Livestock Research for Rural Development 28 (3) 2016 Guide for preparation of papers LRRD Newsletter

Citation of this paper

Variation in the Normalized Difference Vegetation Index (NDVI) in dairy farms in northern Antioquia

Anderson Bastidas Duque*, Rolando Barahona Rosales1 and Mario Cerón-Muñoz

Research Group on Animal Genetics, Breeding and Modeling (GaMMA), Facultad de Ciencias Agrarias, Universidad de Antioquia, Colombia.
1 BIOGEM Research Group, Faculty of Agrarian Sciences, Universidad Nacional de Colombia, sede Medellín, Colombia.
* Joven Investigador Colciencias


Grassland vegetation cover is a factor of great importance to productivity and environmental impact of livestock, thus proper monitoring should be a priority for tropical livestock production. A novel way to do this is monitoring the normalized difference vegetation index (NDVI), calculated by the relationship between the red and near-infrared bands of satellite images, which allows identification of areas with high or low vegetation cover. The present study was undertaken with the purpose of analyzing the variation in NDVI in the north of Antioquia dedicated to dairy farming between 1995 and 2014. To do this, the vegetation index of 103 dairy farms in northern Antioquia was evaluated, using a generalized additive mixed model including the softened effects of the variables latitude, longitude, date and farm. The NDVI varied between 0.4 and 0.8, showing an average value of 0.61 and the municipalities with the greater vegetation index during the period evaluated were: San Pedro de los Milagros, Entrerrios and Donmatías. By contrast, Santa Rosa de Osos and San Jose de la Montana were the municipalities that had the lowest NDVI values. The NDVI data obtained suggests that following the implementation of dairy farming, a stable vegetation index has been maintained over time in the region studied, with no evidence of significant changes in vegetation cover. These findings should be corroborated following an on-site approach which should also be aimed at understanding the main factors underlying this response.

Keywords: livestock, pasture, remote sensing, vegetation cover


La cobertura vegetal de las praderas es un factor de gran importancia para la productividad y el impacto ambiental de la ganadería, con lo que su monitoreo adecuado debe ser una actividad prioritaria para la ganadería tropical. Una forma novedosa de realizar este monitoreo es a través del índice de vegetación de diferencia normalizada (NDVI), calculado por medio de la relación entre la banda del rojo e infrarrojo cercano de imágenes satelitales, que permite identificar áreas con alta o baja cobertura vegetal. El presente estudio se realizó con el propósito de analizar la variación del NDVI, en la zona norte de Antioquia dedicada a la ganadería de leche entre el año 1995 y 2014. Para esto, se evaluó el índice de vegetación de 103 fincas lecheras del norte de Antioquia, mediante un modelo mixto aditivo generalizado incluyendo los efectos suavizados de las variables latitud, longitud, fecha y finca. El NDVI varió entre 0.4 y 0.8, presentando un valor promedio de 0.61 y los municipios que presentaron un mayor índice de vegetación durante el periodo evaluado fueron: San Pedro de los Milagros, Entrerrios y Donmatías. Por el contrario, Santa Rosa de Osos y San José de la Montaña fueron los municipios que presentaron los menores valores de NDVI. La información de NDVI sugiere que en la zona de estudio, desde la implementación de la ganadería de leche, el índice vegetación se ha mantenido estable en el tiempo, sin existir evidencia de cambios en cobertura vegetal. Estos hallazgos deben ser corroborados siguiendo una aproximación de sitio, que debe también ser dirigida a contribuir a la comprensión de los principales factores responsables de esta respuesta.

Palabras clave: cobertura vegetal, ganadería, pastos, teledetección


Antioquia is the department with the greatest milk production in Colombia, producing of approximately 3.5 million liters / day; most of which (70%) are produced in the north subregion (MADR 2012), In recent years, milk production has been growing due to the increase of cattle inventory, together with the adoption of breeding programs and pasture improvement (CorlacR2 2012).

