Análisis comparativo de las zonificaciones climáticas de Caldas-Lang y Holdridge, con la zonificación del clima edáfico del estudio semidetallado de suelos, en la cuenca del río Cauca, departamento del Valle del Cauca

Abstract

COMPARATIVE ANALYSIS OF THE CLIMATIC ZONES OF CALDAS-LANG AND HOLDRIDGE, WITH THE ZONING OF THE SOIL CLIMATE OF THE SEMIDETALLED SOILS STUDY, IN THE CAUCA RIVER BASIN, VALLE DEL CAUCA DEPARTMENT – COLOMBIA

INTRODUCTION In Colombia climatic zoning has been carried out based mainly on the classifications proposed by Köppen, Martonne (1937), Thornthwaite, Caldas-Lang (1962) and Holdridge (IDEAM, 2005). The IDEAM and the Alexander von Humboldt Biological Resources Research Institute (IAvH) have generated climatic maps of the country and its different natural regions using the Holdridge and Caldas-Lang classifications (Espinal & Montenegro, 1963, Etter, 1998, Rodríguez et al. ., 2004). The Geographical Institute Agustín Codazzi (IGAC, 2018) for its soil studies uses a combination of the Caldas and Holdridge classifications. Despite the numerous works on climate zoning carried out in the country, very few have addressed comparative analysis between different classification methods for the same area or region, which allow deciding on which is the most appropriate or convenient to use.

The IGAC (2004), IGAC-CENICAÑA (2005) and IGAC-CVC (2016) carried out semi-detailed studies of soils at a scale of 1: 25,000 in the Cauca river basin, department of Valle del Cauca - Colombia. Studies that have allowed to know the origin, evolution and distribution of the soils in the region and provided a basic tool for the processes of management and management of watersheds. In addition, they have zoning of the edaphic climate generated from the analysis of the temperature and humidity regimes associated with the formation processes of the different soil units. The present study compared two climate classification methods commonly used in river basin management processes in the department of Valle del Cauca: Caldas-Lang and Holdridge. For this purpose, data series of meteorological stations located in the Cauca River hydrographic area were analyzed: 174 for precipitation and 23 for temperature (series of minimum 30 years). Several interpolation methods were tested, and using GIS tools, the temperature and precipitation surface models were generated, and then the climate zoning with the classifications of interest. The quality of the interpolation methods was analyzed by the Determination Coefficient (Adjusted R2), Mean Square Error (MSE), Cross Validation (CV) and the Akaike Information Criterion (AIC). Finally, the climatic zonings in mention were contrasted with the zoning of the edaphic climate of the semi-detailed study of soils for the study area. The statistical comparison parameters of the climatic zonifications allowed to deduce that Holdridge was the method of best fit to the data used. On the other hand, the contrast of the climatic zonifications with the zoning of the edaphic climate showed similarity of the latter with Caldas-Lang at the level of thermal floors. At the moisture province level, the analysis indicated that the comparative zoning did not resemble each other. The zoning of the soil climate provides valuable information on climatic conditions associated with the formation of soils that can complement the analysis of the climatic characterizations of watersheds.

METHODOLOGY Study Area: Cauca river basin in the department of Valle del Cauca (Colombia), 10,738 km2 (900 to 4,200 MAMSL). High climatic and physical variability (Molina Benavides et al., 2016) (figure 1).

Data used: network of meteorological stations of the IDEAM and the CVC (23 temperature and 174 precipitation) (figure 2).

Treatment of climate data: an interest period of 30 years (1980-2010) was considered, as defined by the World Meteorological Organization (WMO, 2015).

Quality control: through the methods of coarse filters, standard deviations, interquartile range, consecutive, duplicates and inter-monthly difference..

Filling missing data: temperature using the R Climatol package (Guijarro, 2014) and precipitation with the CHIRPS tool (Funk et al., 2015).

Generation of climatic surfaces: use of the R programs (statistics) with their interfaces, ArcGIS 10.5 (SIG) and the digital elevation model (DEM) SRTM (Shuttle Radar Topography Mission) of 30 m.

