Abstract

Detailed climate characterization—now essential for adapting to environmental changes and managing natural resources—requires integrated approaches that combine international standardization, high spatial resolution, and regular updates to reference data. The central challenge is updating climate norms in accordance with the recommendations of the World Meteorological Organization (WMO). In Madagascar, the historical zoning system of the General Directorate of Meteorology (DGM), although pragmatic and operational for local stakeholders, is based on an empirical classification (zones A through J) that does not accurately represent a small, distinct area with characteristics of a Mediterranean to temperate summer climate (Csb) or the fine distinction between temperate oceanic (Cfb) and temperate humid (Cfa) climates. This article aims to produce an updated climate map of Madagascar for the reference period 1991-2020 by applying the Köppen-Geiger classification improved by Beck et al. (2018). The method is based on a hierarchical approach that incorporates extreme temperatures and precipitation seasonality, using high-resolution ENACTS data (4 km × 4 km grid). Our results reveal a mosaic of eight active climate types. In the eastern regions (Af tropical rainforest and Am monsoon), abundant rainfall throughout the year requires rigorous management of moisture and runoff. In the west, the savanna climate (Aw) is characterized by severe water stress during the southern winter. In the central highlands, temperate climates (Cfa hot summer and Cfb mild summer) feature regular precipitation, in contrast to the far south, where hot arid climates (BWh) prevail, and the high-altitude Mediterranean climate (Csb) is found locally in the Diana Highlands—an area not previously identified by the DGM.

Keywords

Climate Zones, Climate Classifications, Madagascar, Köppen, Beck

Share and Cite:

1. Introduction

The Köppen-Geiger climate classification is one of the most widely used systems for the global description of Earth’s climates. Based on an empirical approach, it combines temperature, precipitation, and their seasonal variability to define homogeneous climate zones. Originally developed by Köppen (1936) [1], this classification has undergone major revisions, notably by Peel et al. (2007) [2] and later by Beck et al. (2018) [3], who proposed a high-resolution global version based on gridded climate data.

In accordance with the recommendations of the World Meteorological Organization (WMO), climate normals are defined over 30-year periods and are updated every decade (commonly used periods: 1961-1990, 1971-2000, 1981 to 2010, and 1991 to 2020). The period from 1991 to 2020 thus constitutes the current reference norm, and it is essential to update the analyses as soon as each new 10-year period is reached, in order to reflect recent changes in the climate system and provide up-to-date benchmarks for applied studies.

Madagascar, represented by the General Directorate of Meteorology (DGM), is a member country of the WMO. As such, the regular updating of climate information is of particular importance: it provides users (decision-makers, farmers, water resource managers, land-use planners, and researchers) with reference data tailored to their most recent work, and thus operational for planning and managing climate risks.

In this context, this study aims to update Madagascar’s climate mapping by applying the Köppen-Geiger classification, as refined by Beck et al. (2018) over the reference period from 1991 to 2020. This update incorporates essential physical variables such as temperature and precipitation, enabling a detailed and spatially explicit characterization of the island’s climates, in line with current international standards.

2. Methods and Materials

2.1. Climate Data

The data used come from the ENACTS Version 3 (Enhancing National Climate Services) project [4], accessible via Météo Madagascar’s Maproom. They combine in situ observations (weather stations) and satellite estimates to produce a high-resolution 4 km × 4 km raster grid covering the entire territory of Madagascar. The reference period is 1991 to 2020. The available variables are precipitation and temperatures (maximum and minimum).

These data consist of daily values spatially aggregated into raster grids, with each pixel representing a homogeneous geographic cell. This approach makes it possible to map local climates despite the low density of the weather station network in Madagascar.

According to the WMO Guidelines on the Calculation of Climate Normals (WMO-No. 1203, 2017) [5], climate normals are calculated from monthly values, which are themselves derived from daily or hourly observations (Table 1).

Table 1. Method for calculating climatological normals [5].

