Conselho Nacional de Desenvolvimento CientÃfico e TecnolÃgico / O ArquipÃlago de Cabo Verde, situado na costa ocidental africana, sofre influÃncia do deserto de Saara tornando o clima caracterizado por pluviometria muito baixa e distribuÃda irregularmente no espaÃo e no tempo. As chuvas sÃo muito concentradas, gerando grandes escoamentos para o mar. O aumento da disponibilidade hÃdrica requer alÃm da construÃÃo e manutenÃÃo de infraestrutura de captaÃÃo e conservaÃÃo de Ãguas pluviais, uma gestÃo eficiente destes recursos. Atualmente, constitui um dos eixos estratÃgicos da polÃtica do Estado de Cabo Verde, a captaÃÃo, o armazenamento e a mobilizaÃÃo de Ãguas superficiais atravÃs de construÃÃo de barragens. Estudos do comportamento hidrolÃgico e sedimentolÃgico do reservatÃrio e da sua bacia de contribuiÃÃo constituem premissas bÃsicas para adequados dimensionamento, gestÃo e monitoramento da referida infraestrutura. à nesse sentido que o presente estudo objetivou sistematizar informaÃÃes hidrolÃgicas e sedimentolÃgicas da bacia hidrogrÃfica da Barragem de PoilÃo (BP) e apresentou proposta operacional de longo prazo. A Ãrea de estudo ocupa 28 km da Bacia HidrogrÃfica da Ribeira Seca (BHRS) na Ilha de Santiago. A altitude da bacia varia de 99 m, situada na cota da barragem, atà 1394 m. Para o estudo, foram utilizados e sistematizados sÃrie pluviomÃtrica de 1973 a 2010, registros de vazÃo instantÃnea do perÃodo 1984 a 2000 e registros agroclimÃticos da Ãrea de estudo de 1981 a 2004. Para o preenchimento das falhas tanto dos escoamentos como da descarga sÃlida em suspensÃo, foi utilizado o mÃtodo de curva chave. Para estimativa de produÃÃo de sedimentos na bacia, aplicou-se a EquaÃÃo Universal de Perda de Solo (USLE) associada à razÃo de aporte de sedimentos (SDR). O Ãndice de retenÃÃo de sedimentos no reservatÃrio foi estimado pelo mÃtodo de Brune e a distribuiÃÃo de sedimento pelo mÃtodo empÃrico de reduÃÃo de Ãrea descrito por Borland e Miller e revisado por Lara. Para gerar e simular curvas de vazÃo versus garantia foi utilizado cÃdigo computacional VYELAS, desenvolvido por AraÃjo e baseado na abordagem de Campos. TambÃm foi avaliada uma possÃvel reduÃÃo da vazÃo de retirada do perÃodo 2006 a 2026, provocada pelo assoreamento do reservatÃrio. Concluiu-se que, em mÃdia, a precipitaÃÃo anual à de 323 mm, concentrando-se 73% nos meses de agosto e setembro; a bacia de contribuiÃÃo apresenta nÃmero de curva (CN) 76, com abstraÃÃo inicial (Ia) de 26 mm, coeficiente de escoamento de 19% e vazÃo anual afluente de 1,7 hm (Cv = 0,73); a disponibilidade hÃdrica para garantia de 85% à 0,548 hmÂ/ano e nÃo 0,671 hmÂ/ano, como indica o projeto original da barragem. Com uma descarga sÃlida avaliada em 22.185 mÂ/ano, estima-se que, atà o ano de 2026, a capacidade do reservatÃrio seja reduzida a uma taxa de 1,8 % ao ano, devido ao assoreamento, provocando uma reduÃÃo de 41% da disponibilidade hÃdrica inicial. Em 2026, segundo as estimativas dessa pesquisa, as perdas por evaporaÃÃo e sangria deverÃo ser da ordem de 81% da vazÃo afluente ao reservatÃrio. Na base desses resultados se apresentou proposta de operaÃÃo da Barragem de PoilÃo / Cape Verde Islands, located in the West African coast, are influenced by the Sahara desert, making the climate characterized by very low rainfall and great spatial and temporal variability. The rainfall events are highly concentrated generating large flows to the sea. The increase of water availability requires construction and maintenance of infrastructure to capture and conserve rainwater, and also an efficient management of these resources. Nowadays, capture, storage and mobilization of water surface through construction of dams is one of the strategic priorities of the policy of the State of Cape Verde. Studies of hydrology and sedimentology of the only existing reservoir (PoilÃo Dam, 1.2 hmÂ) and its watershed are the basic requirements for adequate design, management and monitoring of this infrastructure. In this sense, the present study aimed at systematizing hydrological and sediment information