Relation of leaf water content with real evapotranspiration and biomass in Caatinga biome , using remote sensing data

Introduction Water stress caused by drought limits plat productivity and crop yields by reducing photosynthesis and leaf growth (Hunt at al. 1987, Boyer, 1982; Bradford and Hsiao, 1982). The caatinga has a strategy for surviving the great water shortage. Detection of water stress is a major application of remote sensing (Knipling, 1970; Wiegand et al., 1981). According to Giongo (2011), several works have been development to estimate biophysical variables of the vegetation through orbital images, such as the of NDVI (Rouse et al., 1973) and the Soil Adjusted Vegetation IndexSAVI (Huete, 1988). More recently, the NDWI Moisture Index has been highlighted in the monitoring of water stress in semi-arid environment. The amount of water in the vegetation is strongly correlated to the nearinfrared, allowing estimation of amount of water in cultures (Hardinsky et al. 1983; Gao, 1996; Oliveira et al., 2010). As the water content of leaves in vegetation canopies increases, the strength of the absorption around 1599 nm increases. Absorption at 819 nm is nearly unaffected by changing water content, so it is used as the reference. Applications include canopy stress analysis, productivity prediction and modeling, fire hazard condition analysis, and studies of ecosystem physiology. Revista Brasileira de Geografia Física v.10, n.05 (2017) 1545-1551. 1546 This study has the objective to analyze the relationship between water content in the leaf with biomass and evapotranspiration in the area of the Caatinga biome located in São José do Sabugi, Paraiba, Brazil. Material and methods Study area The municipality of São José do Sabugi, located in the state of Paraíba Brazil (Figure 1), is included in Borborema mesoregion and Caatinga biome (PROBIO, 2004). It is located in the area of the semi-arid region, presenting a hot and dry climate, with a total annual rainfall of about 600 mm (AESA Paraíba State Water Management Executive Agency) and temperatures ranging from 21oC at 36oC (Galvíncio et al., 2009). In terms of land use and cover, there is a predominance of cattle raising, aesthetical savanna, and aesthetical savanna (PROBIO, 2004), characterized by a tree cover composed of small spiny trees and several cacti, covering a gray stratum. Figure 1. Study área location. Radiometric data The images used for the development of this work were the TM sensor (Thematic Mapper) on board the Landsat-5 satellite obtained from the Image Generation Division of the National Institute for Space Research (INPE). Images of the orbit and point 215/65 with date of passage of the satellite were used on June 19, 2008. Image processing and layout assembly Initially, the image was recorded using a recorded Landsat image obtained from the Remote Sensing Center SISCOM IBAMA and later verification of the record through GPS points collected during field activity carried out in the city. For the processing of Landsat-5 satellite images in relation to SEBAL, models were created in the Model Maker tool of the ERDAS Imagine 9.3 software. The division into classes and the final assembly of the maps was done through ArcGIS 9.3 software. Both are licensed by the Department of Geographical Sciences of the Federal University of Pernambuco. Methodology In this study was used the SEBALSurface Energy Balance algorithm for Land. All, descriptions methodology can see in Oliveira et al (2017); Machado et al. (2014); Oliveira et al. (2010) and Allan et al. (2002). Leaf Water Content Index – LWCI To obtain of the water content of leaf was utilized the equations: LWCI= (-log (1-(TM4-TM5)))/(-log(1-(TM4FTTM5FT))) where TM4FT and TM5FT are the maximum reflectance of TM4 and TM5 The statistics methods used was correlations and regressions with significance level Revista Brasileira de Geografia Física v.10, n.05 (2017) 1545-1551. 1547 0.01. The descriptive statistics of the samples was obtained. Evaluation criteria Variance inflation factors (VIF) measure how much the variance of the estimated regression coefficients are inflated as compared to when the predictor variables are not linearly related. Use to describe how much multicollinearity (correlation between predictors) exists in a regression analysis. Multicollinearity is problematic because it can increase the variance of the regression coefficients, making them unstable and difficult to interpret. Use the following guidelines to interpret the VIF: VIF Status of predictors VIF = 1 Not correlated 1 < VIF < 5 Moderately correlated VIF > 5 to 10 Highly correlated