The northern highland Antioquia forming the Central Cordillera north of the department of Antioquia, is characterized by having mild to moderate slopes crossed by a dense network of drainages (Escobar 2006) .The majority of soils are Hapludands and Fulvudands (Jaramillo 2002), which are poorly developed soils originated from black, sandy loam and friable volcanic ash, slightly acidic, of high cation exchange capacity, low base saturation, low available phosphorus and moderate fertility (Zapata 2006).

According to Gómez (2010), this area is covered mainly by grass pastures (37.9%), natural forests (22.1%), secondary vegetation (21%) and annual crops (9.7%). The Kikuyu grass (Cenchrus clandestinum (Hochst ex Chiov)) is the main grass in the specialized systems bovine milk production in the high tropics of Antioquia (Jaimes et al 2015), due to their habit of stoloniferous growth, high aggressiveness and resistance to trampling (Ramirez 2013; Ruiz et al 2014). It is a grass with high nitrogen requirements, so fertilization takes place after grazing to maintain an adequate supply of biomass (Mejia-Taborda et al 2014).

The floristic diversity of pastures is lower than that observed in the stubble and logged forests, due to the structural homogeneity of the vegetation. Plants from the Poaceae, Cyperaceae and Asteraceae families thrive there, at the edges of the pastures is common to find ferns and among the trees and bushes it is common to observe Cupressus lusitanica (cypress), Eucalyptus sp. (eucalyptus), Pinus patula (pine), Tibouchina lepidota (lasiandra), Myrcia sp. (myrtle) and Weinmannia pubescens (encenillo) among others (Robles 2006).

Over the years, this part of Antioquia has been subjected to different types of land use, with mining being one of the most common, from where soils are then devoted to agricultural activities which include livestock production, as it was the case in northern Antioquia. Together with these processes, there are changes in vegetation cover, either increasing or decreasing it and these variations are also influenced by climatic conditions and human activity (Escobar 2006).

Remote sensing is a technique that allows to gather information without a physical contact (Sobrino 2010). Through the use of satellites, images of the Earth's surface are obtained that can be used in various medium and long term studies of plant and soil processes, forest degradation analysis and detection of land use changes (Meneses-Tobar 2011).

The use of vegetation indices has become an important tool in agriculture, and it allows to monitor areas devoted to livestock production, as in the case of grazing systems used in specialized dairy farms where there is a periodic variation due to grazing (Barrachina et al 2009). The use of satellite images allows monitoring the productivity of pastures over time and identify periods of lower production of biomass, and the presence of diseases that reduce the vegetation index values in the pastures or areas evaluated (Escribano and Hernández Díaz-Ambrona 2013).

There are different techniques applied to the monitoring of vegetation, among which are those of radar, LIDAR and optical type. The radar wave sensors send signals that allow taking pictures in the presence of clouds and darkness. They also allow typifying the vertical structure of forest areas (Zapata and Anaya 2011; Marchionni and Cavayas 2014). LIDAR sensors use laser beams to measure directly the distribution of the vegetation canopy in three dimensions, in addition to terrain topography. These provide better information, but their current limitation are high costs and lower availability of images to study. Meanwhile, satellite images captured by optical sensors have a high temporal resolution and are inexpensive. Although these images are limited in clouded conditions and allow less accurate estimates in areas of very dense vegetation, there have been multiple studies on the variation of the growth of pastures through their use, specifically through the Normalized Difference Vegetation Index (NDVI) (Zapata and Anaya 2011). Although there are several methods for using satellite images to study changes in vegetation cover (Chuvieco 1998), the use of NDVI enables classification of different soil covers of a given area and differentiate active or vigorous vegetation from senescent or ill vegetation, by contrasting the amount of visible (60um - 70um) and infrared (70um - 90um) light which is reflected by green vegetation (López 2012).

