Numéro |
Ann. de l’Associat. Internat. de Climatologie
Volume 1, 2004
|
|
---|---|---|
Page(s) | 133 - 156 | |
DOI | https://doi.org/10.4267/climatologie.1058 | |
Publié en ligne | 9 octobre 2015 |
Estimation des précipitations par télédétection au Mato Grosso (Brésil)
Rainfall estimating using remote sensing in Mato Grosso (Brazil)
1
COSTEL / LETG - UMR 6554 CNRS, Université Rennes 2
2
IRD / Agência Naçional da Agua, Brasilia
3
Universidade Federale do Mato Grosso a Cuiaba
a
vincent.dubreuil@uhb.fr
b
josyane@ana.gov.br
c
maitelli@terra.br
Dans les régions des fronts pionniers du centre du Mato Grosso, les rendements des cultures implantées par les colons depuis une vingtaine d’années (soja, canne à sucre, coton, maïs) sont encore largement tributaires de la variabilité pluviométrique interannuelle; cependant l’estimation à échelle fine des précipitations est impossible étant donné les lacunes du réseau d’observation au sol. Nous avons choisi d’utiliser les données infrarouge GOES-8 à pleine résolution (4km) pour essayer de répondre à cette question. Un premier jeu de données a été constitué en retenant pour chaque mois la valeur maximale de la température observée pour chaque pixel afin de ne conserver que l’émission en provenance du sol dont l’intensité est surtout fonction de la quantité de pluie reçue. Un second jeu de données a été constitué en seuillant les mêmes images à partir de –40°C : ainsi, l’occurrence des nuages à sommet froid pluviogènes peut être également suivie au pas de temps mensuel. Les résultats calculés sur la période de septembre 1999 à août 2000 montrent que les pluies mesurées au sol sont mieux corrélées aux occurrences (r = 0.86) qu’avec les températures maximales (r = - 0.65). Les bons résultats obtenus aux pas de temps mensuels et annuels permettent de réaliser une cartographie des précipitations par satellite qui fait clairement ressortir l’influence de la topographie.
Abstract
Following the national strategy of regional occupation, colonization projects initiated both by the Brazilizan government and private companies, played a major role in the process of deforestation and development of the State of Mato Grosso in the 1970’s. This process involved the expansion of planting crops such as soy-beans, maize and cotton but, because of the lack of meteorological stations before the settlements, many gatherings of crops suffered from the rainfall variability. Concurrently, the low density (one point for 15000km²) and the bad quality (frequently more than 20% of missing values) of the current meteorological network did not allow to draw detailed maps of rainfall. This paper focuses on the interest of GOES-8 images for estimating rainfall with a better spatial accuracy.
The climate in the study area is classified as equatorial or tropical hot and wet. The well-defined dry season lasts from June to August in the northern part of Mato Grosso, while it lasts from April to October in the south. In the same way, the annual average precipitation decreases from 2500mm in the north to less than 1000mm in the Pantanal. We chosed to study the period between September 1999 and august 2000 because of the availability of satellite datas and of rainfall observations for more than 200 stations located between 5°S – 20°S and 50°W – 65°W.
As ground-based techniques such as those using raingauges and radar suffer from spatial and/or temporal coverage problems, the measurement of rainfall from satellites has been a very active field of study. There are two primary means of measuring rainfall from satellite-based sensors. Historically, the first ones were infrared methods which are based on the relationship between cloud top temperatures (observed by geostationnary satellites) and the rainfall intensity. The second ones used datas from passive microwave sensors based on polar-orbiting satellites (DMSP). Although microwave radiances have a good physical connection with rainrate, the low spatial and frequency coverages are a strong limitation for an accurate mesoscale mapping of rainfall. Therefore, we chosed to use infrared images from geostationnary satellites GOES for estimating the monthly rainfall in the state of Mato Grosso. A first dataset was obtained by selecting, for each month, the maximum value of the observed temperature (TBmax). A second one was made counting for each pixel the occurrence of values of temperatures less than –40°C (cold-top cloud occurrence).
The best results were observed for monthly values whereas the correlations for decades values were the lowest ones. For all the months, the results obtained with the occurences of high-level-clouds were better (r = 0.86) than those calculated with the maximum temperature compositing (r = -0.65). This could be related to the high diversity of landscapes in Mato Grosso which leads to a great variability of surface temperatures. The good agreement between the measured rainfall and the cold-top cloud occurrences allowed us to estimate the amount of rainfall for the whole year; this technique permitted an improvement of the knowledge of the spatial distribution of rainfall with a 4km resolution. Satellite datas also confirmed the role of the topography in the amount of rainfall: the main reliefs showed more rainy conditions, phenomenon that cannot be observed with the rainfall ground network.
Mots clés : précipitations / télédétection / Amazonie
Key words: rainfall / remote sensing / Amazonia
© Association internationale de climatologie 2004
This is an Open Access article distributed under the terms of the Creative Commons Attribution License CC-BY-NC (http://creativecommons.org/licenses/by-nc/4.0), which permits unrestricted use, distribution, and reproduction in any medium, except for commercial purposes, provided the original work is properly cited.
Les statistiques affichées correspondent au cumul d'une part des vues des résumés de l'article et d'autre part des vues et téléchargements de l'article plein-texte (PDF, Full-HTML, ePub... selon les formats disponibles) sur la platefome Vision4Press.
Les statistiques sont disponibles avec un délai de 48 à 96 heures et sont mises à jour quotidiennement en semaine.
Le chargement des statistiques peut être long.