of the PoilÃo Dam and providing a long-term operational proposal. The altitude of the basin varies from 99 m, situated on the dam, up to 1394m. For the study, the following data were used and systematized: rainfall dataset from 1973 to 2010; records of instantaneous flow for the period of 1984 to 2000; and agro-climatic records from 1981 to 2004. To fill the gaps of both flow and suspended-sediment discharge, the rating-curve method was used. To estimate sediment yield in the watershed it was applied the Universal Soil Loss Equation (USLE), associated with the Sediment Delivery Ratio (SDR). The sediment release efficiency rate of the reservoir was estimated by Brune method and the sediment distribution inside the reservoir by the empirical area reduction method described by Borland and Miller, and reviewed by Lara. To assess water availability, water yield versus reliability curves were used. These curves were obtained by the computer code VYELAS, developed by AraÃjo and based on Campos approach. The reduction of the water yield (caused by the reservoir silting) within the period from 2006 to 2026 was also assessed. It was concluded that the average annual rainfall in the watershed is 323 mm, concentrating 73% in August and September; the watershed presents a number of curve (CN) of 76 with an initial abstraction (Ia) of 26 mm, the runoff coefficient is 19% and the average annual inflow is 1.7 hm (coefficient of variation 0.73); the water availability at 85% reliability level is estimated at 0.548 hmÂ/year and not 0.671 hmÂ/year as the original design indicates. With a sediment discharge estimated at 22,185 mÂ/year, it was concluded that, by the year 2026, the annual reservoir storage capacity reduction rate is 1.18% (due to siltation), causing a 41% reduction of the initial water availability. At that time, losses due to evaporation and spillway overflow totalize 81% of annual inflow. Based on these results, a PoilÃo reservoir operation proposal is presented

Este trabalho de dissertação objetiva uma proposta metodológica para a análise da relação entre cenários futuros de disponibilidade hídrica superficial e demanda hídrica, levando-se em consideração a evolução dos fatores intervenientes à oferta hídrica superficial e à demanda hídrica, segundo a narrativa de cenários tendenciais de propagação dos padrões históricos de evolução destes fatores e a narrativa de cenários alternativos. Esta análise é feita através de um balanço hídrico entre disponibilidade hídrica e demanda hídrica. Esta proposta metodológica é aplicada sobre a região de abrangência da UGRHI do Tietê-Jacaré, e esta aplicação objetiva contribuir para a elaboração do plano da bacia da UGRHI em questão. Os fatores intervenientes à oferta hídrica superficial dizem respeito às variáveis meteorológicas, que influem diretamente no regime de precipitações, e aos padrões estatísticos históricos de evolução das precipitações. Os fatores intervenientes à demanda hídrica dizem respeito às variáveis sócio-econômicas e parâmetros tecnológicos de eficiência dos usos múltiplos dos recursos hídricos. Ambos os cenários, tendenciais e alternativos, narram a evolução destes fatores intervenientes à disponibilidade hídrica superficial e demanda hídrica nas dimensões climática e sócio-econômica. Na dimensão climática são propostos dois cenários de evolução dos fatores intervenientes à disponibilidade hídrica: o cenário climático tendencial (CCT) e o cenário climático alternativo. O cenário CCT é uma propagação futura dos padrões estatísticos históricos de evolução dos regimes de precipitações. Para este cenário são geradas séries sintéticas de precipitação, sobre a região ora em estudo, a partir da aplicação de um modelo estocástico linear. O cenário CCA leva em consideração os efeitos regionais das projeções de mudanças climáticas do cenário A1F (IPCC, 2000) para a região ora em estudo. Para este cenário são inferidos os efeitos da projeção de aumento da temperatura, sobre as séries geradas, através de relações empíricas regionais e conceituais de algumas variáveis