Domain Adaptation in remote sensing: increasing the portability of land-cover classifiers

Among the types of remote sensing acquisitions, optical images are certainly one of the most widely relied upon data sources for Earth observation. They provide detailed measurements of the electromagnetic radiation reflected or emitted by each pixel in the scene. Through a process termed supervised land-cover classification, this allows to automatically yet accurately distinguish objects at the surface of our planet. In this respect, when producing a land-cover map of the surveyed area, the availability of training examples representative of each thematic class is crucial for the success of the classification procedure. However, in real applications, due to several constraints on the sample collection process, labeled pixels are usually scarce. When analyzing an image for which those key samples are unavailable, a viable solution consists in resorting to the ground truth data of other previously acquired images. This option is attractive but several factors such as atmospheric, ground and acquisition conditions can cause radiometric differences between the images, hindering therefore the transfer of knowledge from one image to another. The goal of this Thesis is to supply remote sensing image analysts with suitable processing techniques to ensure a robust portability of the classification models across different images. The ultimate purpose is to map the land-cover classes over large spatial and temporal extents with minimal ground information. To overcome, or simply quantify, the observed shifts in the statistical distribution of the spectra of the materials, we study four approaches issued from the field of machine learning. First, we propose a strategy to intelligently sample the image of interest to collect the labels only in correspondence of the most useful pixels. This iterative routine is based on a constant evaluation of the pertinence to the new image of the initial training data actually belonging to a different image. Second, an approach to reduce the radiometric differences among the images by projecting the respective pixels in a common new data space is presented. We analyze a kernel-based feature extraction framework suited for such problems, showing that, after this relative normalization, the cross-image generalization abilities of a classifier are highly increased. Third, we test a new data-driven measure of distance between probability distributions to assess the distortions caused by differences in the acquisition geometry affecting series of multi-angle images. Also, we gauge the portability of classification models through the sequences. In both exercises, the efficacy of classic physically- and statistically-based normalization methods is discussed. Finally, we explore a new family of approaches based on sparse representations of the samples to reciprocally convert the data space of two images. The projection function bridging the images allows a synthesis of new pixels with more similar characteristics ultimately facilitating the land-cover mapping across images.

Effects of changing water availability on land use in irrigated mountain oases of Al Jabal Al Akhdar, northern Oman

In Oman, during the last three decades, agricultural water use and groundwater extraction has dramatically increased to meet the needs of a rapidly growing population and major changes in lifestyle. This has triggered agricultural land-use changes which have been poorly investigated. In view of this our study aimed at analysing patterns of shortterm land-use changes (2007-2009) in the five irrigated mountain oases of Ash Sharayjah, Al’Ayn, Al’Aqr, Qasha’ and Masayrat ar Ruwajah situated in the northern Oman Hajar mountains of Al Jabal Al Akhdar where competitive uses of irrigation water are particularly apparent. Comprehensive GIS-based field surveys were conducted over three years to record changes in terrace use in these five oases where farmers have traditionally adapted to rain-derived variations of irrigation water supply, e.g. by leaving agricultural terraces of annual crops uncultivated in drought years. Results show that the area occupied with field crops decreased in the dry years of 2008 and 2009 for all oases. In Ash Sharayjah, terrace areas grown with field crops declined from 4.7 ha (32.4 % of total terrace area) in 2007 to 3.1 ha (21.6 %) in 2008 and 3.0 ha (20.5 %) in 2009. Similarly, the area proportion of field crops shrunk in Al’Ayn, Qasha’ and Masayrat from 35.2, 36.3 and 49.6 % in 2007 to 19.8, 8.5 and 41.3 % in 2009, respectively. In Al’Aqr, the area of field crops slightly increased from 0.3 ha (17.0 %) in 2007 to 0.7 (39.1 %) in 2008, and decreased to 0.5 ha (28.8 %) in 2009. During the same period annual dry matter yields of the cash crop garlic in Ash Sharayjah increased from 16.3 t ha-1 in 2007 to 19.8 t ha-1 in 2008 and 18.3 t ha-1 in 2009, while the same crop yielded only 0.4, 1.6 and 1.1 t ha-1 in Masayrat. In 2009, the total estimated agricultural area of the new town of Sayh Qatanah above the five oases was around 13.5 ha. Our results suggest that scarcity of irrigation water as a result of low precipitation and increased irrigation and home water consumption in the new urban settlements above the five oases have led to major shifts in the land-use pattern and increasingly threaten the centuries-long tradition and drought-resilience of agriculture in the oases of the studied watershed.

