Diego Fernández-Prieto

Targeted Land-Cover Classification

This paper addresses a specific typology of land-cover classification problems, hereinafter referred to as “targeted land-cover classification,” where the objective is the identification of only one or few specific “targeted” land-cover classes of interest, disregarding all the other potential classes present in the area under analysis. Such a challenging problem, which is common to a variety of operational information services and applications (e.g., agriculture, forestry, spatial planning, ecosystem monitoring, disaster management, habitat mapping, etc.), can be effectively solved by traditional supervised classification techniques provided that an exhaustive ground truth is available for all the land-cover classes present in the region of interest. Such a requirement is seldom satisfied and presents several practical drawbacks and limitations, both in terms of time and economic cost that may render this task difficult to achieve in most real-life cases. However, the possibility to perform an effective targeted classification using only ground-truth samples for the class(es) of interest (hence avoiding the burden and cost associated with the collection of a full and exhaustive ground-truth information) would represent a significant advantage. In this paper, we present a novel technique capable of identifying specific land-cover classes of interest by exploiting the ground truth only available for these targeted classes, while providing accuracies comparable to those of traditional fully supervised methods. The proposed technique jointly exploits both the unlabeled samples of the image under investigation and the training samples only available for the targeted classes. In particular, the expectation-maximization algorithm and Markov random fields are employed to estimate the probability density function of both the class(es) of interest and the unknown class representing the merger of all the unknown land-cover classes characterizing the study area for which no ground-truth information is available. An extensive experimental analysis and cross-comparisons with both fully supervised support vector machines and ensembles of multiple one-class support-vector-data-description classifiers on different data sets confirmed the effectiveness and the reliability of the proposed technique.

First results of the earth observation water cycle multi-mission observation strategy (WACMOS)

Observing and monitoring the different components of the global water cycle and their dynamics are essential steps to understand the climate of the Earth, forecast the weather, predict natural disasters like floods and droughts, and improve water resources management. Earth observation technology is a unique tool to provide a global understanding of many of the essential variables governing the water cycle and monitor their evolution from global to basin scales. In the coming years, an increasing number of Earth observation missions will provide an unprecedented capacity to quantify several of these variables on a routine basis. However, this growing observational capacity is also increasing the need for dedicated research efforts aimed at exploring the potential offered by the synergies among different and complementary EO data records. In this context, the European Space Agency (ESA) launched the Water Cycle Multi-mission Observation Strategy (WACMOS) in 2009 aiming at enhancing, developing and validating a novel set of multi-mission based methods and algorithms to retrieve a number of key variables relevant to the water cycle. In particular the project addressed four major scientific challenges associated to a number of key variables governing the water cycle: evapotranspiration, soil moisture, cloud properties related to surface solar irradiance and precipitation, and water vapour. This paper provides an overview of the scientific results and findings with the ultimate goal of demonstrating the potential of strategies based on utilizing multi-mission observations in maximizing the synergistic use of the different types of information provided by the currently available observation systems and establish the basis for further work.

Editorial "Advances in Earth observation for water cycle science"

Since observing the Earth from space became possible more than forty years ago, satellite Earth Observation (EO) missions have become central to the monitoring and understanding of the Earth system, its different components and how they interact with each other. The continuous growth and improvements in the quality of the data and information provided by satellites has resulted in significant progress and advances in a broad range of scientific and application areas including the understanding and characterisation of the global water cycle hydrology and water management. The water cycle is a complex process driven mainly by solar radiation. The evaporation of water from open water and wet soil surfaces is controlled by energy and water availability and near-surface atmospheric conditions (air temperature, humidity and wind-speed), while transpiration of water is primarily controlled by vegetation. The result of evaporation and transpiration is the presence of water vapour in the atmosphere, a prerequisite for cloud formation. If cloud condensation nuclei are present and if the atmospheric state allows for condensation, clouds are formed which are then globally distributed by winds. In the presence of precipitating clouds, water returns back to the Earth’s surface where it accumulates in rivers, lakes and oceans. Surface water may also infiltrate into the soil, moistening the soil layers and accumulating as groundwater replenishing aquifers. Aquifers can store water for several (thousands of) years, provide water for human activities, or discharge it naturally to the surface or to the oceans. The response of the hydrological cycle to global warming is expected to be far reaching (Bengtsson, 2010), and because different physical processes control the change in water vapour and consequently evaporation and precipitation, a more extreme distribution of precipitation is expected leading to, in general, wet areas become wetter and dry areas become dryer (IPCC, 2008). In this context, relying on accurate and continuous observations of the long-term dynamics of the different key variables governing the energy and water cycle processes from global to local scale is essential to further increase not only our understanding of the different components of the water cycle both in its spatial and temporal variability, but also to characterise the processes and interactions between the terrestrial and atmospheric aspects of the energy and water cycle, and how this coupling may influence climate variability and predictability. Such global and continuous observations can only be secured by the effective use of Earth Observation (EO) satellites as a major complement to in-situ observation networks. In the years to come, EO technology will enter into a new era, where the increasing number of more sophisticated missions will provide scientists with an unprecedented capacity to observe and monitor the different components of the water cycle from the local to the global scales. Already today, global observations of several key parameters governing the global water cycle (e.g. precipitation, soil moisture, water vapor, evaporation and transpiration, water levels, gravityderived groundwater measurements, etc. ...) are feasible. In addition, significant progress has been made in the area of data assimilation enhancing the capabilities to integrate EObased products into suitable land surface and hydrological models; hence opening new opportunities for science and applications.