meteorológicas que influem sobre o comportamento do regime de precipitações. Para a estimativa da disponibilidade hídrica superficial, segundo os dois cenários climáticos supracitados, foi usado o modelo HEC-HMS 2.1 (USACE) de modelagem e simulação de processos de chuva-vazão. Na dimensão sócio-econômica são propostos dois cenários de evolução dos fatores intervenientes à demanda hídrica: o cenário sócio-econômico tendencial (CSET) e o cenário sócio-econômico alternativo (CSEA). O cenário CSET é uma propagação futura dos padrões estatísticos históricos de evolução dos fatores sócio-econômicos e parâmetros tecnológicos intervenientes à demanda hídrica. O cenário CSEA leva em consideração os efeitos regionais das projeções de mudanças sócio-econômicas do cenário A1F (IPCC, 2000) para a região ora em estudo. Como ferramenta computacional para a estimativa de evolução dos fatores intervenientes à demanda hídrica foi desenvolvido, pelo autor dessa dissertação, o modelo MDSA ( Modelo de Demanda Setorial por Recursos Hídricos) baseado na metodologia do modelo NoWUM (Nordeste Water Use Model) (GAISER, T. et al., 2003). O resultado principal deste trabalho de dissertação foi obtido quando da comparação entre os cenários de disponibilidade hídrica superficial e demanda hídrica. Para todos os cenários comparativos foi constatada a evolução para uma situação crítica da relação entre disponibilidade hídrica superficial e demanda hídrica, principalmente nos períodos de estiagem de chuvas, ou períodos de recessão dos escoamentos superficiais. A ocorrência futura da situação crítica, ou escassez hídrica, varia de acordo com a maior ou menor pressão sobre os recursos hídricos, de acordo com os cenários sócio-econômicos, e de acordo com uma disponibilidade maior ou menor de chuvas sobre a região ora em estudo, de acordo com os cenários climáticos. Embora se possa esperar e seja evidente, de forma intuitiva, a ocorrência da escassez hídrica, a proposta, deste trabalho de dissertação, é a elaboração de uma metodologia, ou conjunto de métodos, para a estimativa quantitativa das variáveis de oferta e disponibilidade e demanda hídrica em todas as dimensões espacial e temporal. / This work aims a methodological proposal for the analysis of future scenarios of surface water availability and water demand, being taken into consideration the evolution of the intervening factors of surface water availability and water demand, according to the narrative of trend scenarios of propagation of the historical standards, and the narrative of alternative scenarios. This methodological proposal is applied on the region of the UGRHI of Tietê-Jacaré, and this objective application is to contribute for the elaboration of the watershed plan of the UGRHI. The intervening factors to superficial water availability concern to the meteorological variables that influence directly in the precipitation regimen and to the historical statistical standards of evolution of precipitations. The intervening factors to the water demand concern to the socioeconomic variables and technological parameters of efficiency of the multiple uses of the water resources. Both scenarios, trend and alternative, tell the evolution of these intervening factors into the superficial water availability and water demand in the climatic and socioeconomic dimensions. In the climatic dimension two scenarios of evolution of the intervening factors to the water availability are considered: the climatic tendency scenario (CCT) and the climatic alternative scenario (CCA). The scenario CCT is a future propagation of the historical statistical standards of evolution of regimes of precipitations. For this scenario synthetic precipitation series are generated, on the region in study, from the application of a linear stochastic model. The scenario CCA takes in consideration the regional effect of the projections of climatic changes of scenario A1F (IPCC, 2000) for the region in study. For this scenario the effect of the projection of increase of the temperature are inferred, on the generated series, through