Mechanisms for the land/sea warming contrast exhibited by simulations of climate change

The land/sea warming contrast is a phenomenon of both equilibrium and transient simulations of climate change: large areas of the land surface at most latitudes undergo temperature changes whose amplitude is more than those of the surrounding oceans. Using idealised GCM experiments with perturbed SSTs, we show that the land/sea contrast in equilibrium simulations is associated with local feedbacks and the hydrological cycle over land, rather than with externally imposed radiative forcing. This mechanism also explains a large component of the land/sea contrast in transient simulations as well. We propose a conceptual model with three elements: (1) there is a spatially variable level in the lower troposphere at which temperature change is the same over land and sea; (2) the dependence of lapse rate on moisture and temperature causes different changes in lapse rate upon warming over land and sea, and hence a surface land/sea temperature contrast; (3) moisture convergence over land predominantly takes place at levels significantly colder than the surface; wherever moisture supply over land is limited, the increase of evaporation over land upon warming is limited, reducing the relative humidity in the boundary layer over land, and hence also enhancing the land/sea contrast. The non-linearity of the Clausius–Clapeyron relationship of saturation specific humidity to temperature is critical in (2) and (3). We examine the sensitivity of the land/sea contrast to model representations of different physical processes using a large ensemble of climate model integrations with perturbed parameters, and find that it is most sensitive to representation of large-scale cloud and stomatal closure. We discuss our results in the context of high-resolution and Earth-system modelling of climate change.

Understanding land-sea warming contrast in response to increasing greenhouse gases. Part I: Transient adjustment

Climate model simulations consistently show that surface temperature over land increases more rapidly than over sea in response to greenhouse gas forcing. The enhanced warming over land is not simply a transient effect caused by the land–sea contrast in heat capacities, since it is also present in equilibrium conditions. This paper elucidates the transient adjustment processes over time scales of days to weeks of the surface and tropospheric climate in response to a doubling of CO2 and to changes in sea surface temperature (SST), imposed separately and together, using ensembles of experiments with an atmospheric general circulation model. These adjustment processes can be grouped into three stages: immediate response of the troposphere and surface processes (day 1), fast adjustment of surface processes (days 2–5), and adjustment of the whole troposphere (days 6–20). Some land surface warming in response to doubled CO2 (with unchanged SSTs) occurs immediately because of increased downward longwave radiation. Increased CO2 also leads to reduced plant stomatal resistance and hence restricted evaporation, which increases land surface warming in the first day. Rapid reductions in cloud amount lead in the next few days to increased downward shortwave radiation and further warming, which spreads upward from the surface, and by day 5 the surface and tropospheric response is statistically consistent with the equilibrium value. Land surface warming in response to imposed SST change (with unchanged CO2) is slower. Tropospheric warming is advected inland from the sea, and over land it occurs at all levels together rather than spreading upward from the surface. The atmospheric response to prescribed SST change in about 20 days is statistically consistent with the equilibrium value, and the warming is largest in the upper troposphere over both land and sea. The land surface warming involves reduction of cloud cover and increased downward shortwave radiation, as in the experiment with CO2 change, but in this case it is due to the restriction of moisture supply to the land (indicated by reduced soil moisture), whereas in the CO2 forcing experiment it is due to restricted evaporation despite increased moisture supply (indicated by increased soil moisture). The warming over land in response to SST change is greater than over the sea and is the dominant contribution to the land–sea warming contrast under enhanced CO2 forcing.