2.2. Derived Climate Variables

Average temperature $T_{\text{mean}}$

The annual average temperature $T_{\text{mean}}$ for a given period (month, year) is calculated as the arithmetic mean of $T_{\max }$ and $T_{\min }$ :

$T_{\text{mean}}=\frac{T_{\max }+T_{\min }}{2}$ (1)

Next, the annual average temperature $T_{\text{ann}}$ over a series of n years is:

$T_{\text{ann}}=\frac{1}{n}{\displaystyle \sum _{i=1}^n{T}_{\text{mean},i}}$ (2)

where $T_{\text{mean},i}$ is the average temperature for year i. This quantity is used to define major climate groups (tropical, temperate, etc.) [3].

Average Monthly Temperature $T_{\text{mean\_month}}$

$T_{\text{mean\_min\_month}}=\frac{1}{\left|D_m\right|}{\displaystyle \sum _{d\in D_m}{T}_{\min ,d}}$ (3)

$T_{\text{mean\_max\_month}}=\frac{1}{\left|D_m\right|}{\displaystyle \sum _{d\in D_m}{T}_{\max ,d}}$ (4)

$T_{\text{mean\_month}}=\frac{T_{\text{max\_mean}}+T_{\text{min\_mean}}}{2}$ (5)

where:

• $T_{\max ,d}$ and $T_{\min ,d}$ : maximum and minimum daily temperatures on day d (˚C).

• $D_m$ : the set of days in month m over the entire period from 1991 to 2020 (30 years).

For each month $m$ , we first calculate the average of all daily maximum temperatures and the average of all daily minimum temperatures over the 30 year period. The monthly average temperature is the arithmetic mean of these two values.

$T_{\text{mean\_month}}$ is the arithmetic mean of the 12 monthly mean temperatures. Each monthly temperature is first calculated as the average of the daily mean temperatures [3], which are themselves obtained by (T_max + T_min)/2.

Temperature of the warmest month $T_{\max ,\text{warm}}$

$T_{\max ,\text{warm}}={\max }_{m=1,\cdots ,12}{T}_{\text{mean\_month}\left(m\right)}$ (6)

The maximum value among the 12 monthly mean temperatures. It correspond to the mean temperature of the warmest month [3]. Used for the thermal subtypes of temperate climates (a, b, c) and to define polar climates (ET if $T_{\max ,\text{warm}}<10˚\text{C}$ , EF if $T_{\max ,\text{warm}}<0˚\text{C}$ ).

Temperature of the coldest month $T_{min,cold}$

$T_{\min ,\text{cold}}=\underset{m=1,\cdots ,12}{\min }{T}_{\text{mean\_month}\left(m\right)}$ (7)

The minimum value among the 12 monthly average temperatures. It represents the average temperature of the coldest month and is used to distinguish between groups [3]:

• $T_{\min ,\text{cold}}\ge 18˚\text{C}$ → Group A (tropical)

• $3˚\text{C}<T_{\min ,\text{cold}}<18˚\text{C}$ → Group C (temperate)

• $T_{\min ,\text{cold}}\le -3˚\text{C}$ → Group D (continental)

Average monthly precipitation $P_{\text{mean\_month}}$

$P_{\text{mean\_month}}=\frac{1}{N_y}{\displaystyle \sum _{d\in D_m}{P}_d}$ (8)

where:

• $P_d$ : precipitation on day $d$ (mm).

• $D_m$ : set of all days in month m over the entire period from 1991 to 2020 (30 years).

• $N_y=30$ : number of years.

For each month $m$ , we sum all the daily precipitation values for that month over the 30-year period, then divide by the number of years (30). This yields the average precipitation for that month per year (mm/month) [3].