regional and conceptual empirical relations of some meteorological variables that influence on the behavior of the precipitation behavior. For the estimate of the superficial water availability, according to two above-mentioned climatic scenarios, the model HEC-HMS 2,1 (USACE) of modeling and simulation of hydrological processes was used. In the socioeconomic dimension two scenarios of evolution of the intervening factor to the water demand are considered: the tendency socioeconomic scenario (CSET) and alternative socioeconomic scenario (CSEA). Scenario CSET is a future propagation of the historical statistical standards of evolution of the socioeconomic factors and intervening technological parameters to the water demand. Scenario CSEA takes in consideration the regional affect of the projections of socioeconomic changes of scenario A1F (IPCC, 2000) for the region in study. As a computational tool for the estimate of the evolution of the intervening factors to the water demand it was developed the MDSA model (Model of Sectorial Demand for Water Resources) based in the methodology of the NoWUM model (Northeast Water Use Model) (GAISER, T. et al.). The main result of this work was gotten when of the comparison between the scenarios of superficial water availability and water demand. For all the comparative scenarios were evidenced the evolution for a critical situation of the relation between superficial water availability and water demand, mainly in the periods of no rains, or periods of contraction of the superficial drainings. The future occurrence of the critical situation, or water scarcity, varies in accordance to the greater or minor pressure on the water resources, in accordance with the socioeconomic scenarios, and in accordance with a bigger or lesser availability of rains on the region in study, in accordance with the climatic scenarios. Although if it can wait and either evident the occurrence of the water scarcity, the proposal, of this work, is the elaboration of a methodology, or set of methods, for the quantitative estimate of the variable of availability and water demand in all the space and temporal dimensions.

Semi-arid valleys in northern Utah are home to the majority of the state population and are dependent upon winter snowpack in surrounding mountains for water for irrigation, hydropower and municipal use. Water is delivered to the urban areas in the spring as discharge in rivers draining the mountains. Understanding the natural variability and cycles of wet and dry periods enables water managers to make informed water allocations. However, the complex regional climate teleconnections are not well understood and the shortness of the instrumental period does not allow for a full understanding of natural variability. Paleo proxies can be used to extend the instrumental record and better capture natural variability. This study uses dendrochronology to reconstruct streamflows of the Logan River in northern Utah over the last several centuries to provide water managers with a better understanding of natural variability. This reconstruction involved sampling and creating three Douglas-fir, one limber pine and two Rocky Mountain juniper chronologies in northern Utah. Combined with existing chronologies, three flow reconstructions of the Logan River were created: one using only within basin chronologies, one using all considered chronologies and one long chronology. Employing regional chronologies resulted in the most robust models, similar to other findings. Results indicate that the last several centuries exhibited greater variability and slightly higher mean annual flows than in the instrumental record (1922-2011). These reconstructions were created using species well established within the dendroclimatology literature such as of Douglas-fir and limber pine and the lesser used Rocky Mountain juniper. The success of Rocky Mountain juniper suggests that it can be a useful species for dendroclimatology in other areas lacking more widely recognized species in semi-arid climates (e.g., pinyon pine).