Modelling sediment supply and transport in the River Lugg: strategies for controlling sediment loads

The River Lugg has particular problems with high sediment loads that have resulted in detrimental impacts on ecology and fisheries. A new dynamic, process-based model of hydrology and sediments (INCA- SED) has been developed and applied to the River Lugg system using an extensive data set from 1995–2008. The model simulates sediment sources and sinks throughout the catchment and gives a good representation of the sediment response at 22 reaches along the River Lugg. A key question considered in using the model is the management of sediment sources so that concentrations and bed loads can be reduced in the river system. Altogether, five sediment management scenarios were selected for testing on the River Lugg, including land use change, contour tillage, hedging and buffer strips. Running the model with parameters altered to simulate these five scenarios produced some interesting results. All scenarios achieved some reduction in sediment levels, with the 40% land use change achieving the best result with a 19% reduction. The other scenarios also achieved significant reductions of between 7% and 9%. Buffer strips produce the best result at close to 9%. The results suggest that if hedge introduction, contour tillage and buffer strips were all applied, sediment reductions would total 24%, considerably improving the current sediment situation. We present a novel cost-effectiveness analysis of our results where we use percentage of land removed from production as our cost function. Given the minimal loss of land associated with contour tillage, hedges and buffer strips, we suggest that these management practices are the most cost-effective combination to reduce sediment loads.

Change, rigidity & delay in the UK system of land-use development control

The British system of development control is time-consuming and uncertain in outcome. Moreover, it is becoming increasingly overloaded as it has gradually switched away from being centred on a traditional ‘is it an appropriate land-use?’ type approach to one based on multi-faceted inspections of projects and negotiations over the distribution of the potential financial gains arising from them. Recent policy developments have centred on improving the operation of development control. This paper argues that more fundamental issues may be a stake as well. Important market changes have increased workloads. Furthermore, the UK planning system's institutional framework encourages change to move in specific directions, which is not always helpful. If expectations of increased long-term housing supply are to be met more substantial changes to development control may be essential but hard to achieve.

Planning delay and the responsiveness of English housing supply

There is growing international interest in the impact of regulatory controls on the supply of housing. Most research focuses on the supply impacts of prescribed limits on land use but housing supply may also be affected by the process of planning monitoring and approval but this is hard to measure in detail. The UK has a particularly restrictive planning regime and a detailed and uncertain process of development control linked to it, but does offer the opportunity of detailed site-based investigation of planning delay. This paper presents the findings of empirical research on the time taken to gain planning permission for selected recent major housing projects in southern England. The scale of delay found was far greater than is indicated by average official data measuring the extent to which local authorities meet planning delay targets. Hedonic modelling indicated that there is considerable variation in the time it takes local authorities to process planning applications. Housing association developments are processed more quickly than those of large developers and small sites appear to be particularly time-intensive. These results suggest that delays in development control may be a significant contributory factor to the low responsiveness of UK housing supply to upturns in market activity.

Competition for land

A key challenge for humanity is how a future global population of 9 billion can all be fed healthily and sustainably. Here, we review how competition for land is influenced by other drivers and pressures, examine land-use change over the past 20 years and consider future changes over the next 40 years. Competition for land, in itself, is not a driver affecting food and farming in the future, but is an emergent property of other drivers and pressures. Modelling studies suggest that future policy decisions in the agriculture, forestry, energy and conservation sectors could have profound effects, with different demands for land to supply multiple ecosystem services usually intensifying competition for land in the future. In addition to policies addressing agriculture and food production, further policies addressing the primary drivers of competition for land (population growth, dietary preference, protected areas, forest policy) could have significant impacts in reducing competition for land. Technologies for increasing per-area productivity of agricultural land will also be necessary. Key uncertainties in our projections of competition for land in the future relate predominantly to uncertainties in the drivers and pressures within the scenarios, in the models and data used in the projections and in the policy interventions assumed to affect the drivers and pressures in the future.