One use of the NDVI is to evaluate how vegetation changes reflect the dynamics of the environment, as the vegetation is an important indicator of the effects of climate change, which is depicted by the distribution of vegetation in time and space (Aguilar et al 2012; Rodriguez-Moreno and Bullock 2013). In addition, NDVI is used to estimate the quantity, quality and growth of the vegetation (Hernández et al 2014). Thus, the use of satellite images to monitor vegetation is common and has been used to monitor the phenology in different crops (Hmimina et al 2013).

The aim of this study was to analyze the variation of Normalized Difference Vegetation Index (NDVI) in areas in the north of Antioquia dedicated to dairy farming between 1995 and 2014.

Materials and Methods

Information of NDVI was calculated from images between 1995 and 2014 of the satellites Landsat 5,7 and 8 on 103 dairy farms located in the municipalities of San Pedro de los Milagros, Entrerrios, Santa Rosa de Osos, Donmatías and San José de la Montaña, in the northern region of Antioquia department (see Figure 1). These farms were associated to the "Strengthening of the dairy production chain in the Northern District of Antioquia" project carried by University of Antioquia and the National University of Colombia - Medellín and funded by the Colombian General Royalties System and the Government of Antioquia.

Figure 1. Location of the municipalities in the department of Antioquia, where the vegetation index data were obtained.

A total of 88 satellite images were obtained which were subjected to an atmospheric correction on the B3 (red) and B4 (near infrared) bands when images came from Landsat 5 and 7, and bands B4 (red) and B5 (near infrared) when images came from Landsat 8, using the dark object subtraction method to estimate the amount of light dispersion due to particles and aerosols in the atmosphere and subtract it digitally. In addition, a mask of valid pixels was applied, where the pixels are classified into three categories: Clouds, no-clouds and gaps and edges. This was carried out through a mask generated by an unsupervised "k-means" classification, whose fundamental principle is the classification of the pixels in k number of classes by proximity to their means. Subsequently, the image was cut to the size of the study area. To do this, a survey of each farm was obtained by GPS and a shape of the perimeter to each farm was created. The index was calculated by the following relationship:

NDVI: (ρNIR – ρR) / (ρNIR + ρR)

Where: ρNIR: near infrared band and ρR: red band

To model the relationship between vegetation index and the other variables, the following information of the image was taken into account: latitude (m, between 6.39 °N and 6.87 °N), longitude (p, between -75.68 °W and -75.35 °W), year (between 1995 and 2014), day of the year (d, between 1 and 365) and Julian date (j) and located on the farm (f). Analyses were performed using generalized additive mixed models (GAMM), using the MGCV library (Wood 2011) of the R Software project (R Core Team 2015). A total of 18 models were evaluated; in which the variables were subjected to different interactions and smoothing effects to determine the model that better relates the vegetation index with the other variables (Table 1).

Results and discussion

The value of the obtained NDVI interval varies between minus one (-1) and plus one (+1). Of these, only the positive values correspond to areas of vegetation, values close to zero signify no vegetation, and negative values belong to clouds, snow, water, areas of bare soil and rocks (Escribano and Hernández Díaz-Ambrona 2013).