<p>According to several scientific reports, climate change will have an impact on water provision and thus agriculture, which depends on soil moisture for plant survival. India is a country that is heavily dependent on agriculture as a source of income. One of the country’s future challenges is securing water for irrigation. Cotton in India is an important cash crop which is grown under high evapotranspirative demand, using about 15% of the national water resources, making the crop vulnerable to changes in water availability.</p><p>The purpose of this study is to evaluate the resilience of cotton production with regards to water availability and pesticide use in Punjab and Andhra Pradesh. Three aspects of resilience: latitude, resistance and precariousness has been used to analyse three variables, precipitation, irrigation and pesticide in order to understand how these cotton growing systems are going to be affected by climate change. By bringing together existing data from several scientific reports and governmental websites, assumptions could be made whether these systems are resilient or if they are reaching a threshold. The results show that the cotton growing regions of Punjab are highly vulnerable when it comes to water provision in the region and that they might be reaching a threshold. Changes in climate are predicted to affect precipitation and temperature in the area, which in time might ultimately affect water resources in the region. Groundwater depletion and water logging are already prevailing problems in the area where almost all cotton production is irrigated. Cotton farmers in Andhra Pradesh are struggling with pest infestation which induces them to overconsume pesticides, affecting not only water quality in the area, but also farmers’ livelihood. It is likely that climate change will not minimize the outbreaks; on the contrary it might benefit some pests, which might increase the consumption of pesticide in the region. Coastal districts are more exposed to extreme weather which can harm cotton cultivation.</p>

The purpose of my research was to test the hypothesis that the availability of fresh, naturally occurring water may limit the distribution of Florida Key Deer (Odocoileus virginianus clavium). More specifically, I was trying to determine if there was enough fresh, drinkable water for the deer on each of the islands. To test the hypothesis, I developed a model that simulated likely seasonal fluctuations in fresh water availability in naturally occurring waterholes within the Key Deer range. I estimated 60 scenarios representing different weather (precipitation and evaporation) conditions, different literature estimates of the daily water requirement of Key Deer and also different upper salinity thresholds for drinkable water. Results showed that 1) even under the most favorable conditions in terms of fresh water availability, there was not enough fresh, drinkable water for the deer on any of the islands. Results also showed that 2) high salinity was important in determining the fresh water availability to the deer, in addition to the lack of water volume. Although these results suggest a prolonged seasonal shortage of fresh, naturally occurring water on each of the islands, deer were present on all of the islands during all seasons. One possible reason for the lack of correlation between Key Deer distribution and naturally occurring waterholes is the availability of man-made water sources (e.g. birdbaths, swimming pools).

The purpose of my research was to test the hypothesis that the availability of fresh, naturally occurring water may limit the distribution of Florida Key Deer (Odocoileus virginianus clavium). More specifically, I was trying to determine if there was enough fresh, drinkable water for the deer on each of the islands. To test the hypothesis, I developed a model that simulated likely seasonal fluctuations in fresh water availability in naturally occurring waterholes within the Key Deer range. I estimated 60 scenarios representing different weather (precipitation and evaporation) conditions, different literature estimates of the daily water requirement of Key Deer and also different upper salinity thresholds for drinkable water. Results showed that 1) even under the most favorable conditions in terms of fresh water availability, there was not enough fresh, drinkable water for the deer on any of the islands. Results also showed that 2) high salinity was important in determining the fresh water availability to the deer, in addition to the lack of water volume. Although these results suggest a prolonged seasonal shortage of fresh, naturally occurring water on each of the islands, deer were present on all of the islands during all seasons. One possible reason for the lack of correlation between Key Deer distribution and naturally occurring waterholes is the availability of man-made water sources (e.g. birdbaths, swimming pools).

