•ANN, SVM and empirical equations were evaluated for the estimation of ETo in Brazil.•K-means was used to group weather stations with similar climatic characteristics.•Data from previous days were ...used as additional inputs to ANN and SVM models.•Clustering and previous data provided performance increments.•ANNs with data from previous days were the best evaluated alternatives.
Reference evapotranspiration (ETo) is a variable of great importance for several purposes, such as hydrological studies and irrigation scheduling. The FAO-56 Penman-Monteith (FAO-56 PM) equation is recommended to estimate ETo given its good accuracy. However, the estimation of ETo poses a challenge when the availability of meteorological data is limited since the FAO-56 PM equation requires data on temperature, relative humidity, solar radiation and wind speed. This study evaluates, for the first time, the performance of alternative equations, artificial neural network (ANN) and support vector machine (SVM), for the estimation of daily ETo across the entirety of Brazil using measured data on temperature and relative humidity or only temperature. Two strategies, not yet used in Brazil, were used to develop the ANN and SVM models: (i) the definition of groups of weather stations with similar climatic characteristics, using the K-means clustering algorithm, to develop models specific for each group; and (ii) the addition of previous meteorological data as input for the models. Data from 203 weather stations distributed across Brazil were used. The ANN and SVM models showed higher performances than the equations that were studied, even when they were calibrated. The evaluated strategies (clustering and previous days) provided considerable performance gains. For the temperature-based models, the best performance was obtained by the ANN developed with the strategy of clustering and the use of data from two previous days as input; however, due to the similar performance and greater generalization capacity, the ANN developed without clustering and with the use of data from four previous days is recommended. For the temperature- and relative humidity-based models, the ANN developed with data from four previous days was the best option.
Reference evapotranspiration (ETo) is a fundamental parameter for hydrological studies and irrigation management. The Penman-Monteith method is the standard to estimate ETo and requires several ...meteorological elements. In developing countries, the number of weather stations is insufficient. Thus, free products of remote sensing with evapotranspiration information must be used for this purpose. In this context, the objective of this study was to estimate monthly ETo from potential evapotranspiration (PET) made available by MOD16 product. In this study, the monthly ETo estimated by Penman-Monteith method was considered as the standard. For this, data from 265 weather station of the National Institute of Meteorology (INMET), spread all over the Brazilian territory, were acquired for the period from 2000 to 2014 (15 years). For these months, monthly PET values from MOD16 product for all Brazil were also downloaded. By using machine learning algorithms and information from WorldClim as covariates, ETo was estimated through images from the MOD16 product. To perform the modeling of ETo, eight regression algorithms were tested: multiple linear regression; random forest; cubist; partial least squares; principal components regression; adaptive forward-backward greedy; generalized boosted regression and generalized linear model by likelihood-based boosting. Data from 2000 to 2012 (13 years) were used for training and data of 2013 and 2014 (2 years) were used to test the models. The PET made available by the MOD16 product showed higher values than those of ETo for different periods and climatic regions of Brazil. However, the MOD16 product showed good correlation with ETo, indicating that it can be used in ETo estimation. All models of machine learning were effective in improving the performance of the metrics evaluated. Cubist was the model that presented the best metrics for r2 (0.91), NSE (0.90) and nRMSE (8.54%) and should be preferred for ETo prediction. MOD16 product is recommended to be used to predict monthly ETo, which opens possibilities for its use in several other studies.
Brazil has extensive forests and savannas on deep weathered soils and plays a key role in the discussions about carbon sequestration, but the distribution of soil organic carbon (SOC) stocks up to ...1 m depth has not been investigated in Brazil using machine learning techniques. In this study, we applied a methodological framework to optimize the prediction of SOC stocks for the entire Brazilian territory and determine how the environmental heterogeneity of Brazil influences the SOC stocks distribution. We used a legacy dataset of 8227 soil profiles which consisted of 37,693 samples. For each profile, the vertical distribution of SOC and bulk density were interpolated to standard depths (0–5, 5–15, 15–30, 30–60 and 60–100 cm) using mass preserving equal-area quadratic splines. The covariates database was composed of 74 variables including bioclimatic (temperature and precipitation) data, soil and biome maps, vegetation indexes and morphometric maps derived from a digital elevation model, with a 1 km spatial resolution. To obtain the best prediction performance, we tested four machine learning algorithms: Random Forests, Cubist, Generalized Linear Model Boosting and Support Vector Machines. Random Forests showed the best performance in predicting SOC stocks for all depths, with the highest performance at 30–60 cm for training (R2 = 0.32) and validation (R2 = 0.33); hence, it was selected for the spatial prediction of SOC stocks. The most important covariates selected by Random Forests using the recursive feature elimination were: soil class, sum of monthly mean temperature (SAMT), precipitation, slope height and vegetation indexes (NDVI, GPP). In total, Brazilian soils store approximately 71.3 PgC within the top 100 cm, where the first 0–30 cm contains almost 36 PgC. Approximately 31% of the total SOC stocks (22.2 PgC) occurs in protected areas (2.6 million km2), which are not subjected to land use pressure and carbon losses. Although the Amazon biome has the highest amount of stored SOC (36.1 PgC), its soils do not represent a good potential for carbon accumulation. Among soil classes, the Luvisols showed the lowest SOC density (6.45 kg m−2) and the Histosols presented the highest values (14.87 kg m−2). More than 57% of the total SOC was found in nutrient-poor, deep-weathered Ferralsols and Acrisols, which are the dominant soils in Brazil. The presented methodological framework covers all steps of prediction process, building maps with known accuracy and has great potential to be used in future soil carbon inventories at large scales. Concerning conservation issues, the results highlight the importance of nature reserves for protecting SOC in the long-term.