Average Annual Precipitation $P_{\text{ann}}$

Annual precipitation $P_{\text{ann}}$ is calculated as the average of the annual totals $P_{\text{mean\_month}\left(i\right)}$ (mm/year)

$P_{\text{ann}}=\frac{1}{n}{\displaystyle \sum _{i=1}^n{P}_{\text{mean\_month}\left(i\right)}}$ (9)

For each month $m$ , we calculate the sum of all daily precipitation amounts for that month over the 30 year period (1991 to 2020), then divide by the number of years ($N_y=30$ ). This yields the average precipitation for that month per year. Next, $P_{\text{ann}}$ is the sum of the 12 average monthly precipitation values. This variable is essential for calculating the aridity threshold ($P_{\text{threshold}}$ ) and for classifying climates A and B [3].

Precipitation in the driest month ($P_{\text{dry}}$ )

$P_{\text{dry}}=\underset{m=1,\cdots ,12}{\min }{P}_{\text{mean\_month}\left(m\right)}$ (10)

$P_{\text{dry}}$ is the minimum value among the 12 monthly average precipitation values. It correspond to the precipitation in the driest month of the year and is used to distinguish tropical subtypes (Af, Am, Aw) as well as to verify the conditions of temperate climates with dry summers (Cs) [3].

Seasonal precipitation and threshold ($P_{\text{threshold}}$ )

First, we define the totals and averages for the periods April through September and October through March [3]:

$T_{\text{AprSep}}=\frac{1}{6}{\displaystyle \sum }_{m=4}^9{T}_{\text{mean\_month}\left(m\right)}$ (11)

$T_{\text{OctMar}}=\frac{1}{6}{\displaystyle \sum }_{m\in \left\{10,11,12,1,2,3\right\}}{T}_{\text{mean\_month}\left(m\right)}$ (12)

$P_{\text{AprSep}}={\displaystyle \sum }_{m=4}^9{P}_{\text{mean\_month}\left(m\right)}$ (13)

$P_{\text{OctMar}}={\displaystyle \sum }_{m\in \left\{10,11,12,1,2,3\right\}}{P}_{\text{mean\_month}\left(m\right)}$ (14)

The warm season (summer) and the cold season (winter) are determined by comparing temperatures [3]:

$\begin{array}{l}{T}_{\text{AprSep}}>T_{\text{OctMar}}\text{then}\{\begin{array}{l}\text{summer}=\text{April through September}\\ \text{winter}=\text{October through Mars}\end{array}\\ \text{Otherwise}\{\begin{array}{l}\text{summer}=\text{October through Mars}\\ \text{winter}=\text{April through September}\end{array}\end{array}$ (15)

On calcule alors [3]:

$P_{\text{summer},\text{tot}}={\displaystyle \sum }_{m\in \text{summer}}{P}_{\text{mean\_month}\left(m\right)}$ (16)

$P_{\text{winter},\text{tot}}={\displaystyle \sum }_{m\in \text{winter}}{P}_{\text{mean\_month}\left(m\right)}$ (17)

$P_{\text{smin}}=\underset{m\in \text{summer}}{\text{min}}{P}_{\text{mean\_month}\left(m\right)}$ (18)

$P_{\text{wmin}}=\underset{m\in \text{winter}}{\text{min}}{P}_{\text{mean\_month}\left(m\right)}$ (19)

$P_{\text{smax}}=\underset{m\in \text{summer}}{\text{max}}{P}_{\text{mean\_month}\left(m\right)}$ (20)

$P_{\text{wmax}}=\underset{m\in \text{winter}}{\text{max}}{P}_{\text{mean\_month}\left(m\right)}$ (21)

$pc{t}_{\text{winter}}=\frac{P_{\text{winter},\text{tot}}}{P_{\text{ann}}}\times 100$ (22)

$pc{t}_{\text{summer}}=\frac{P_{\text{summer},\text{tot}}}{P_{\text{ann}}}\times 100$ (23)

The precipitation threshold for arid climates is [3]:

$P_{\text{threshold}}=\{\begin{array}{ll}2T_{\text{mean\_ann}}\hfill & \text{if}pc{t}_{\text{winter}}>70\hfill \\ 2T_{\text{mean\_ann}}+28\hfill & \text{if}pc{t}_{\text{summer}}>70\hfill \\ 2T_{\text{mean\_ann}}+14\hfill & \text{otherwise}\hfill \end{array}$ (24)

This threshold depends on the annual mean temperature and the seasonality of precipitation.