A first estimation of SMOS-based ocean surface T-S diagrams

A first estimation of satellite-based ocean surface T-S diagrams is performed by using SMOS Sea Surface Salinity (SSS) and OSTIA Sea Surface Temperature (SST) and comparing them with in situ measurements interpolated fields obtained by the Argo-buoys for the North Atlantic and over the entire year 2011. The key objectives at the base of this study are: (1) To demonstrate the feasibility of generating routinely satellite-derived surface T-S diagrams, obviating the lack of extensive sampling of the surface open ocean, (2) To display the T-S diagrams variability and the distribution/dynamics of SSS, altogether with SST and the relative density with respect to in situ measurements, and (3) To assess the SMOS SSS data added value in detecting geophysical signals not sensed/resolved by the Argo measurements. To perform the latter analysis, the satellite-Argo mismatches have been overlapped with geophysical parameters of precipitation rates, surface heat and freshwater fluxes and wind speed data. Ongoing and future efforts focus on enlarging the study area and the temporal frame of the analysis and aim at developing a method for the systematic identification of surface water masses formation areas by remotely sensed data.

How do Spatial Scale, Noise, and Reference Data affect Empirical Estimates of Error in ASAR-Derived 1 km Resolution Soil Moisture?

The performance of the advanced synthetic aperture radar (ASAR) global mode (GM) surface soil moisture (SSM) data was studied over Australia by means of two widely used bivariate measures, the root-mean-square error (RMSE) and the Pearson correlation coefficient (R). By computing RMSE and at multiple spatial scales and for different data combinations, we assessed how, and at which scales, the spatial sampling error, noise, and the choice of the reference data impact on RMSE and . The results reveal large changes in RMSE and with continental average values of 8% and 18% for the RMSE of relative soil moisture saturation and between 0.4 and 0.7 for depending on the spatial scale of aggregation and the choice of reference data. The combined effect of noise and spatial sampling error accounted for a 79% RMSE increase at 1 km and predominated over the error due to the choise of the reference data also at 5 km scale. The effect of noise on RMSE strongly diminished at spatial scales ≥2 km. By contrast, the impact of uncertainties in the reference data was larger on than on RMSE. This highlights the better potential of to estimate the benefit of observations prior to data assimilation. Based on our results, it is further suggested that a potential way for an improved ASAR GM SSM error assessment is to: 1) aggregate the data to ≥2 km resolution to minimize the noise; 2) subtract the spatial sampling error within the coarse resolution footprint; and 3) remove the reference uncertainty using advanced techniques such as triple collocation.

Remote sensing of surface ocean PH exploiting sea surface salinity satellite observations

The overall process commonly referred to as Ocean Acidification (OA) is nowadays gathering increasing attention for its profound impact at scientific and socio-economic level. To date, the majority of the scientific studies into the potential impacts of OA have focused on models and in situ datasets. Satellite remote sensing technology have yet to be fully exploited and could play a significant role by providing synoptic and frequent measurements for investigating OA processes on global scales. Within this context, the purpose of the ESA “Pathfinders-OA” project is to quantitatively and routinely estimate surface ocean pH by means of satellite observations in several ocean regions. Satellite Ocean Colour, Sea Surface Temperature and Sea Surface Salinity data (with an emphasis on the latter) will be exploited. A proper merging of these different datasets will allow to compute at least two independent proxies among the seawater carbonate system parameters and therefore obtain the best educated guess of the surface ocean pH. Preliminary results of the anomaly and variability of the ocean pH maps are presented.

