Análise conjunta de regressões : ajustamento e zonagem agrícola

A Análise conjunta de regressões, ACR, é uma técnica utitilizada para estudar a interação gentipo x ambiente baseada em regressões. Nesta técnica ajusta-se uma regressão linear por cultivar. Nestas regressões a variável controlada é o índice ambiental que mede a produtividade dos vários ambientes. Nas culturas anuais, os ambientes compreendem aos pares (local, ano). Os valores dos índices ambientais e dos coeficientes das regressões são ajustados simultaneamente. Até agora a ACR tem sido aplicada a uma única cultura de cada vez. Neste trabalho vamos procurar ultrapassar essa limitação através da modelação dos logaritmos dos índices ambientais tendo-se desenvolvido um modelo da forma: zi, j  v  l j  j , i  1,...,n, J  1,...,n onde zi , j é o logaritmo do índice ambiental para o i-essimo ambiente na j  essima cultura , v um valor médio geral , li j  essimo cultivar. o efeito do i  essimo local e  j o efeito do Ao utilizar esta modelação, os locais corresponderão a estações experimentais de forma a poder-se ter várias culturas no mesmo local. Ora, as estações experimentais são escolhidas por forma a serem representativas das regiões onde estão implantadas. Assim, os índices ambientais correspondentes às várias estações experimentais e, consequentemente, às respetivas regiões, pudesse ser utilizados para agrupar regiões contíguas com índices semelhantes obtendo-se assim, um elemento interessante para a Zonagem agrícola no que diz respeito às culturas que se trabalha. Pode-se ainda procurar uma Zonagem para grupos de cultivares. Por exemplo, adiante trabalharemos com dados da cevada e trigo os quais são cereais. ABSTRACT: Joint Regression Analysis, JRA, is one of the techniques for the study of genotypeXenvironment interaction based on the use of regressions .In JRA a linear regression of the yields of each cultivar on a controlled variable, the environment index ,is adjusted .The index miss erasures the productivity of each environment .In yearly cultures the environments correspond to the pairs (location ,years) .These indexes and the correlation coefficients are adjusted simultaneously. Up to now JRA has been applied to single crops .Now we try to overcome this restriction through modeling of the logarithms of the environmental indexes .We developed a model τ i , j = v + l j +λ j , i = 1,..., b, J = 1,...J where τ i , j is the logarithm of the environmental index for the i-th environment and the j-th crop , v is the general mean , li is the effect of the i-th environment and λ j is the effect of the j-th crop . When applying this model the location will correspond to experimental situations in order to have several crops in the same locations .Now experimental stations are chosen to be representative of the regions in which they are located .Then the l1 ,..., lb can be used to group contiguous regions with similar location effects .We thus get an useful tool for Agricultural Zoning for the crops we used or, even, for the group to which those crops belong . For instance we worked with barley and wheat that are cereals.

Seasonal dynamics and operational monitoring of hedgerow olive tree transpiration in response to applied water

We used 2012 sap flow measurements to assess the seasonal dynamics of daily plant transpiration (ETc) in a high-density olive orchard (Olea europaea L. cv. ‘Arbequina’) with a well-watered (HI) control treatment A to supply 100 % of the crop water needs, and a moderately (MI) watered treatment B that replaced 70% of crop needs. To assure that treatment A was well-watered, we compared field daily ETc values against ETc obtained with the Penman-Monteith (PM) combination equation incorporating the Orgaz et al. (2007) bulk daily canopy conductance (gc) model, validated for our non-limiting conditions. We then tested the hypothesis of indirectly monitoring olive ETc from readily available vegetation index (VI) and ground-based plant water stress indicator. In the process we used the FAO56 dual crop coefficient (Kc) approach. For the HI olive trees we defined Kcb as the basal transpiration coefficient, and we related Kcb to remotely sensed Soil Adjusted Vegetation Index (SAVI) through a Kcb-SAVI functional relationship. For the MI treatment, we defined the actual transpiration ETc as the product of Kcb and the stress reduction coefficient Ks obtained as the ratio of actual to crop ETc, and we correlated Ks with MI midday stem water potential (ψst) values through a Ks-ψ functional relationship. Operational monitoring of ETc was then implemented with the ETc = Kcb(SAVI)Ks(ψ)ETo relationship stemmed from the FAO56 approach and validated taking as inputs collected SAVI and ψst data reporting to year 2011. Low validation error (6%) and high goodness-of-fit of prediction were observed (R2 = 0.94, RSME = 0.2 mm day-1, P = 0.0015), allowing to consider that under field conditions it is possible to predict ETc values for our hedgerow olive orchards if SAVI and water potential (ψst) values are known.

Above ground biomass functions with vegetation indices for multiple use systems of two evergreen oaks

Remote sensing is a promising approach for above ground biomass estimation, as forest parameters can be obtained indirectly. The analysis in space and time is quite straight forward due to the flexibility of the method to determine forest crown parameters with remote sensing. It can be used to evaluate and monitoring for example the development of a forest area in time and the impact of disturbances, such as silvicultural practices or deforestation. The vegetation indices, which condense data in a quantitative numeric manner, have been used to estimate several forest parameters, such as the volume, basal area and above ground biomass. The objective of this study was the development of allometric functions to estimate above ground biomass using vegetation indices as independent variables. The vegetation indices used were the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Simple Ratio (SR) and Soil-Adjusted Vegetation Index (SAVI). QuickBird satellite data, with 0.70 m of spatial resolution, was orthorectified, geometrically and atmospheric corrected, and the digital number were converted to top of atmosphere reflectance (ToA). Forest inventory data and published allometric functions at tree level were used to estimate above ground biomass per plot. Linear functions were fitted for the monospecies and multispecies stands of two evergreen oaks (Quercus suber and Quercus rotundifolia) in multiple use systems, montados. The allometric above ground biomass functions were fitted considering the mean and the median of each vegetation index per grid as independent variable. Species composition as a dummy variable was also considered as an independent variable. The linear functions with better performance are those with mean NDVI or mean SR as independent variable. Noteworthy is that the two better functions for monospecies cork oak stands have median NDVI or median SR as independent variable. When species composition dummy variables are included in the function (with stepwise regression) the best model has median NDVI as independent variable. The vegetation indices with the worse model performance were EVI and SAVI.