•We modelled and mapped the SOC stocks and uncertainty at 1 m soil depth for Brazilian territory.•Random Forest showed highest performance for SOC stocks prediction.•Soil class, climate variables and vegetation indexes most influence SOC stocks.•Amazon biome holds the highest total SOC stocks.•Protected areas hold about 31% of the total SOC stocks in Brazil.
Crop NDVI Monitoring Based on Sentinel 1 Filgueiras, Roberto; Mantovani, Everardo Chartuni; Althoff, Daniel ...
Remote sensing (Basel, Switzerland),
06/2019, Letnik:
11, Številka:
12
Journal Article
Recenzirano
Odprti dostop
Monitoring agricultural crops is necessary for decision-making in the field. However, it is known that in some regions and periods, cloud cover makes this activity difficult to carry out in a ...systematic way throughout the phenological cycle of crops. This circumstance opens up opportunities for techniques involving radar sensors, resulting in images that are free of cloud effects. In this context, the objective of this work was to obtain a normalized different vegetation index (NDVI) cloudless product (NDVInc) by modeling Sentinel 2 NDVI using different regression techniques and the Sentinel 1 radar backscatter as input. To do this, we used four pairs of Sentinel 2 and Sentinel 1 images on coincident days, aiming to achieve the greatest range of NDVI values for agricultural crops (soybean and maize). These coincident pairs were the only ones in which the percentage of clouds was not equal to 100% for 33 central pivot areas in western Bahia, Brazil. The dataset used for NDVInc modeling was divided into two subsets: training and validation. The training and validation datasets were from the period from 24 June 2017 to 19 July 2018 (four pairs of images). The best performing model was used in a temporal analysis from 02 October 2017 to 08 August 2018, totaling 55 Sentinel 2 images and 25 Sentinel 1 images. The selection of the best regression algorithm was based on two validation methodologies: K-fold cross-validation (k = 10) and holdout. We tested four modeling approaches with eight regression algorithms. The random forest was the algorithm that presented the best statistical metrics, regardless of the validation methodology and the approach used. Therefore, this model was applied to a time series of Sentinel 1 images in order to demonstrate the robustness and applicability of the model created. We observed that the data derived from Sentinel 1 allowed us to model, with great reliability, the NDVI of agricultural crops throughout the phenological cycle, making the methodology developed in this work a relevant solution for the monitoring of various regions, regardless of cloud cover.
Evaluating and monitoring forest areas during a restoration process is indispensable to estimate the success or failure of management intervention and to correct the restoration trajectory through ...adaptive management. However, the field measurement of several indicators in large areas can be expensive and laborious, and establishing reference values for indicators is difficult. The use of supervised classification techniques of high resolution images, combined with an expert system to generate management recommendations, can be considered promising tools for monitoring and evaluating restoration areas. The objective of the present study was to elaborate an expert system of management recommendation generation for areas under restoration, which were monitored by two different remote sensors: UAV (Unmanned Aerial Vehicle) and LiDAR (Light Detection and Ranging). The study was carried out in areas under restoration with about 54 ha and five years of implementation, owned by Fibria Celulose S.A. (recently acquired by Suzano S.A.), in the southern region of Bahia State, Brazil. We used images from Canon S110 NIR (green, red, near infrared) on UAV and LiDAR data compositions (intensity image, digital surface model, digital terrain model, normalized digital surface model). The monitored restoration indicator entailed land cover separated into three classes: Canopy cover, bare soil and grass cover. The images were classified using the Random Forest (RF) and Maximum Likelihood (ML) algorithms and the area occupied by each land cover classes was calculated. An expert system was developed in ArcGIS to define management recommendations according to the land cover classes, and then we compared the recommendations generated by both algorithms and images. There was a slight difference between the recommendations generated by the different combinations of images and classifiers. The most frequent management recommendation was “weed control + plant seedlings” (34%) for all evaluated methods. The image monitoring methods suggested by this study proved to be efficient, mainly by reducing the time and cost necessary for field monitoring and increasing the accuracy of the generated management recommendations.