Table 1. Generalized mixed additive models and distribution family used to assess the vegetation index in the northern region of Antioquia.
Number Model Family
1 yijklm = α + s(ai ,dj) + s(mk, pl) + eijklm Gaussian
2 yijkl = α + s(di) + s(mj, pk) + eijkl Gaussian
3 yijkl = α + s(ji) + s(mj, pk) + eijkl Quasi
4 yijkl = α + s(ji) + s(mj, pk) + eijkl Poisson
5 yijkl = α + s(ji) + s(mj, pk) + eijkl Quasipoisson
6 yijklm = α + s(ji) + fj+ te(mk, pl) + eijklm Quasi
7 yijklm = α + s(ji) + fj+ te(mk, pl) + eijklm Poisson
8 yijklm = α + s(ji) + fj+ te(mk, pl) + eijklm Quasipoisson
9 yijk = α + s(ji) + s(fj) + eijk Poisson
10 yijk = α + s(ji) + s(fj) + eijk Quasipoisson
11 yijk = α + s(ji) + s(fj) + eijk Quasi
12 yijk = α + s(ji) + s(fj) + eijk Gaussian
13 yijk = α + te(mi, pj) + eijk Gaussian
14 yijk = α + te(mi, pj) + eijk Quasi
15 yijk = α + te(mi, pj) + eijk Poisson
16 yijk = α + te(mi, pj) + eijk Quasipoisson
17 yijk = α + s(mi, pj) + eijk Poisson
18 yijk = α + s(mi, pj) + eijk Quasipoisson
y is the value of the vegetation index, α is the intercept, s is the smoothed function of one variable or a combination of two variables, te is the tensor product of two variables, m is the meridian, p is the parallel, a is the year, d is the day of the year, j is the Julian date, f is the farm, and e is the residue.

The model with the greater R2 (0.32) and lower BIC, was the one that considered the combination of year (ai) and the day of the year (dj) as a smoothed function and the combination of latitude (mk) and longitude (pl):

yijklm = α+ s(ai, dj) + s(mk, pl) + eijklm

As shown in Figure 2, in the dairy region of northern Antioquia, the vegetation index presented low values (<0.50) in January, February, November and December throughout the period evaluated. More specifically, between 1995 and 2002, the NDVI showed its lowest value (0.40) in days 1 to 20 (January) and 320 to 345 days (November and December), showing highly significant differences (p <0.01). In analyzing figure 2, it must be remembered that IDEAM (2007; 2012) reported that between 1997 and 1998 occurred the strongest El Niño of the millennium in Colombia, resulting in a considerable reduction in the volume of rainfall in the region. In addition, during 2002 and 2010 occurred other droughts of lower intensity. Increased vegetation index (> 0.60) between days 60 and 100 (March and April) coincided with the start of the rainy season reported for the Andean region (IDEAM 2012). De la Casa and Obando (2006) found a relationship between vegetation index and precipitation, indicating that plant development was reduced when rainfall was less than 40mm/month.

The northern region of Antioquia had NDVI values between 0.40 and 0.80 during the period evaluated, and thus the vegetation index was found near the ranges reported by other authors. Roldán and Poveda (2006) assessed the temporal variability of NDVI in five regions of Colombia and found that for the Andean region NDVI values were between 0.38 and 0.69. In turn, for pasture-only zones in Spain, Escribano and Hernández Díaz-Ambrona (2013) reported that NDVI ranged between 0.20 and 0.68.

Figure 2. Temporal variation of vegetation index from 1995 to 2014 and the days 1 to 365 of the year in the municipalities of
San Pedro de los Milagros, Donmatías, Entrerrios, Santa Rosa de Osos and San José de la Montaña, Antioquia.

While the lowest NDVI values reported in the current study (between 0.40 and 0.50) are not as low as those of other studies (Escribano and Hernández Díaz-Ambrona 2013; Jobbágy et al 2013; Hernández et al 2014), the temporal variation observed in NDVI showed periods of significantly diminished vegetation cover. Cristiano (2010) found that vegetation decreased between 39 and 45% when it was under conditions of water and nutritional stress, due to a reduction in the growth of stems and leaves. Similarly, Robles (2006) mentioned that when the vegetation index values were low, the growth of the vegetation cover decreased, temperature control was reduced and the presence of eroded areas increased.

Figure 2 shows that in the last 10 years grass growth rate increased and it is important to identify the reasons for this increase, as it may be associated to either increased environmental temperature or greater agricultural input (i.e. fertilizer) usage. While livestock production has been regarded as a process that transforms vegetation cover (Moreno and Cuartas 2015), the temporal variation observed in this study suggests that the vegetation cover has been maintained over time, an important observation due to the broken topography of the region and the different types of use to which that land has been subjected to (Escobar 2006). Due to increased food demand, the percentage of land dedicated to pasture production has increased in this region. Whether this would prompt producers to adopt different agricultural practices to ensure the sustainability of their production remains to be demonstrated. However, it is clear that farm productivity can be ensured through the adoption of improvements in pasture management (Murgueitio et al 2015) or changes in stocking rates (Sossa and Barahona 2015) and that in some cases, this also leads to increased vegetation cover (Cuartas et al 2014).