Temperature: from the Regnie model (Rauthe et al., 2013), which combines multiple linear regression and the Inverse Distance Weighting (IDW) (Alzate et al., 2017), with which the three (3) equations were obtained.

Precipitation: Natural Neighbor interpolation methods (Sibson, 1980), Inverse Distance Weighting (IDW) (Philip & Watson, 1982) and Kriging (Kirge, 1951) were tested, verified by cross-validation and calculation of mean square error (MSE) . Edaphic climate zoning: product of the semi-detailed studies of soils at 1: 25,000 scale carried out by the IGAC (2004), IGAC-CENICAÑA (2005) and IGAC-CVC (2016), based on field work with 845 profile and 6,012 scuttle and boxes (figure 3).

Comparative analysis Caldas-Lang versus Holdridge: Akaike Information Criterion (AIC) (Martínez, 2009).

Climatic zoning versus zoning of soil climate: Friedman test and the Wilcoxon test (Gregory and Foreman, 2014). From a point matrix (2,675) of 2 km between points.

RESULTS

Surface models (temperature and precipitation)

Temperature: Three equations were obtained with the Regnie model, of which the one that considered the coordinates and the altitude, showed to be the most adequate to generate the surface model according to the criteria used. According to the model obtained, the temperature varies between 4 ° C in the highest zone up to 25 ° C in the lower part with a predominance of values higher than 18 ° C.

Precipitation: The model obtained with the Natural Neighbor interpolation method obtained the best values in the Mean Square Error (MSE) parameters and the calculation of the Prediction Effectiveness Estimate (E). On the western slopes of the Cordillera Central, the averages are slightly above 1,500 mm and, above 3,500 meters above sea level, rainfall begins to decrease to around 1,000 mm or less, as in the Barragán páramo, Bugalagrande river basin (North East).

Climatic zoning Caldas-Lang: 15 climatic units were obtained; with predominance of the semi-humid Tempered (26%) and Warm Semi-Arid (19%) climatic units. The Arid Warm unit (3%) is registered in the south central area of the geographical valley.

Holdridge: 13 climatic units were obtained, with predominance of the Humid Premontane unit (36%), followed by the Dry Premontane (8.3%). The Montano dry bass climatic unit, which occupied 0.1% of the area, is observed as a small sector.

Edaphic climate zoning: includes 12 climatic units, where the most representative unit was Warm Dry (31%), followed by Humid Tempered (26%). The following figure shows the result of the three zoning.

Climatic zoning Holdridge, Caldas-Lang vs. zoning climate For the Friedman and Wilcoxon tests at the level of thermal floors and humidity provinces, it was hypothesized that the three classified in the same way versus at least one of them differed in their classification. From the 2,675 points taken on each of the zoning, the evaluation was made separately, that is, the obtained in the thermal floor and the humidity province were compared. Thermal floors: the three zonifications showed similar proportions in the thermal floors, being notable the differences in the Warm and Tempered categories.

The Friedman test indicated that there are differences in at least one of the zoning, since the p-value obtained was below an alpha of 5%.

To identify which zoning was different from the others, the Wilcoxon peer test was done.

There were similarities between the zoning of Soils (IGAC) and Caldas-Lang (0.66), while Holdridge showed no similarities with the others.

Moisture provinces: The percentage distribution of the humidity provinces of the compared zoning, in general terms, was similar to that observed in the thermal floors. Caldas-Lang represented a different (Very Dry), which coincided with the sites of less precipitation in the study area (average annual precipitation values less than 800 mm / year). The following figure shows the results of the zoning by province of humidity.

The Friedman test indicated that there are differences in at least one of the zoning, since the p-value obtained was below an alpha of 5%.

With the Wilcoxon test, it is obtained that all the zonifications are different, that is, there were not enough coincidences to determine that they were giving the same result.

This allows us to infer that no pair of zonings classifies significantly equal to the other, each one has different results

DISCUSSION The parameters of statistical comparison of the climatic zonifications allowed to deduce that the method of classification of Holdridge presents a better statistical adjustment than Caldas-Lang. Although this last zoning represented a greater number of climatic units, it did not mean that this aspect gave it an advantage, since the applied statistical tests selected the classification method that lost the least information or better adjustment to the data used.