Bioclimatic envelope models have been extensively used to predict the vegetation dynamics in response to climate changes. However, they are prone to the uncertainties arising from General Circulation ...Models (GCMs), classification algorithms and predictors, with low-resolution results and little detail at the regional level. Novel research has focused on the improvement of these models through a combination of climate and soil predictors to enhance ecological consistency. In this framework, we aimed to apply a joint edaphoclimatic envelope to predict the current and future vegetation distribution in the semiarid region of Brazil, which encompasses several classes of vegetation in response to the significant environmental heterogeneity. We employed a variety of machine learning algorithms and GCMs under RCP 4.5 and 8.5 scenarios of Coupled Model Intercomparison Project Phase 5 (CMIP5), in 1 km resolution. The combination of climate and soil predictors resulted in higher detail at landscape-scale and better distinction of vegetations with overlapping climatic niches. In forecasts, soil predictors imposed a buffer effect on vegetation dynamics as they reduced shifts driven solely by climatic drift. Our results with the edaphoclimatic approach pointed to an expansion of the dry Caatinga vegetation, ranging from an average of 16% to 24% on RCP 4.5 and RCP8.5 scenarios, respectively. The shift in environmental suitability from forest to open and dry vegetation implies a major loss to biodiversity, as well as compromising the provision of ecosystem services important for maintaining the economy and livelihoods of the world's largest semiarid population. Predicting the most susceptible regions to future climate change is the first step in developing strategies to mitigate impacts in these areas.
Display omitted
•Impacts of climate change on the Brazilian semi-arid region are assessed.•Vegetation distribution models are improved with the use of soil predictors.•Uncertainties from General Circulation Models are reduced in forecasts.•Climate change scenarios led to major impacts on the vegetation distribution.•Results showed the expansion of Caatinga vegetation in the RCP 4.5 and 8.5 scenarios.
Increasingly, applications of machine learning techniques for digital soil mapping (DSM) are being used for different soil mapping purposes. Considering the variety of models available, it is ...important to know their performance in relation to soil data and environmental variables involved in soil mapping. This paper investigated the performance of eight machine learning algorithms for soil mapping in a tropical mountainous area of an official rural settlement in the Zona da Mata region in Brazil. Morphometric maps generated from a digital elevation model, together with Landsat-8 satellite imagery, and climatic maps, were among the set of covariates to be selected by the Recursive Feature Elimination algorithm to predict soil types using machine learning algorithms. Mapping performance was assessed using the confusion matrix, and the Z-test among the Kappa indexes of the matrices. In a conventional soil survey, the soils described and classified in the Brazilian System of Soil Classification Argissolos Vermelho-Amarelos Distróficos – PVAd (Acrisols), Cambissolos Háplicos Tb Distróficos - CXbd (Cambisols), Gleissolos Háplicos Háplicos Tb Distróficos - GXbd (Gleysols), Latossolos Amarelos Distróficos - LAd (Xanthic Ferralsos), Latossolos Vermelho-Amarelos Distróficos - LVAd (Rhodic Ferralsols), and Neossolos Litólicos Distróficos - RLd (Neossols) were grouped into composite mapping units (MU) using the conventional method. The eight algorithms showed similar performance without statistical difference (Kappa 0.42-0.48). The mapping of soils with varying slopes (LAd, LVAd, CXbd) showed lower accuracy, whereas soils on hydromorphic lowlands (GXbd) were classified more accurately. In map algebra, the result was rather satisfactory, with 63-67 % agreement between the conventional soil map and maps produced by machine learning. The areas with the largest disagreement in the DSM occurred in the LAd unit due to subtle color variation in the Latossolos mantle without a clear relation to any environmental variable, highlighting difficulties in DSM regarding hill slope landforms. Model performance was satisfactory, and good agreement with the conventional soil map demonstrates the importance of the DSM as a potential complementary tool for assisting soil mapping in mountainous areas in Brazil for the purpose of land use planning.