Beck et al. (2018) use equation (24) to calculate this threshold:

• If more than 70% of annual precipitation falls in winter: $P_{\text{threshold}}=2\times T_{\text{mean\_ann}}$

• If more than 70% of annual precipitation falls in summer: $P_{\text{threshold}}=2\times T_{\text{mean\_ann}}+28$

• In other cases (uniform distribution): $P_{\text{threshold}}=2\times T_{\text{mean\_ann}}+14$

where $T_{\text{mean\_ann}}$ is the annual mean temperature (˚C). Summer (winter) is defined as the warmest (coldest) six-month period between April-September and october-march.

Once this threshold is calculated, the criteria for arid climates are [3]:

• Arid climate (B): $P_{\text{ann}}<10\times P_{\text{threshold}}$

• Desert (BW): $P_{\text{ann}}<5\times P_{\text{threshold}}$

• Steppe (BS): $P_{\text{ann}}\ge 5\times P_{\text{threshold}}$ (and $P_{\text{ann}}<10\times P_{\text{threshold}}$ )

Number of warm months (num_months_gt_10)

$\text{num\_months\_gt\_}10=N_{>10}={\displaystyle \sum _{m=1}^{12}1\text{l}\left({\overline{T}}_m>10˚\text{C}\right)}$ (25)

where $1\text{l}(\cdot )$ is the indicator function.

Counts the number of months in which the average monthly temperature exceeds 10˚C. Used to distinguish the thermal subtypes of temperate climates [3]:

• $N_{>10}\ge 4$ : subtype b (temperate summer)

• $N_{>10}<4$ : subtype c (cool summer)

2.3. Climate Classification (Beck et al., 2018)

The classification is hierarchical: it first identifies the major climate group (A, B, C, D, E) based on the temperature of the coldest month ($T_{\text{coldest}}$ ) and the temperature of the warmest month ($T_{\text{warmest}}$ ). Next, each group is subdivided into subtypes based on the seasonality of precipitation, summer temperature, and the number of months with an average temperature exceeding 10˚C. Table 2 summarizes the hierarchical structure of all groups and subtypes.

Table 2. Hierarchical structure of groups and subtypes in the Köppen-Geiger classification (adapted from Beck et al., 2018).

Definitions of parameters and legends:

• $T_{\text{ann}}$ : annual average temperature

• $T_{\text{max\_warm}}$ : maximum temperature of the warmest month

• $T_{\text{min\_cold}}$ : minimum temperature of the coldest month

• $P_{\text{ann}}$ : total annual precipitation

• $P_{\text{dry}}$ : precipitation of the driest month

• $P_{\text{threshold}}$ : Köppen precipitation threshold (a function of $T_{\text{ann}}$ and seasonality)

• $P_{\text{smin}}/P_{\text{wmin}}$ : minimum summer/winter precipitation

• $P_{\text{smax}}/P_{\text{wmax}}$ : maximum summer/winter precipitation

• num_months_gt_10: number of months with $T_{\text{moy}}$ > 10˚C

• Cs (dry summer): $P_{\text{smin}}<P_{\text{wmin}}$ , $P_{\text{wmax}}>3\times P_{\text{smin}}$ , $P_{\text{smin}}$ < 40 mm

• Cw (dry winter): $P_{\text{wmin}}<P_{\text{smin}}$ , $P_{\text{smax}}>10\times P_{\text{wmin}}$