In the highland tropics, it is possible to obtain increases in vegetation cover through the adoption of different practices which include the application of proper fertilization rates and the implementation of new grazing systems such as silvopastoral systems. As for fertilization, Mejía-Taborda et al (2014) and Ruiz et al (2014) reported that the use of different doses of nitrogen and compound fertilizers (> 50kg / ha) increased grass biomass per hectare between 88 and 170%, respectively. As for silvopastoral systems, Mahecha (2002) and Rivera et al (2015) reported increases of 30% in the availability of forage, ensuring sustainable growth of the vegetation and adequate soil cover (Ojeda et al 2003).

Improvements in vegetation cover also included changes in the critical times for grazing in the region. This shifts in NDVI could be associated to environmental factors and/or management changes (De Bello 2006; Paz-Pellat et al 2009). This is observed between days 300 and 365, that early during the evaluation period (around 1995) was the point of the year when there was lower NDVI, and that has shown noticeable increases in NDVI as the evaluation period progressed towards 2014. The opposite occurred for the year days 35-120, when NDVI has slowly declined in recent years (Figure 2). Thus, the vegetation index evaluated across the different municipalities was affected by the change in environmental variables and the conditions of the terrain (Verhulst et al 2010). As grazing is an activity that leads to temporary increases and decreases of vegetation cover, further analysis of NDVI should be improved by tracking changes in animal numbers in the area.

The effect of the smoothed function of the interaction between longitude and latitude was highly significant (p <0.01), indicating that there are differences in the amount of vegetation due to the geographical position of the municipalities included in the study. As shown Figure 3, the NDVI showed a homogeneous spatial variation with an average value of 0.61, indicating that the distribution of vegetation cover in the monitored region is within the ranges reported in the literature, similar to the value of 0.60 determined by Roldán and Poveda (2006) for the Andean region, while for grassland and plantations areas in Argentina, Jobbágy et al (2013) found NDVI averages of 0.48 and 0.73, respectively.

On Figure 3, those in yellow are areas of low NDVI values, and these were more evident in some areas of the municipalities of Santa Rosa de Osos and San José de la Montaña, suggesting that production conditions within those farms affected the proper development of a healthy vegetation cover. One reason for this could be having greater stocking rates which could be the case of Santa Rosa de Osos, where land is expensive. Another factor that may have limited the development of a healthy vegetation cover in these municipalities can be related to the topographical conditions of the terrain. Indeed, Lasanta et al (2004), Lasanta and Vicente-Serrrano (2006) reported that the slope, altitude and shape of the slope are important variables in the spatial distribution of vegetation, with the areas of gentle slopes being both the most grazed by cattle and where most pressure is placed on the growth of vegetation. By contrast, farms in the municipalities of San Pedro de los Milagros, Entrerrios, and Don Matías had the greater NDVI values, identifying the southern part of the region evaluated as the one that allows better development of the vegetation. In addition, the pig-grasses-milk system frequently found in the northern highlands of Antioquia, presents further development in towns like Don Matías and San Pedro de los Milagros (Quirós et al 1997). This allows obtaining a great amount of organic fertilizers that contribute to the growth and development of kikuyu pastures.

Figure 3. Average NDVI values in the municipalities of San Pedro de los Milagros, Don Matías,
Entrerrios, Santa Rosa de Osos and San José de la Montaña, Antioquia.


