Article

Estimación de cobertura arbórea mediante imágenes satelitales multiespectrales de alta resolución

Agrociencia, ISSN 1405-3195, Vol. 40, Nº. 3, 2006, pags. 383-394
Source: OAI

ABSTRACT Para evaluar la utilidad de imágenes de alta resolución espacial, se compararon mediciones directas en campo como densitómetro esférico, muestreo de intersección de líneas y estimadores de razón y regresión, con procedimientos que utilizan imágenes satelitales de alta y mediana resolución espacial (Ikonos, QuickBird-2, SPOT-4 y LANDSAT-7) para estimar el porcentaje de cobertura de copa arbórea en un pinar ubicado en el ejido San Rafael Ixtapalucan, Municipio de Tlahuapan, Estado de Puebla, México. Las imágenes Ikonos y QuickBird son adecuadas técnicamente para determinar el porcentaje de cobertura de copa en este tipo de vegetación. Sin embargo, es conveniente considerar los costos de la compra de las imágenes y que se requiere de personal capacitado para su uso.

0 Followers
 · 
86 Views
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Abstract. Remote sensors combined with geospatial analytical methods provide important tools for measuring forest biophysical variables at a lower cost and at broader spatial and temporal scales than traditional forest inventories. The objective of this study was to analyze the relationship between data of the National Inventory of Forest and Soils (INFyS) of Mexico and spectral data from images of the SPOT platform for spatial estimation of basal area, timber volume and overlapping tree cover in the temperate and mesophyll forests of Hidalgo, Mexico. Four approaches to the analysis were used to generate models that describe the inventory and the distribution of the variables of interest: 1) multiple linear regression, 2) K-nearest neighbor (K-nn), 3) ratio and regression estimators, and 4) traditional forest inventory. The estimations derived from the first three methods are within the confidence interval of 95 % of the traditional forest inventory, and the values derived from ratio and regression estimators produced narrower confidence intervals. The analysis of the results indicate significant correlation between the INFyS data and the spectral bands of the SPOT satellite, particularly with the green, near infrared and mid infrared bands, as well as with the indexes and simple ratios based on these bands. Key words: remote sensors, forest inventory, basal area, volume, overlapping tree cover. Resumen. Los sensores remotos, en combinación con métodos de análisis geoespacial, ofrecen herramientas importantes para la medición de variables biofísicas del bosque con costo menor que el inventario forestal tradicional y en escalas espaciales y temporales mayores. El objetivo de este estudio fue analizar la relación entre los datos del Inventario Nacional Forestal y de Suelos (INFyS) de México y datos espectrales provenientes de imágenes de la plataforma SPOT para estimar espacialmente el área basal, el volumen maderable y la cobertura arbórea trasla-pada en los bosques templado y mesófilo de Hidalgo, México. Cuatro enfoques de análisis se aplicaron para generar modelos que describen el inventario y la distribución de las variables de interés: 1) regresión lineal múltiple, 2) K vecino más cercano (K-nn), 3) estimadores de razón y regresión y, 4) inventario fo-restal tradicional. Las estimaciones derivadas de los tres primeros métodos se encuentran dentro del intervalo de confianza del 95 % del inventario forestal tradicional y los valores deriva-dos mediante estimadores de razón y regresión produjeron los intervalos de confianza mas estrechos. El análisis de los resulta-dos indicó la correlación significativa entre los datos del INFyS y las bandas espectrales del satélite SPOT, particularmente con la verde, el infrarrojo cercano e infrarrojo medio, así como con índices y cocientes simples basados en estas bandas. Palabras clave: sensores remotos, inventario forestal, área basal, volumen, cobertura arbórea traslapada.
    Agrociencia 12/2014; 48:847-862. · 0.05 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents relations between spectrum data of the SPOT 5 HRG spatial high resolution sensor and aboveground tree carbon Mg ha(-1)) in a Pinus patula forest in Zacualtipan, Hidalgo, Mexico. First it was necessary to quantity the biomass (Mg ha(-1)). The multiple linear regression and the non parametric method of the nearest neighbor (k-nn) were used. The analysis of results suggests the presence of a high correlation between forest variables and the spectrum indexes associated with vegetation moisture. During validation, the correlation coefficients between the values observed and estimated for the regression methods and k-nn were highly significant (p = 0.01), and showed their potential for predicting the presence of aboveground tree carbon. The root mean square error (RCME) of the k-nn estimates was 22.24 Mg ha(-1) (35.43 %). The total estimate calculated by using k-nn was the closest to that obtained through traditional stratified sampling. From the results obtained, the contribution of the SPOT 5 images and k-nn method to the development of carbon inventories is confirmed.
    Agrociencia 01/2009; 43(2):209-220. · 0.05 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper presents relations between spectrum data of the SPOT 5 HRG spatial high resolution sensor and aboveground tree carbon Mg ha–1) in a Pinus patula forest in Zacualtipán, Hidalgo, México. First it was necessary to quantify the biomass (Mg ha–1). The multiple linear regression and the non parametric method of the nearest neighbor (k–nn) were used. The analysis of results suggests the presence of a high correlation between forest variables and the spectrum indexes associated with vegetation moisture. During validation, the correlation coefficients between the values observed and estimated for the regression methods and k–nn were highly significant (p = 0.01), and showed their potential for predicting the presence of aboveground tree carbon. The root mean square error (RCME) of the k–nn estimates was 22.24 Mg ha–1 (35.43 %). The total estimate calculated by using k–nn was the closest to that obtained through traditional stratified sampling. From the results obtained, the contribution of the SPOT 5 images and k–nn method to the development of carbon inventories is confirmed.
    Agrociencia 03/2009; 43(2):209-220. · 0.05 Impact Factor

Full-text (2 Sources)

Download
11 Downloads
Available from
Aug 12, 2014