAquila city (Italy) on April 6, 2009, while the dataset is composed of two X-band COSMO-SkyMed SAR images acquired before and after the event. A macro-seismic survey map was available to evaluate the obtained results.

A Stable Gaussian Fitting Procedure for the Parameterization of Remote Sensed Thermal Images

An image analysis procedure based on a two dimensional Gaussian fitting is presented and applied to satellite maps describing the surface urban heat island (SUHI). The application of this fitting technique allows us to parameterize the SUHI pattern in order to better understand its intensity trend and also to perform quantitative comparisons among different images in time and space. The proposed procedure is computationally rapid and stable, executing an initial guess parameter estimation by a multiple regression before the iterative nonlinear fitting. The Gaussian fit was applied to both low and high resolution images (1 km and 30 m pixel size) and the results of the SUHI parameterization shown. As expected, a reduction of the correlation coefficient between the map values and the Gaussian surface was observed for the image with the higher spatial resolution due to the greater variability of the SUHI values. Since the fitting procedure provides a smoothed Gaussian surface, it has better performance when applied to low resolution images, even if the reliability of the SUHI pattern representation can be preserved also for high resolution images.

Triple Collocation to Assess Classification Accuracy Without a Ground Truth in Case of Earthquake Damage Assessment

The assessment of satellite image classifications is usually carried out using a test sample assumed as the ground truth, from which a confusion matrix is derived. There are cases where the reference data, even those coming from a ground survey, are affected by errors and do not represent a reliable truth. In the field of geophysical parameter retrieval, the triple collocation (TC) technique is applied for validating remotely sensed products when the source of test data (e.g., ground data) does not represent a reliable reference. TC is able to retrieve the error variances of three systems observing the same target parameter, assuming that their errors are independent. In this paper, we exploit the same idea to test the classification accuracy in cases where the ground truth is not available. We extend the TC approach to the classification problem for a general number of classes, but we solve it numerically for a two-class problem (i.e., collapsed and noncollapsed buildings). The specific case refers to the detection of L’Aquila 2009 earthquake damage from very high-resolution optical data. The image classification, performed by exploiting an object-based analysis, is compared with those from two different ground surveys carried out after the earthquake by different teams and with different purposes. This paper demonstrates the power of the TC approach for assessing the classification accuracy with no reliable ground truth available, and provides an insight into the problem of assessing damage, from satellite and on ground, in a very critical and unsafe situation, like the one occurring after an earthquake. Moreover, it was found that the remotely sensed product can have an order of accuracy comparable to that of at least one of the ground surveys.

Downscaling of the land surface temperature over urban area using Landsat data

In this work, the land surface temperature (LST) retrieval using Landsat Thematic Mapper (TM) data over the city of Florence, Italy, characterized also by the presence of rural and vegetated zones, was compared with a high-resolution (1 m ground pixel size) thermal image provided by an airborne survey made on July 18, 2010. Two Landsat scenes were processed before and after the flight, with the aim to evaluate the impact of the Landsat TM resolution of the thermal channel (120 m) on the LST estimation over an urban texture. The thermal data were downscaled at 30 m using a statistical approach employing spectral indices: such a method highlights the potentials and limits of the LST downscaling performed over an heterogeneous urban area.

Detecting sparse earthquake damages in high density urban settlements by VHR SAR data

Nowadays, space-borne Synthetic Aperture Radar (SAR) sensors, can achieve spatial resolutions in the order of 1 m. However, the exploitation of SAR at very high resolution (VHR) for detecting sparse and isolated damages in urban areas, caused by earthquakes, is still a challenging task. Within urban settlements, the scattering mechanisms are extremely complex and simple change detection analyses or classification procedures can hardly be performed. In this work the 2009, L’Aquila (Italy), earthquake has been considered as case study. Despite about 300 people were killed by the earthquake, few buildings were completely collapsed, and many others were heavily/partially damaged, resulting in a quite sparse damage distribution. We have visually analyzed pairs of VHR SAR data acquired by COSMO-SkyMed satellites, in SPOTLIGHT mode, before and after the earthquake. Such analyses were performed to understand the SAR response of damaged structures surrounded by unaffected buildings, with the aim to identify possible strategies to map the damaged buildings by using an automatic classification procedure. The preliminary analyses based on RGB images, generated by combining pre- and post-event backscattering images, allowed us to figure out how the completely collapsed and the partially damaged buildings are characterized in the SAR response. These outcomes have been taken into account to set up a decision tree algorithm (DTA). Decision rules and related thresholds were identified by statistically analyzing the values of backscattering and derived features. This study point out that many pieces of information and discrimination rules must be exploited to obtain reliable results when dealing with non-extensive and sparse damage within a dense urban settlement.