Although the northern region of Antioquia has been subjected for many years to the agronomic and cultural practices required in dairy production, it has a vegetation index within the range reported for other regions. Analysis of the NDVI data suggests that following the implementation of dairy farming, a stable vegetation index has been maintained over time in that region, and following this approach, no significant reductions in vegetation cover are evident. These findings should be corroborated following an on-site approach which should also be aimed at understanding the main factors underlying this response.


This work is part of the project "Fortalecimiento de la producción de la cadena láctea del distrito Norte Antioqueño", agreement No. 2013AS180031 signed between the Ministry of Agriculture and Rural Development, the Department of Antioquia, la Universidad Nacional de Colombia, Medellín and the University of Antioquia, with resources from the Sistema General de Regalías - SGR. The support from the Committee for the development of the investigation CODI (Estrategia para la Sostenibilidad 2016 grupo GaMMA) and from the Jóvenes Investigadores e Innovadores Program of COLCIENCIAS by the first author (call 645 2014) is also gratefully acknowledged.


Aguilar C, Zinnert J, Polo M and Young D 2012 NDVI as an indicator for changes in water availability to woody vegetation. Ecological Indicators. 23:290-300.

Barrachina M, Cristóbal J, Tulla A y Pons X 2009 Análisis de la producción de biomasa de los prados y pastos de Vall Fosca (Pirineo Axial-Nogueres). Teledetección: Agua y desarrollo sostenible. XIII Congreso de la Asociación Española de Teledetección. pp.225-228.

CorlacR2 (Consejo Regional Lácteo) 2012 Acuerdo de competitividad de la cadena láctea de Antioquia. Consejo Regional Lácteo. pp. 26.

Cuartas C A, Naranjo J F, Tarazona A M, Murgueitio E, Chará J D, Ku J, Solorio F J, X Flores M X, Solorio B and Barahona R 2014 Contribution of intensive silvopastoral systems to animal performance and to adaptation and mitigation of climate change. Revista Colombiana de Ciencias Pecuarias. 27(2):76-94.

Chuvieco E 1998 El factor temporal en teledetección: evolución fenomenológica y análisis de cambios. Revista de Teledetección. N° 10:1-9.

Cristiano P M 2010 Estimación de la productividad primaria mediante el modelo de eficiencias y sensores remotos. Biblioteca digital de la Facultad de Ciencias Exactas y Naturales – Universidad de Buenos Aires. Tesis. p. 107.

De Bello F 2006 Consecuencias de cambios de presión ganadera sobre la estructura de la vegetación a lo largo de gradientes climáticos. Ecosistemas. 15(2): 106-112.

De La Casa A y Obando G 2006 Relación entre la precipitación y los índices de vegetación durante el comienzo del ciclo anual de lluvias en la provincia de Córdoba, Argentina. Instituto Nacional de Tecnología Agropecuaria (INTA). Revista de Investigaciones Agropecuarias. 35(1): 67-85.

Escobar L 2006 Plan de acción trienal 2007-2009. Corporación Autónoma Regional del Centro de Antioquia. CORANTIOQUIA. Versión 1. pp. 57.

Escribano J and Hernández Díaz-Ambrona C 2013 Estimación de la producción de pastos en Dehesas mediante índices de vegetación. pp. 465-472.

Gómez M 2010 Fenología reproductiva de especies forestales nativas presentes en la jurisdicción de CORANTIOQUIA, un paso hacia su conservación. Corporación Autónoma Regional del Centro de Antioquia, CORANTIOQUIA. Vol 1. pp. 228.

Hernández C, Escribano J and Tarquis A 2014 Comparación del Índice de Diferencia Normalizada obtenido a diferentes escalas en pastos de Dehesa. En: 53ª Reunión Científica de la Sociedad Española para el Estudio de los Pastos. Potes Cantabria. Pastos Y Pac -2014 2020:121-128.

