Comparación entre siete capas climáticas digitales para identificar áreas con alta precipitación anual en México (al menos 1 500 mm en promedio)
Por:
Rincón-Gutiérrez A., Ricker M., Mas Y.J.-F.
Publicada:
1 ene 2024
Resumen:
Spatial modeling of precipitation is fundamental to understand the distribution of natural vegetation, plant growth, and the consequences of climate change. Here we compare for mainland Mexico the estimates of high annual precipitation (= 1 500 mm on average) from seven digital climate layers, which are available from the Internet. Is it relevant with which layer the areas with high precipitation are identified? Is it advisable to use a precipitation layer elaborated with satellite data, instead of ground-based rain gauges? It turned out that there are huge differences among the estimates of the seven digital layers: The area with average annual precipitation = 1 500 mm varies according to the selected layer between 143 964 km2 (“Cuervo”) and 232 439 km2 (“Satelital”), or 7.4% to 11.9% of Mexico’s land area. The latter area is 1.6 times the former. The total area, where at least one of the seven layers estimates an average annual precipitation = 1 500 mm, is 287 042 km2 (14.7% of Mexico’s land area). The coefficient of variation for the estimated annual precipitation among the seven layers varies from 1.8 to 60.9%. The category with a relatively low coefficient of variation covers 74.7% of the 287 042 km2, the intermediate category 24.1%, and the one with the highest variation 1.2%. All layers, except “Satellite”, are based on data from Mexico’s National Water Commission (CONAGUA). For no layer is it possible to trace which climatological stations were used for its elaboration, and only for some layers the number of used stations is indicated. Other limitations for the analysis of the layers were that we could not find the following information: What statistical corrections were made to the raw data? How were missing precipitation data (which are common in the data from the CONAGUA) handled? What are the precise specifications for the methodology used for spatial interpolation? What environmental factors were taken into account? What parameters of statistical accuracy can be reported? On the other hand, we used 504 climatological stations from CONAGUA, where for the years 1951 to 2010 (or two 30-year periods of climatological normals) an average of = 1 500 mm had been measured. Taking into account the missing data, the 504 climatological stations covered on average only 32.7 years of this 60-year interval. For the data from the climatological stations, we also did not find adequate methodological information: Which measuring instruments were used, and how accurate were they? What is the reason for the large number of missing data? Was there any process of correction and improvement of the data? We analyzed the differences between the precipitation measured at the 504 climatological stations and that estimated by the layer, determining the average (“relative bias”) and the interval between the 10% and 90% quantiles of the differences (“dispersion 1”). The “Satellite” layer had the largest bias (-481 mm) and the largest dispersion (1 707 mm), while “Garcia” had the smallest elative bias (-23 mm) and “UNIATMOS” had the smallest dispersion (781 mm). The averages of the absolute differences (“dispersion 2”), according to the layer from 226 to 630 mm, were also high. The situation of large inconsistencies among layers was similar at 110 525 sites located equidistant between pairs of stations. At these sites, the precipitation estimates of the layers resulted from interpolating data between climatological stations. The extreme was at a site, where one layer estimated 886 mm and the other 4 929 mm of precipitation, 5.6 times the first value. The large variation in precipitation estimated by the layers was considerably explained by the topographic heterogeneity in a circle of 1 km diameter around the 504 stations. As supplementary information, we provide two Excel files with the annual precipitation data from the CONAGUA, and the layer estimates at the station sites. In conclusion, it is relevant which of the seven layers are used to identify areas with high precipitation, and a layer based exclusively on satellite data is not (yet) recommended. The “UNIATMOS” layer stood out here with the comparatively best parameters. In any case, the spatial estimation of high annual precipitation in Mexico still presents major accuracy problems, which results in three recommendations: First, it would be important to improve the scientific basis of empirical data of precipitation and other climate variables for Mexico, with less missing data in the future, with description of methods and quality indicators, and if possible with a larger number of climate stations in regions with high precipitation; second, reproducible methods should be included in the description of climate layers; and third, it would be advisable to model a specific layer for high precipitation for the corresponding surface, estimated here at 14.7% of the Mexico’s surface, where topography is taken into account in detail. © 2024 Instituto de Geografia. All rights reserved.
Filiaciones:
Rincón-Gutiérrez A.:
Posgrado en Ciencias Biológicas, Instituto de Biología, Universidad Nacional Autónoma de México, Ciudad Universitaria, Coyoacán, Ciudad de México, 04510, Mexico
Ricker M.:
Departamento de Botánica, Instituto de Biología, Universidad Nacional Autónoma de México, Ciudad Universitaria, Coyoacán, Ciudad de México, 04510, Mexico
Mas Y.J.-F.:
Laboratorio de Análisis Espacial, Centro de Investigaciones en Geografía Ambiental, Universidad Nacional Autónoma de México, Antigua Carretera a Pátzcuaro 8701, colonia Ex-Hacienda de San José de La Huerta, Morelia, Michoacán, 58190, Mexico
All Open Access; Gold Open Access
|