Surface water availability models often use monthly simulation time steps for reasons of data availability, model parameter parsimony, and reduced computational time. Representing realistic streamflow variability, however, requires modeling time steps with sub-monthly or daily temporal resolution. Adding daily time step simulation capability to the Water Rights Analysis Package (WRAP) and the Texas Water Availability Modeling (WAM) System is a growing area of need and interest in water rights permitting, water supply planning, and environmental protection. This research consisted of the following tasks: 1. Key modeling issues are identified that are relevant to daily time step modeling, but are otherwise not considered with monthly simulations. These key modeling issues include disaggregating monthly naturalized flows into daily flows, routing changes to flow through the stream network, reducing impacts to water availability in a priority order based water right system through the use of streamflow forecasting, distributing water right targets from monthly to daily amounts, and integrating flood control reservoir operations into the existing conservation reservoir modeling framework. 2. Two new programs for WRAP are developed to address the key daily time step modeling issues. The new programs include a pre-processor program, DAY, and a daily simulation program, SIMD. 3. A case study of the Brazos River Basin WAM is presented using daily time steps with SIMD. The purpose of the case study is to present an implementation of the daily modeling capabilities. 4. The case study simulation results are used as a basis to draw conclusions regarding monthly versus daily simulation outcomes. The research, as presented through the Brazos River Basin WAM case study, illustrated that incorporating realistic daily streamflow variability into the simulation of a priority order based water allocation system can substantially affect the results obtained for time series of critical period reservoir storage contents, the determination of long-term water right reliability, and the distribution of unappropriated and regulated flows. The modeling capabilities developed by this research advance the state of water availability modeling with sub-monthly time steps by addressing the key modeling issues related to streamflow variability and routing.

Streamflow characteristics change over time as a result of water resources development and management projects, water use, watershed land use changes, and climate changes. The main objective of this thesis is to assess the significance of the impacts of human activities such as construction of reservoirs, water supply diversions, increased water use and return flows on streamflows by the recently completed Texas WAM (Water Availability Modeling) system. The major river basins in the state of Texas were selected as suitable study basins. The particular objective is accomplished by the assessment of WAM monthly and annual naturalized and regulated flows, based on using the WRAP (Water Rights Analysis Package) model, which represents the river/reservoir management model. WAM flow frequency analysis was performed for the simulated flows. The flow ratio indices developed showed the divergence of the actual flows from their natural behavior for the entire monthly flow frequency flow spectrum ranging from minimum flows to high flows. This study describes the combined effects of reservoir construction, increased water use, water resources development projects and land use changes on the river flow regime.

Drought and poor soil fertility are often cited as major constraints to rainfed lowland rice production in Laos, particularly in the central and southern parts of the country, where uncertainty in the growing season is common, due mainly to a combination of unreliable rainfall and coarse textured soils with a low water holding capacity and high rates of deep percolation. The soil conditions, together with fluctuations in rainfall distribution, are regarded as the most serious constraints to achieving high and stable grain yields in the rainfed lowland rice ecosystem in this region. Improvements in rainfed lowland rice productivity depend, in part, on improved predictions of water availability, as well as better understanding of soil nutrient and water stress limitations to grain yield. The use of a soil water balance model, in conjunction with the quantification of soil nutrient availability, can help in estimating free water levels in the rice fields, thereby helping determine the duration of the growing period, as well as, helping with predictions of potential yield reduction due to water stress and soil nutrient limitations. Linking the simulated results with Geographic Information System (GIS) can help quantify the spatial pattern of these attributes at a provincial or regional scale. This study was aimed at quantifying the spatial distribution of water availability, including the frequency and severity of water stress development