Hmimina G,Dufrene E, Pontailler J, Delpierre N, Aubinet M, Caquet B, Grandcourt A, Burban B, Flechard C, Granier A, Gross P, Heinesch B, Longdoz B, Moureaux C, Ourcival J, Rambal S, Saint André S and Soudani K 2013 Evaluation of the potential of MODIS satellite data to predict vegetation phenology in different biomes: An investigation using ground-based NDVI measurements. Remote Sensing of Environment. 132: 145-158.

IDEAM 2007 Modelo institucional del IDEAM sobre el efecto climático de los fenómenos El Niño y La Niña en Colombia. Bogotá. pp. 81.

IDEAM 2012 Posibles efectos naturales y socioeconómicos del fenómeno “El Niño” en el periodo 2012 – 2013 en Colombia. Bogotá. pp.22.

Jaimes L, Cerón J y Correa H 2015 Efecto de la época del año y la etapa de lactancia sobre el consumo alimenticio de vacas Holstein pastoreando kikuyo (Cenchrus clandestinus) en Colombia. Livestock Research for Rural Development 27, ·Article #244. from

Jaramillo D 2002 Introducción a la ciencia del suelo. Universidad Nacional de Colombia. Facultad de Ciencias. Medellín. pp. 613.

Jobbágy E G, Acosta A M and Nosetto M D 2013 Rendimiento hídrico en cuencas primarias bajo pastizales y plantaciones de pino de las sierras de Córdoba (Argentina). Ecología Austral 23(2): 87-96.

Lasanta T, Vicente-Serrano S M and Romo A 2004 Influencia de la topografía en la estacionalidad de la actividad vegetal: análisis en el Pirineo occidental Aragonés a partir de imágenes NOAA-AVHRR. Boletín de la A.G.E. Nro 38: 175-197.

Lasanta T and Vicente-Serrano S M 2006 Factores en la variabilidad espacial de los cambios de cubierta vegetal en los Pirineos. Cuadernos de investigación geográfica. 32:57-80.

López P 2012 Tendencia del Índice de Vegetación de Diferencia Normalizada para una zona del centro de México. Universidad Autónoma de Querétaro, Facultad de Ciencias Naturales. Tesis. pp. 42.

MADR (Ministerio de Agricultura y Desarrollo Rural) 2012 Política del sector lácteo en Antioquia.

Mahecha L 2002 El silvopastoreo: una alternativa de producción que disminuye el impacto ambiental de la ganadería bovina. Revista Colombiana de Ciencias Pecuarias. 15(2): 226-231.

Marchionni D y Cavayas F 2014 La teledetección por radar como fuente de información litológica y estructural. Análisis espacial de imágenes SAR de RADARSAT-1. GEOACTA. 39(1): 62-89.

Mejía-Taborda A C, Ocho-Ochoa R and Medina-Sierra M 2014 Efecto de diferentes dosis de fertilizante compuesto en la calidad de pasto kikuyo (Pennisetum clandestinum Hochst. Ex. Chiov.). Pastos y Forrajes. 37(1): 31-37.

Meneses-Tobar C 2011 El índice normalizado diferencial de la vegetación como indicador de la degradación del bosque. Revista Internacional de Silvicultura e Industrias Forestales. 62:39-46.

Moreno Betancur D J y Cuartas Hernández S E 2015 Sobrevivencia y crecimiento de plántulas de tres especies arbóreas en áreas de Bosque Montano Andino degradadas por ganadería en Colombia. Acta Biológica Colombiana. 20(2):85-100.

Murgueitio E, Xóchitl Flores M, Calle Durán Z, Chará J D, Barahona R, Molina Durán C H and Uribe F 2014 Productividad en Sistemas silvopastoriles intensivos en América Latina. En: Florencia Montagnini, Eduardo Somarriba, Enrique Murgueitio, Hugo Fassola, Beatriz Eibl (eds.) Sistemas agroforestales. Funciones productivas, socioeconómicas y ambientales. CATIE, Turrialba, Costa Rica. Editorial CIPAV, Cali, Colombia. Informe Técnico Nº 402 pp 59-101.