Cost effectiveness of mitigation measures for water security and water quality in the River Thames catchment: impacts of changing climate, land use and water scarcity

The catchment of the River Thames, the principal river system in southern England, provides the main water supply for London but is highly vulnerable to changes in climate, land use and population. The river is eutrophic with significant algal blooms with phosphorus assumed to be the primary chemical indicator of ecosystem health. In the Thames Basin, phosphorus is available from point sources such as wastewater treatment plants and from diffuse sources such as agriculture. In order to predict vulnerability to future change, the integrated catchments model for phosphorus (INCA-P) has been applied to the river basin and used to assess the cost-effectiveness of a range of mitigation and adaptation strategies. It is shown that scenarios of future climate and land-use change will exacerbate the water quality problems, but a range of mitigation measures can improve the situation. A cost-effectiveness study has been undertaken to compare the economic benefits of each mitigation measure and to assess the phosphorus reductions achieved. The most effective strategy is to reduce fertilizer use by 20% together with the treatment of effluent to a high standard. Such measures will reduce the instream phosphorus concentrations to close to the EU Water Framework Directive target for the Thames.

Application of landsat images for quantifying the energy balance under conditions of land use changes in the semi-arid region of Brazil

In the Nilo Coelho irrigation scheme, Brazil, the natural vegetation has been replaced by irrigated agriculture, bringing importance for the quantification of the effects on the energy exchanges between the mixed vegetated surfaces and the lower atmosphere. Landsat satellite images and agro-meteorological stations from 1992 to 2011 were used together, for modelling these exchanges. Surface albedo (α0), NDVI and surface temperature (T0) were the basic remote sensing retrieving parameters necessary to calculate the latent heat flux (λE) and the surface resistance to evapotranspiration (rs) on a large scale. The daily net radiation (Rn) was obtained from α0, air temperature (Ta) and short-wave transmissivity (τsw) throughout the slob equation, allowing the quantification of the daily sensible heat flux (H) by residual in the energy balance equation. With a threshold value for rs, it was possible to separate the energy fluxes from crops and natural vegetation. The averaged fractions of Rn partitioned as H and λE, were in average 39 and 67%, respectively. It was observed an increase of the energy used for the evapotranspiration process inside irrigated areas from 51% in 1992 to 80% in 2011, with the ratio λE/Rn presenting an increase of 3 % per year. The tools and models applied in the current research, can subsidize the monitoring of the coupled climate and land use changes effects in irrigation perimeters, being valuable when aiming the sustainability of the irrigated agriculture in the future, avoiding conflicts among different water users. © 2012 SPIE.

Jatropha mahafalensis for rural energy supply in south-western Madagascar?

In many parts of the eastern African region wood-based fuels will remain dominant sources of energy in coming decades. Pressure on forests, especially in semi-arid areas will therefore continue increasing. In this context, the role of liquid biofuels as substitutes for firewood and charcoal, to help reducing pressure on woody biomass and contributing to a better energy security of rural communities, has remained controversial among researchers and practitioners. At household level, the economic and technical feasibility of straight vegetable oil (SVO) was assessed mainly on Jatropha curcas, with unpersuasive results. So far nothing is known about the suitability as an energy carrier of Jatropha mahafalensis Jum. & H. Perrier, the only endemic representative of the Jatropha genus in Madagascar. This paper explores the potential of this plant as a biofuel feedstock in the agro-pastoral area of Soalara, in the semi-arid south-western part of Madagascar. Only hedge-based production was considered to rule out competition over land with food crops. Yield data, the length of currently existing hedges and energy consumption patterns of households were used to assess the quantitative potential and economic viability of J. mahafalensis SVO for lighting and cooking. Tests were conducted with cooking and lighting devices to assess their technical suitability at household level. The paper concludes that J. mahafalensis hedges have some potential to replace paraffin for lighting (though without much economic benefit for the concerned households), but not to replace charcoal or firewood for cooking. The paper recommends that rural energy strategies in similar contexts do not focus only on substituting current fuels with SVO, but should also take into consideration other alternatives. In the case of cooking, there seems to be substantially more potential in increasing the efficiency of current fuel production and consumption technologies (kilns and stoves); and in the case of lighting, solutions based on SVO need to be compared against other options such as portable solar devices.