Introduction
Water stress caused by drought limits plat productivity and crop yields by reducing photosynthesis and leaf growth (Hunt at al. 1987, Boyer, 1982;Bradford and Hsiao, 1982).The caatinga has a strategy for surviving the great water shortage.Detection of water stress is a major application of remote sensing (Knipling, 1970;Wiegand et al., 1981).
According to Giongo (2011), several works have been development to estimate biophysical variables of the vegetation through orbital images, such as the of NDVI (Rouse et al., 1973) and the Soil Adjusted Vegetation Index-SAVI (Huete, 1988).More recently, the NDWI Moisture Index has been highlighted in the monitoring of water stress in semi-arid environment.The amount of water in the vegetation is strongly correlated to the nearinfrared, allowing estimation of amount of water in cultures (Hardinsky et al. 1983;Gao, 1996;Oliveira et al., 2010).
As the water content of leaves in vegetation canopies increases, the strength of the absorption around 1599 nm increases.Absorption at 819 nm is nearly unaffected by changing water content, so it is used as the reference.Applications include canopy stress analysis, productivity prediction and modeling, fire hazard condition analysis, and studies of ecosystem physiology.
This study has the objective to analyze the relationship between water content in the leaf with biomass and evapotranspiration in the area of the Caatinga biome located in São José do Sabugi, Paraiba, Brazil.

Study area
The municipality of São José do Sabugi, located in the state of Paraíba -Brazil (Figure 1), is included in Borborema mesoregion and Caatinga biome (PROBIO, 2004).It is located in the area of the semi-arid region, presenting a hot and dry climate, with a total annual rainfall of about 600 mm (AESA -Paraíba State Water Management Executive Agency) and temperatures ranging from 21ºC at 36ºC (Galvíncio et al., 2009).
In terms of land use and cover, there is a predominance of cattle raising, aesthetical savanna, and aesthetical savanna (PROBIO, 2004), characterized by a tree cover composed of small spiny trees and several cacti, covering a gray stratum.

Image processing and layout assembly
Initially, the image was recorded using a recorded Landsat image obtained from the Remote Sensing Center -SISCOM -IBAMA and later verification of the record through GPS points collected during field activity carried out in the city.For the processing of Landsat-5 satellite images in relation to SEBAL, models were created in the Model Maker tool of the ERDAS Imagine 9.3 software.The division into classes and the final assembly of the maps was done through ArcGIS 9.3 software.Both are licensed by the Department of Geographical Sciences of the Federal University of Pernambuco.

Methodology
In this study was used the SEBAL-Surface Energy Balance algorithm for Land.All, descriptions methodology can see in Oliveira

Leaf Water Content Index -LWCI
To obtain of the water content of leaf was utilized the equations: where TM4FT and TM5FT are the maximum reflectance of TM4 and TM5 The statistics methods used was correlations and regressions with significance level 0.01.The descriptive statistics of the samples was obtained.

Evaluation criteria
Variance inflation factors (VIF) measure how much the variance of the estimated regression coefficients are inflated as compared to when the predictor variables are not linearly related.Use to describe how much multicollinearity (correlation between predictors) exists in a regression analysis.
Multicollinearity is problematic because it can increase the variance of the regression coefficients, making them unstable and difficult to interpret.

Results
The descriptive statistics was obtained to N=66, Table 1.In mean LWCI was 1.88%, is low and NDVI was 0.462 is great for area, in June we have the end of rain period..390.12866 A good relationship between vegetation index and leaf water content, with r = 0.76 for SAVI and 0.64 for NDVI.For evapotranspiration the correlation was r = 0.386, Table 2.According to Figure 2 and 3 the standard of the spatial variations of NDVI, SAVI and LWCI was similar.Thus, is possible to estimate the LWCI using the NDVI.
The equations obtained between LWCI, NDVI and SAVI, where the r was 0.85.Or, the model development to estimate LWCI using NDVI and SAVI showed good results, Table 3 and Table  4..The VIF is highly correlated.
Results similar to those obtained in this study can be seen in Anazawa et al. (2001) in which it evaluated the relationship between LWCI and NDVI in different forest and cultivated fields.Vina et al. ( 2011) also found a good relationship between vegetation indexes and water content in the plant.The Table 5 showed coefficients to new model.The equation is: LWCI=22.798SAVI-9.In general, the Figure 4 showed that the new model underestimate the observed data.
The Figure 5 showed the spatial variation of the plant water content estimated with LWCI new modelling to São José do Sabugi.In general the São José do Sabugi municipality show low water content in june.Ist important to talk that June is a month in this area that has your major water content in plant because is the end of the rain period, but yet is low.

Conclusions
It is possible to estimate the water content in the plant using NDVI.Data from drones images it is estimated NDVI concludes that it is possible to estimate water content in the plant using images of drones.
It is recommended that further research be developed on LWCI for drought monitoring, climate mitigation mechanisms, water balance estimation, evepotranspiration estimation and carbon sequestration.

Figure 4 -
Figure 4 -Relationship between observed and expected of the LWCI.

Table 3 -
Model Summary b