In this study, we used a large national database to assess the rainfall erosivity (RE) patterns in time and space over the Brazilian territory. Thereby, RE and erosivity density (ED) values were ...obtained for 5166 rainfall gauges. Also, the concentration of the RE throughout the year and the RE's gravity center locations were analyzed. Finally, homogeneous regions regarding RE values were delimited and estimative regression models were established. The results show that Brazil's mean annual RE value is 5620 MJ mm ha−1 h−1 year−1, with considerable spatial variation over the country. The highest RE magnitudes were found for the north region, while the northeast region shows the lowest values. Regarding the RE's distribution throughout the year, in the southern region of Brazil, it is more equitable, while in some spots of the northeastern region, it is irregularly concentrated in specific months. Further analyses revealed that for most of the months, the RE's gravity centers for Brazil are in the Goiás State and that they present a north-south migration pattern throughout the year. Complementarily, the ED magnitudes allowed the identification of high-intensity rainfall spots. Additionally, the Brazilian territory was divided into eleven homogeneous regions regarding the RE patterns and for each defined region, a regression model was established and validated. These models' statistical metrics were considered satisfactory and, thus, can be used to estimate RE values for the whole country using monthly rainfall depths. Finally, all database produced are available for download. Therefore, the values and maps shown in this study are relevant for improving the accuracy of soil loss estimates in Brazil and for the establishment of soil and water conservation planning on a national scale.
Display omitted
•Rainfall erosivity and erosivity density values were obtained for 5166 rainfall gauges over the Brazilian territory.•The annual rainfall erosivity magnitudes range from 252 to 23,916 MJ mm ha−1 h−1 year−1.•The rainfall erosivity gravity center migrates latitudinally throughout the year.•The erosivity density analysis made possible the identification of high-intensity precipitation spots in all year long.•Regionalized regression models were established and validated for eleven homogeneous regions regarding the erosivity patterns.
Estimation of reference evapotranspiration (ETo) is very relevant for water resource management. The Penman-Monteith (PM) equation was proposed by the Food and Agriculture Organization (FAO) as the ...standard method for estimation of ETo. However, this method requires various weather data, such as air temperature, wind speed, solar radiation and relative humidity, which are often unavailable. Thus, the objective of this study was to compare the performance of multivariate adaptive regression splines (MARS) and alternative equations, in their original and calibrated forms, to estimate daily ETo with limited weather data. Daily data from 2002 to 2016 from 8 Brazilian weather stations were used. ETo was estimated using empirical equations, PM equation with missing data and MARS. Four data availability scenarios were evaluated as follows: temperature only, temperature and solar radiation, temperature and relative humidity, and temperature and wind speed. The MARS models demonstrated superior performance in all scenarios. The models that used solar radiation showed the best performance, followed by those that used relative humidity and, finally, wind speed. The models based only on air temperature had the worst performance.
It remains unclear whether temperature and precipitation exert independent control on tropical vegetation and soil C pools. Likewise, it is unknown whether the feedbacks of tropical C pools to ...climate constraints vary with nutrient availability. These aspects are critical to improving our ability to predict the response of tropical C pools to climate dynamics. This review aimed to assess climate data and the spatial distribution of vegetation and soil C pools across the Brazilian territory to investigate i) whether mean annual precipitation (MAP) and temperature (MAT) exert independent effects on tropical C pools; ii) whether vegetation and soil C pools exhibit hierarchical feedbacks to climate; and iii) how these feedbacks reflect soil nutrient availability. To account for MAP and MAT effects on tropical C cycling, we calculated Ecosystem Effective Moisture (EEM), i.e., the difference between MAP and potential evapotranspiration. We gathered substantial evidence suggesting that under high MAT and MAP controlling EEM, plants exchange more C for water and resorb more nutrients (especially P), which limitations in plant litter reduce microbial-derived C inputs into soil organic matter. Frequent soil saturation under high EEM favors denitrification rates (“open” N cycle), allowing continuous mineralization of litter and shallow soil C pools to release nutrients, sustaining high plant C pools. With decreasing MAP levels, ecosystem C pools depend on MAT controlling evapotranspiration and EEM. Accordingly, decreasing MAP under high MAT reduces EEM, with vegetation and soil C pools co-limited by low net primary productivity (NPP), frequent fire and/or nutrient losses. Otherwise, decreasing MAP and coupled to cool temperatures allow EEM to remain positive, forcing plants to increase deep-rooting and/or shed their leaves, which nutrients are immobilized with microbial-derived C into mineral-organic associations, favoring high soil C pools. Combined, the evidence gathered suggests that the sensitivity of tropical ecosystems to global increases in temperature should not be overlooked, especially if coupled to reductions in precipitation. Overall, the horizontal distribution of vegetation and soil C pools is best described by EEM rather than temperature or precipitation alone, whereas the vertical partition of C in plant-soil systems reflects biotic responses to climate-nutrient constraints.