• !is_Cs: neither Cs

• !is_Cw: neither Cw

• Cf (no dry season): !is_Cs & !is_Cw

2.4. Spatial Classification Method

The classification is implemented as a raster: each pixel with coordinates (x, y) is assigned a climate class based on a hierarchical function [Beck et al., 2018]:

$\begin{array}{l}\text{Class}\left(x,y\right)=f(T_{\text{ann}},T_{\text{max\_warm}},T_{\text{min\_cold}},P_{\text{ann}},P_{\text{dry}},P_{\text{threshold}},P_{\text{smin}},P_{\text{wmin}},\\ P_{\text{smax}},P_{\text{wmax}},\text{num}\_\text{months}\_\text{gt}\_10)\end{array}$

The classification follows a multi-step decision-making process [Beck et al., 2018]:

Step 1: Identification of polar climates (Group E)

If $T_{\text{max\_warm}}$ < 0˚C: EF climate (ice cap)

If 0 ≤ $T_{\text{max\_warm}}$ < 10˚C: ET climate (tundra)

Step 2: Identification of arid climates (group B)

The Pthreshold value is first calculated based on the seasonal distribution of precipitation (Equation 24):

• If winter precipitation accounts for more than 70% of the annual total: $P_{\text{threshold}}=2\times T_{\text{ann}}$

• If summer precipitation accounts for more than 70% of the annual total: $P_{\text{threshold}}=2\times T_{\text{ann}}+28$

• Otherwise: $P_{\text{threshold}}=2\times T_{\text{ann}}+14$

Then:

• If $P_{\text{ann}}<5\times P_{\text{threshold}}$ : desert climate

• BWh (hot desert) if $T_{\text{ann}}$ ≥ 18˚C

• BWk (cold desert) if $T_{\text{ann}}$ < 18˚C

• If $5\times P_{\text{threshold}}\le P_{\text{ann}}<10\times P_{\text{threshold}}$ : steppe climate

• BSh (hot steppe) if $T_{\text{ann}}$ ≥ 18˚C

• BSk (cold steppe) if $T_{\text{ann}}$ < 18˚C

Step 3: Determining the primary group (A, C, D)

For non-polar and non-arid regions:

• Groupe A (Tropical): if $T_{\text{min\_cold}}$ ≥ 18˚C

• Groupe C (Temperate): if −3˚C < $T_{\text{min\_cold}}$ < 18˚C

• Groupe D (Continental): if $T_{\text{min\_cold}}$ ≤ −3˚C

Step 4: Assigning sub-types

For Group A (Tropical):

• Af (tropical rainforest): if $P_{\text{dry}}$ ≥ 60 mm

• Am (tropical monsoon): if $P_{\text{dry}}$ < 60 mm et $P_{\text{ann}}\ge 25\times \left(100-P_{\text{dry}}\right)$

• Aw (tropical savanna): if $P_{\text{dry}}$ < 60 et $P_{\text{ann}}<25\times \left(100-P_{\text{dry}}\right)$

For Group C (Temperate):

First, we define the seasonality indicators:

• is_Cs = $P_{\text{smin}}<P_{\text{wmin}}$ , $P_{\text{wmax}}>3\times P_{\text{smin}}$ , $P_{\text{smin}}$ < 40 mm

• is_Cw = $P_{\text{wmin}}<P_{\text{smin}}$ , $P_{\text{smax}}>10\times P_{\text{wmin}}$

• is_Cf = !is_Cs & !is_Cw

Then:

• Csa: is_Cs and $T_{\text{max\_warm}}$ ≥ 22˚C

• Csb: is_Cs, $T_{\text{max\_warm}}$ < 22˚C and num_months_gt_10 ≥ 4

• Csc: is_Cs (default rule)

• Cwa: is_Cw and $T_{\text{max\_warm}}$ ≥ 22˚C

• Cwb: is_Cw, $T_{\text{max\_warm}}$ < 22˚C and num_months_gt_10 ≥ 4

• Cwc: is_Cw (default rule)