Preliminary results of SMOS salinity retrieval by using Support Vector Regression (SVR)

A prospective sounding of the capabilities of a novel salinity retrieval by means of Support Vector Regression has been performed. Co-located SMOS measurements and additional auxiliary parameters have been considered, whilst salinity data collected by ARGO buoys represented the ground-truth to be matched by the algorithm. Salinity fields estimated by the SVR are in good agreement with the ground-truth, suggesting that the chosen approach can be promising, despite its robustness and versatility needs to be assessed over wider areas and time lags, and in various combinations of SMOS features.

Derivation of an experimental satellite-based T-S diagram

A preliminary attempt of deriving a purely satellite-based Temperature-Salinity (T-S) diagram is presented, with the overall aim of assessing to what extent is possible, and in which geographical areas, to identify and trace water masses by satellite. This has been performed by using recent SMOS and Aquarius satellite SSS products in conjunction with spaceborne SST data. A baseline T-S diagram is arranged from climatology data, differentiating 7 ocean zones and mapping them into the T-S domain. Therefore, a comparison with satellite data is carried out, highlighting, for this preliminary test, which are the most challenging zones and where, in turn, they mutually agree in a reasonable way.

Analyzing the Mediterranean Water Cycle Via Satellite Data Integration

The water cycle components are being retrieved by an increasing number of satellite missions. However, the monitoring of the water cycle by satellite Earth Observations is still a challenge. Data sets suffer from numerous systematic and random errors and they are often not coherent with each other. We focus here on the Mediterranean basin, one of the regions most sensitive to climate change. A satellite-based analysis of the water cycle is undertaken using a collection of available satellite data sets. Our satellite data set combination uses a simple bias correction and weighted average, and provides a better water budget closure results than any raw satellite data set. Our almost purely satellite data set allows to better describe the full water cycle, not only over the continents, but also in the atmosphere and over the ocean. The limitation/possibilities of this satellite multi-component data set are described: (1) although improved, the water cycle is still not closed by satellite data and the satellite community should focus on this issue, (2) our combined data set shows good coherency with the ERA-I reanalysis which is the reference so far, both in terms of seasonal climatology and long-term trends. This means that, even if the water budget is not yet closed by satellite data, our monitoring of the water cycle using satellite observations is improving, even over complex regions such as the Mediterranean basin.

How much water is used for irrigation? A new approach exploiting coarse resolution satellite soil moisture products

Abstract Knowledge of irrigation is essential for ensuring food and water security, and to cope with the scarcity of water resources, which is expected to exacerbate under the pressure of climate change and population increase. Even though irrigation is likely the most important direct human intervention in the hydrological cycle, we have only partial knowledge on the areas of our planet in which irrigation takes place, and almost no information on the amount of water that is applied for irrigation. In this study, we developed a new approach exploiting satellite soil moisture observations for quantifying the amount of water applied for irrigation. Through the inversion of the soil water balance equation, and by using satellite soil moisture products as input, the amount of water entering into the soil, and hence irrigation, is determined. Through synthetic experiments, we first assessed the impact of soil moisture measurement uncertainty and temporal resolution, also as a function of climate, on the accuracy of the method. Second, we applied the proposed approach to currently available coarse resolution satellite soil moisture products retrieved from the Soil Moisture Active and Passive mission (SMAP), the Soil Moisture and Ocean Salinity (SMOS) mission, the Advanced SCATterometer (ASCAT), and the Advanced Microwave Scanning Radiometer 2 (AMSR-2). Nine pilot sites in Europe, USA, Australia and Africa were used as case study to test the method in a real-world application. The synthetic experiment showed that the method is able to quantify irrigation, with satisfactory performance from satellite data with retrieval errors lower than ∼0.04 m³/m³ and revisit times shorter than 3 days. In the case studies based on real satellite data, qualitative assessments (due to missing in situ irrigation observations) showed that over regions in which satellite soil moisture products perform well, and which are characterized by prolonged periods without rainfall, the method shows good results in quantifying irrigation. However, at sites in which rainfall is sustained throughout the year, the proposed method fails in obtaining reliable performances. Similarly, low performances are obtained in areas where satellite products uncertainties are too large, or their spatial resolution is too coarse with respect to the size of the irrigated fields.