during the growing period, and to estimate the effects of soil fertility and water stress on rice productivity in Savannakhet province in southern Lao PDR. The current status in relation to the variability in field water availability, soil fertility, farm crop management practices and grain yield throughout Savannakhet province were quantified, first by collecting farm data from 53 and 48 farmers’ fields in the 2007 and 2008 cropping seasons, respectively, These farms were selected to be representative of a wide range of soil and climate conditions within the province. The results show that rainfall distribution pattern, soil type and toposequence position of paddy fields, are crucial factors contributing to the temporal variation in field water availability. The overall yield loss due to water stress associated with late season drought was estimated to be rather small (10%, 5% and 3% for the top, middle and bottom toposequence positions of rice fields, respectively) in the 2008 wet season. On the other hand, application of chemical fertilizer has a large effect on final grain yield, with 50 kg N ha-1 and 30 kg P2O5 ha-1 increasing yield by 600 to 800 kg and 800 to 1000kg ha-1 respectively, in the 2007 and 2008 cropping seasons. A new soil water balance (SWB) model that incorporates the effect of low soil clay content on deep percolation, was developed to quantify field water availability and the length of growing period (LGP) for various rainfall lowland rice cropping environments. The model estimates the amount of water stored in a soil profile, the profile being divided into two layers: Layer 1 (0-200 mm) consists of standing water and the topsoil layer, while Layer 2 (200-1000 mm) is the subsoil layer. The SWB model was validated with field experimental data obtained in the 2002 and 2008 cropping seasons. The simulated free water levels were close to those recorded for the observed field data, with a small mean average error, lower root mean square error, and significant correlation coefficient and index of agreement over all sites across the three toposequence positions of paddy fields. Maps of the length, start and end of growing period (LGP, SGP and EGP, respectively) for rainfed lowland rice in Savannakhet province, were developed using the SWB model, with inputs of median weekly climatic data and soil water characteristics. The province was delineated into three main LGP zones with a short LGP zone (less than 21 weeks) in the east, northwest and some rice fields in the south-western corner of the province; an intermediate LGP zone (21 to 24 weeks) was defined in the central and western part of the province; and a long LGP zone (greater than 24 weeks) for the south and for some rice fields in the western part of the province. The variation in the SGP from year-to-year was due largely to the variation in rainfall early in the wet-season (e.g. April), while EGP was strongly dependent upon the clay content of the soils being cropped. The SWB model was combined with other models that estimate yield potential, soil nutrient supply and yield reduction by low soil water level, to characterize and map the suitability zones for rainfed lowland rice in Savannakhet province. The overall results of the model performance on yield estimates were satisfactorily, with a significant correlation coefficient (0.54**) and high index of agreement (0.68) over the 2007 and 2008 seasons. The model classified three main rice agro-ecological zones according to the suitability of climate and soil conditions. The majority of the lowland rice growing areas are classified as moderately suitable to marginally suitable, while the potential area classified as being high suitable is very small. A large potential response of rice yields to fertilizer inputs is predicted for most of rice growing areas in the province. The best sowing time for achieving a high yield, as evaluated by the model, is the first half of June. Appropriate crop phenology and increasing fertilizer use efficiency that matches with water availability and soil conditions in each rice agro-ecological zone, are important in achieving improvements in rice productivity, as substantial improvements in rice fields cannot be achieved by improving water availability alone, where paddy fields are dominated by soils with low level of indigenous fertility. Although the model is capable of quantifying field water availability and crop yield due to the limitations associated with low levels of soil nutrients and water stress, the model has the potential for further improvements in two areas. First, the estimates of water loss need to be modified by incorporating variable factors such as slope of paddy field, which can affects lateral water movement and hence free water level. Second, the model should incorporate some key agronomic variables, such as internal efficiencies and recovery efficiencies of applied fertilizer, which depend on variety, crop management and climatic conditions, and these factors can be modelled.