• Cfa: is_Cf and $T_{\text{max\_warm}}$ ≥ 22˚C

• Cfb: is_Cf, $T_{\text{max\_warm}}$ < 22˚C and num_months_gt_10 ≥ 4

• Cfc: is_Cf (default rule)

For Group D (Continental):

• Dfa: $T_{\text{max\_warm}}$ ≥ 22˚C

• Dfb: $T_{\text{max\_warm}}$ < 22˚C and num_months_gt_10 ≥ 4

• Dfc: $T_{\text{max\_warm}}$ < 22˚C and num_months_gt_10 < 4 (default rule)

Each pixel is thus classified independently, allowing for the production of a continuous and spatially explicit climate map.

3. Results

Figure 1. Climate classification of Madagascar (1991-2020).

Table 3. Climate classes in Madagascar (1991-2020).

Madagascar, located between 12˚S and 26˚S, has a wide variety of climates due to its mountainous terrain (a north-south-oriented central mountain range, with Peak Boby reaching over 2800 m) and the influence of maritime trade winds. The map, generated using ENACTS data (4-km grid) and the classification by Beck et al. (2018), highlights a spatial distribution organized along an east-west gradient and a latitudinal gradient (see Figure 1).

Applying the classification method developed by Beck et al. (2018) to ENACTS data (1991-2020) across the entire territory of Madagascar reveals a range of eight distinct climate classes. This diversity reflects the well-known contrast between the east and west: a perpetually humid eastern slope influenced by maritime trade winds (classes Af and Am), a significantly drier western and southern coastline (classes BWh, BSh, and Aw), and the central highlands, which enjoy a temperate highland climate with distinct thermal seasons (classes Cfa and Cfb). Table 3 details, for each climate class present in Madagascar, the quantitative criteria (indices, temperature and precipitation thresholds) as well as a description of its local climatic characteristics.

Schematic spatial organization (map reading)

Based on the known general characteristics of Madagascar’s climate, the following patterns are observed:

• East Coast (from approximately 13˚S to 24˚S): dominated by Af, Cfb, and Am climate zones (tropical rainforest, temperate oceanic, and monsoon). Moist oceanic trade winds discharge their moisture over the eastern slope, generating very high annual precipitation (2000 to 3500 mm). The dry season is short or nonexistent.

• BSh (hot steppe) in southwestern Madagascar; Cfa (humid temperate) in the northeastern, central, and southern parts; Cfb (oceanic temperate) in the central-eastern part of the island.

• Northern, northwestern, central, and south-central regions (along the entire length, from 15˚S to 20˚S): Aw climate (tropical savanna with a dry winter).

• Far south and southwest (Cape Androka, Tulear, from 21˚S to 25˚S): transition to BWh (hot desert) in the southwestern part of the Atsimo-Andrefana region and in the southern part of the Anosy region. Annual rainfall is less than 400 mm (sometimes less than 200 mm along the coast in the southwestern part).

4. Discussion

4.1. Geographical Correspondences between the DGM Climate Zones (1981-2010) and the Köppen-Geiger Classes Refined by Beck et al. (1991-2020)

Based on the DGM’s description of climate zones [6] (see Figure 2 in the APPENDIX) and the implied locations of the regions of Madagascar, the following correspondences can be established (based on knowledge of Madagascar’s climate) (See Table 4).

Table 4. Proposed correspondence between the DGM climate typology and the Köppen-Geiger classification, as updated by Beck et al. (2018), for Madagascar.

4.2. Divergences Notable Differences: Missing or Additional Classes

Table 5 compares the DGM system with the refined classification by Beck et al. (2018), highlighting the shortcomings of the former: the absence of the Mediterranean temperate Csb zone, which is found locally in the Diana highlands, and the grouping of the humid temperate Cfa and Cfb zones into a single zone, even though summer temperatures vary significantly with elevation.