Given that the two climatic zonifications presented the same number of homologated thermal floors, the difference between the number of climatic units obtained obeyed the greater number of provinces of humidity that Caldas-Lang had. The limits of the climate classes used by Caldas-Lang were narrower or closer than those used by Holdridge. That is, the Lang Factor had more subdivisions or categories than the types of Potential Evapotranspiration proposed by Holdridge. The contrast of the climatic zonifications with the zoning of the edaphic climate showed similarity of the latter with Caldas-Lang at the level of thermal floors. The comparison at the level of provinces of humidity showed no similarity between zoning. In terms of percentage distribution of thermal floors and humidity provinces, the comparative zoning was similar. There was a coincidence in the number of thermal floors, with notable differences in the Warm and Tempered categories. The spatial distribution was similar in the Humid, Humid and Very Humid homologated provinces. Likewise, there was little coincidence in the spatial distribution of the Pluvial humidity province in the three climatic zonifications. The Caldas-Lang zoning is the only one that represented the province of Very Dry humidity, which coincided with the sites of least precipitation in the study area (average annual precipitation values less than 800 mm / year). For the specific case of the upper area of the Bugalagrande river basin (Barragán), while the climatic zones of Holdridge and Caldas-Lang, are based on one (1) temperature and precipitation station, the zoning of the soil climate used more than 170 field observations (22 profiles and about 150 scuttle and boxes), in addition to the typical vegetation of the very dry area as shown by the following images

The zoning of the soil climate in semi-detailed studies at 1: 25,000 scale provides valuable information on climatic conditions associated with the formation of soils that can complement the analysis of the climatic characterizations of watersheds, as well as a better evaluation of the capacity for use and management of the floors. In this sense, greater importance should be given to the semi-detailed soil studies required by the MADS technical guide for the preparation of the POMCA, considering that they are the product of important field observations and comprehensive laboratory and office analyzes.

CONCLUSIONS The parameters of statistical comparison of the climatic zonifications allowed to deduce that the method of classification of Holdridge was better than Caldas-Lang. Although this last zoning represented a greater number of climatic units, it did not mean that this aspect gave it an advantage, since the applied statistical tests selected the classification method that lost the least information or better adjustment to the data used.

Given that the two climatic zonifications presented the same number of thermal floors, the difference between the number of climatic units obtained obeyed the greater number of provinces of humidity that the Caldas-Lang model obtained. The limits of the climate classes used by the Caldas-Lang model were narrower or closer than those used by the Holdridge model. That is, the Lang factor had more subdivisions or categories than Holdridge's Potential Evapotranspiration types.

The contrast of the climatic zonifications with the zoning of the edaphic climate showed similarity of the latter with the Caldas-Lang model at the level of thermal floors. The comparison at the level of provinces of humidity showed no similarity between zoning. In terms of percentage distribution of thermal floors and humidity provinces, the results of the compared zoning were similar. There was a coincidence in the number of thermal floors, with notable differences in the Warm and Tempered categories. The spatial distribution was similar in the provinces of dry, humid and very humid humidity. Likewise, there was little coincidence in the spatial distribution of the Pluvial humidity province for the three climatic zonifications.

The Caldas-Lang zoning is the only one that represented the province of Very Dry humidity, which coincided with the sites of least precipitation in the study area (average annual precipitation values less than 800 mm / year).

The zoning of the soil climate in semi-detailed studies at 1: 25,000 scale provides valuable information on climatic conditions associated with the formation of soils that can complement the analysis of the watershed climate characterizations, as well as favoring a better evaluation of the capacity for use and soil management.

The coverage of climatic stations is fundamental to obtain a good approximation with the proposed models, as can be demonstrated in the c located northeast of the study area where the results were compared zoning at the humidity province level. The Figure 6 shows that the zoning of Holdridge, Caldas-Lang and Soils yielded two, three and four categories of humidity provinces respectively, the latter being the most forceful when compared with the reality of the area.