Ojeda P, Restrepo J, Villada D and Gallego J 2003 Sistemas silvopastoriles, una opción para el manejo sustentable de la ganadería. Fundación para la investigación y desarrollo agrícola (Fidar). Programa Nacional de Transferencia de Tecnología Agropecuaria (PRONATTA). Centro Internacional de Agricultura Tropical (CIAT).pp. 54.

Paz-Pellat F, Odi-Lara M, Cano-Gozález A, Bolaños-González M and Zarco-Hidalgo A 2009 Equivalencia ambiental en la productividad de la vegetación. Agrociencia. 43(6): 635-648.

Quirós D, Arévalo A, Lopera R, Loaiza C y Suárez G 1997 Estudio descriptivo de los factores relacionados con el desarrollo del arreglo cerdos-pastos-leche, como sistema de producción en el altiplano norte de Antioquia. Corporación Colombiana de Investigación Agropecuaria. Boletín de investigación 6. pp. 96.

R Core Team 2015 R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.URL

Ramirez A 2013 Evaluación del crecimiento de pastos usando índices de vegetación calculados a partir de información satelital. Tesis. Universidad Nacional de Colombia-sede Medellín. pp. 91.

Rivera J E, Cuartas C A, Naranjo J F, Tafur O, Hurtado E A, Arenas F A, Chará J and Murgueitio E 2015 Efecto de la oferta y el consumo de Tithonia diversifolia en un sistema silvopastoril intensivo (SSPi), en la calidad y productividad de leche bovina en el piedemonte Amazónico colombiano. Livestock Research for Rural Development. Article #189. 27(10):1-9.

Robles C 2006 Caracterización de la diversidad y uso de la flora silvestre en el municipio de Donmatías-Antioquia. Corporación Autónoma Regional del Centro de Antioquia. Programa Conocimiento y Mejoramiento de los Recursos Naturales. Proyecto Manejo y Conservación de la Flora. pp. 76.

Rodríguez-Moreno V M and Bullock S H 2013 Comparación espacial y temporal de índices de la vegetación para verdor y humedad y aplicación para estimar LAI en el desierto Sonorense. Revista Mexicana de Ciencias Agrícolas. 4(4): 611-623.

Roldán P and Poveda G 2006 Variabilidad espacio-temporal de los índices NDVI y EVI. Aplicación a cinco regiones colombianas. Meteorología colombiana. 10:47-59.

Ruiz J D, Villar D, Correa H J, Noreña J M, Roldán M and Rios J C 2014 Niveles de nitrato en pasto kikuyo (Cenchrus clandestinus (Hochst.- ex Chiov.) Morrone) fertilizado con urea en el altiplano de Antioquia, Colombia. Revista CES Medicina Veterinaria y Zootecnia. 9(1): 52-57

Sobrino J 2010 Teledetección. Universitat de Valencia. España. pp. 449.

Sossa C P and Barahona R 2015 Comportamiento productivo de novillos pastoreando en trópico de altura con y sin suplementación energética. Revista de la Facultad de Medicina Veterinaria y Zootecnia. 62(1): 67-80.

Verhulst N, Govaerts B and Fuentes Ponce M 2010 Sensor portátil GreenSeekerTM para la medición del índice diferencial de vegetación normalizado (NDVI): Una herramienta para la evaluación integral de las prácticas agronómicas. Parte A: Conceptos y estudios de caso. México, D.F.; CIMMYT. pp. 16.

Wood S N 2011 Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society (B) 73(1):3-36.

Zapata B y Anaya J 2011 Los sensores remotos en los proyectos de mitigación de gases de efecto invernadero. Revista Entorno Geográfico. 7-8: 66-83.

Zapata R 2006 Química de los procesos pedogenéticos. Facultad de Ciencias. Universidad Nacional-Sede Medellín. pp. 358.

Received 31 December 2015; Accepted 29 January 2016; Published 1 March 2016

Go to top