Drought and poor soil fertility are often cited as major constraints to rainfed lowland rice production in Laos, particularly in the central and southern parts of the country, where uncertainty in the growing season is common, due mainly to a combination of unreliable rainfall and coarse textured soils with a low water holding capacity and high rates of deep percolation. The soil conditions, together with fluctuations in rainfall distribution, are regarded as the most serious constraints to achieving high and stable grain yields in the rainfed lowland rice ecosystem in this region. Improvements in rainfed lowland rice productivity depend, in part, on improved predictions of water availability, as well as better understanding of soil nutrient and water stress limitations to grain yield. The use of a soil water balance model, in conjunction with the quantification of soil nutrient availability, can help in estimating free water levels in the rice fields, thereby helping determine the duration of the growing period, as well as, helping with predictions of potential yield reduction due to water stress and soil nutrient limitations. Linking the simulated results with Geographic Information System (GIS) can help quantify the spatial pattern of these attributes at a provincial or regional scale. This study was aimed at quantifying the spatial distribution of water availability, including the frequency and severity of water stress development during the growing period, and to estimate the effects of soil fertility and water stress on rice productivity in Savannakhet province in southern Lao PDR. The current status in relation to the variability in field water availability, soil fertility, farm crop management practices and grain yield throughout Savannakhet province were quantified, first by collecting farm data from 53 and 48 farmers’ fields in the 2007 and 2008 cropping seasons, respectively, These farms were selected to be representative of a wide range of soil and climate conditions within the province. The results show that rainfall distribution pattern, soil type and toposequence position of paddy fields, are crucial factors contributing to the temporal variation in field water availability. The overall yield loss due to water stress associated with late season drought was estimated to be rather small (10%, 5% and 3% for the top, middle and bottom toposequence positions of rice fields, respectively) in the 2008 wet season. On the other hand, application of chemical fertilizer has a large effect on final grain yield, with 50 kg N ha-1 and 30 kg P2O5 ha-1 increasing yield by 600 to 800 kg and 800 to 1000kg ha-1 respectively, in the 2007 and 2008 cropping seasons. A new soil water balance (SWB) model that incorporates the effect of low soil clay content on deep percolation, was developed to quantify field water availability and the length of growing period (LGP) for various rainfall lowland rice cropping environments. The model estimates the amount of water stored in a soil profile, the profile being divided into two layers: Layer 1 (0-200 mm) consists of standing water and the topsoil layer, while Layer 2 (200-1000 mm) is the subsoil layer. The SWB model was validated with field experimental data obtained in the 2002 and 2008 cropping seasons. The simulated free water levels were close to those recorded for the observed field data, with a small mean average error, lower root mean square error, and significant correlation coefficient and index of agreement over all sites across the three toposequence positions of paddy fields. Maps of the length, start and end of growing period (LGP, SGP and EGP, respectively) for rainfed lowland rice in Savannakhet province, were developed using the SWB model, with inputs of median weekly climatic data and soil water characteristics. The province was delineated into three main LGP zones with a short LGP zone (less than 21 weeks) in the east, northwest and some rice fields in the south-western corner of the province; an intermediate LGP zone (21 to 24 weeks) was defined in the central and western part of the province; and a long LGP zone (greater than 24 weeks) for the south and for some rice fields in the western part of the province. The variation in the SGP from year-to-year was due largely to the variation in rainfall early in the wet-season (e.g. April), while EGP was strongly dependent upon the clay content of the soils being cropped. The SWB model was combined with other models that estimate yield potential, soil nutrient supply and yield reduction by low soil water level, to characterize and map the suitability zones for rainfed lowland rice in Savannakhet province. The overall results of the model performance on yield estimates were satisfactorily, with a significant correlation coefficient (0.54**) and high index of agreement (0.68) over the 2007 and 2008 seasons. The model classified three main rice agro-ecological zones according to the suitability of climate and soil conditions. The majority of the lowland rice growing areas are classified as moderately suitable to marginally suitable, while the potential area classified as being high suitable is very small. A large potential response of rice yields to fertilizer inputs is predicted for most of rice growing areas in the province. The best sowing time for achieving a high yield, as evaluated by the model, is the first half of June. Appropriate crop phenology and increasing fertilizer use efficiency that matches with water availability and soil conditions in each rice agro-ecological zone, are important in achieving improvements in rice productivity, as substantial improvements in rice fields cannot be achieved by improving water availability alone, where paddy fields are dominated by soils with low level of indigenous fertility. Although the model is capable of quantifying field water availability and crop yield due to the limitations associated with low levels of soil nutrients and water stress, the model has the potential for further improvements in two areas. First, the estimates of water loss need to be modified by incorporating variable factors such as slope of paddy field, which can affects lateral water movement and hence free water level. Second, the model should incorporate some key agronomic variables, such as internal efficiencies and recovery efficiencies of applied fertilizer, which depend on variety, crop management and climatic conditions, and these factors can be modelled.