Table 5. Comparative analysis of the limitations of the DGM climate zoning in relation to the refined Köppen-Geiger classification of Beck et al. (2018).

4.3. Factors Explaining Discrepancies

Spatial resolution and interpolation method:

• New climate classification (method by [3]): ENACTS data with a resolution of 4 km, which is effective at capturing subtle variations in topography (rain shadow effects, thermal inversion). This allows for the distinction between Cfa and Cfb and the identification of Csb in high-mountain areas.

• DGM: ENACTS data with a resolution of 10 km (for precipitation and temperatures), which is less precise. However, the DGM accounts for coastal breezes and local exposure (data on winds, humidity, and duration of sunshine), which gives it good robustness in climate class classification for areas in southern Madagascar.

Different reference periods:

• New climate classification (method by [3]): 1991 to 2020 (recent period, including the years 2010 to 2020, which were marked by severe droughts in the South).

• DGM: 1981 to 2010 for temperatures and winds, 1983 to 2010 for precipitation, 1982 to 2010 for relative humidity, and 2000 to 2010 for sunshine duration. These different periods are older and wetter in the South (the 1980s and 1990s were less arid). Thus, the semi-arid zone J of the DGM is likely more limited in extent than Beck’s BWh classification, which would explain why Beck classifies part of the Southwest as BWh.

5. Conclusions

Applying the Köppen-Geiger climate classification, as refined by Beck et al. (2018) [3], to ENACTS data for the period 1991-2020 reveals a complex and contrasting climatic mosaic in Madagascar, comprising eight active climate types, ranging from tropical rainforests (Af) to hot deserts (BWh). The resulting map highlights a clear east-west contrast: the eastern regions, influenced by maritime trade winds, are dominated by Af (tropical rainforest) and Am (monsoon) climates, while the western plains are largely characterized by the Aw savanna climate. The central highlands feature temperate climates (Cfa and Cfb) that are unique in the tropics, with their mild temperatures directly linked to elevation. Finally, the extreme southwest is home to hot desert (BWh) and hot steppe (BSh) climates. This mapping serves as a valuable tool for agronomy, biogeography, and water resource planning, providing a spatially explicit view of the climatic constraints and advantages of each region.

A comparison between the standardized classification by Beck et al. (2018) [3] and the historical zoning by the General Directorate of Meteorology (DGM) shows that these two approaches are more complementary than contradictory. On the one hand, Beck’s map provides a detailed, international, and up-to-date view (covering the period from 1991 to 2020), highlighting a Mediterranean zone with a temperate summer (Csb) and a clear distinction between the humid temperate climates of the highlands (Cfa vs. Cfb). It is particularly recommended for basic research in ecology, biodiversity, and regional climate modeling. On the other hand, the DGM zoning, based on an empirical grid of ten zones (A through J), is more pragmatic and operational for local stakeholders (Ministry of Agriculture, water management, land-use planning). Its zones are directly correlated with ranges of precipitation and temperature.

This comparison illustrates—beyond specific cases—the crucial importance of two methodological parameters in any climate characterization: spatial resolution and the reference period. The same territory can, in fact, appear more or less arid or more or less temperate, depending on the tools used and the time frame considered. Ideally, we should aim for a hybrid classification that combines the rigor and universality of the Köppen-Geiger thresholds with the clarity and operational simplicity of the DGM zones. Such an integrated approach would reconcile the demands of international research with the practical needs of development stakeholders in Madagascar, thereby contributing to greater resilience in the face of climate change.

Acknowledgements

The authors gratefully acknowledge the “Direction Générale de la Météorologie” of Madagascar for providing the daily temperature and precipitation data used in this study.

Appendix

Climatic classification of Madagascar by the “Direction Générale de la Météo-rologie”:

Figure 2. Climate zones of Madagascar (1981-2010) [6].

Conflicts of Interest

The authors declare no conflicts of interest.

References