Analysis of the diurnal cycles for a better understanding of the mean annual cycle of forests greenness in Central Africa

doi:10.1016/j.agrformet.2016.04.005

Agricultural and Forest Meteorology

Volume 223, 15 June 2016, Pages 81-94

https://doi.org/10.1016/j.agrformet.2016.04.005 Get rights and content

Highlights

•
The two dry seasons and two rainy seasons are unique in terms of climate and EVI.
•
EVI annual cycle in Central Africa is driven by diurnal and annual cycles in climate.
•
Water and light influence on forest greenness changes during the annual cycle.
•
Water availability increase determines the March-May vegetative season start.
•
High light availability explains EVI rapid increase and high level in March-April.

Abstract

Central Africa hosts the world's second largest tropical forest after the Amazonian basin. However, as compared to its Amazonian counterpart, the Central Africa forests receive much less rain (∼1500 mm/year in mean spread over two rainy seasons in March-May and September-November, and two dry seasons). They also experience a slower deforestation rate, so that the main threat for the next decades might come from climate variations. Nonetheless, their response to the annual cycle of solar radiation and rainfall/clouds is still poorly known. Analysing high resolution remote sensing data of Enhanced Vegetation Index, rainfall, cloudiness, and solar radiation for a target region located between 0 and 5°N and 12–19°E, we explore the climatic drivers of the forests greenness mean annual cycle. Three main points emerge; first, the diurnal cycle is a key-scale for understanding the mean annual cycles of rainfall and incoming solar radiation at surface, then how climate shapes the greenness mean annual evolution; second, neither the two dry seasons nor the two rainy seasons resemble each other in terms of cloud cover, solar radiation and rainfall, and their links with greenness levels; third, whereas the first rainy season (March-May) appears optimal for greenness especially because of favorable light conditions, water availability is the main controlling factor during the main dry season and at the start of the first vegetative season (February). Regarding the little dry season (mid-June-mid-August) and the second rainy season (September-October), light availability might be the main limiting factor. These findings pave the way for further studies of the climate interannual variability and change impacts on the Central Africa forests, taking into account time-scale interactions.

Introduction

Tropical forests are a key component of the global climate system as they act as water vapor sources (Spracklen et al., 2012) and carbon dioxide sinks (Lewis et al., 2009) and also correspond to the main ascending sectors of the thermal direct circulations across the tropical zone. For these reasons their evolution in response to both human pressure and climate change is critical. Most of the previous studies devoted to tropical forests' sensitivity to climate focused on the Amazonian forests and suggested that light availability is the main driver of their photosynthetic activity and leaf area seasonal and interannual variations (Huete et al., 2006, Myneni et al., 2007). Although studies pointed to the artificial nature of these variations – dry season greening could result from an artificial increase of canopy reflectance at near-infrared wavelengths caused by variations in the sun-satellite sensor geometry (Morton et al., 2014, Samanta et al., 2012) – more recent studies confirm that the seasonality observed is not an artefact of changes in sun position in the sky or how sensors measure the reflected radiation field (Bi et al., 2015). This call for the use of better corrected vegetation products such as those issued from the MAIAC algorithm (Hilker et al., 2012, Hilker et al., 2015).

The Central African forests which extend over Cameroon, Gabon, Guinea, and the Central African, Congo and Democratic Congo Republics, display a greenness mean annual cycle very different from the Amazonian forests: it is bimodal (vs unimodal), and seems controlled first by the rainfall mean annual cycle (Gond et al., 2013). At the interannual time-scale, the control of the Central African forests' greenness variability by rainfall has also been recently highlighted by Zhou et al. (2014) for the April-June season: a significant greenness decrease is synchronous to a rainfall decrease. However, looking both at greenness and rainfall mean annual cycles shapes (Gond et al., 2013) and at correlations between the interannual variations of greenness and rainfall (Zhou et al., 2014), there are indications that rainfall is not the sole determinant of forests' greenness in Central Africa.

As a first step towards a better understanding of the sensitivity of the Central African forests to present-day climate variability, and a better evaluation of its response to climate change, a detailed analysis of its greenness mean annual cycle is performed. We also examine those of several climate variables considered as potential drivers, i.e., rainfall, cloudiness and solar radiation. Indeed it is still unclear (1) what is the respective weight of these parameters on the forests' greenness mean annual cycle in Central Africa and (2) how these parameters relate to each other, notably because neither the cloudiness nor the solar radiation mean annual cycles are sufficiently known over the region.

We explore these questions focusing on the area of Central Africa located between 0 and 5°N and 12–19°E (Fig. 1, red square) which encompasses the Southeastern Cameroon, the Southwestern Central African Republic, the Northern Congo Republic and the Northwestern Congo Democratic Republic for which 11 types of forest, and their respective seasonal profiles of Enhanced Vegetation Index (EVI) have been precisely determined by Gond et al. (2013). In-situ data over Central Africa and our study area in particular are scarce. This is particularly true for rainfall, temperature, and solar radiation while in-situ measurements of evapotranspiration, photosynthetic activity, and leaf area do not exist here. Therefore, high resolution remote sensed products are exclusively used and presented in Section 2. Section 3 briefly depicts the methods used. Results are provided in Section 4: after presenting and comparing the mean annual cycles of EVI and the climate parameters considered, the focus is on the diurnal cycles of cloud cover and solar radiation. Then regressions and residual analyses are performed to disentangle the respective roles of rainfall and light on forests' greenness. Section 5 discusses the results and closes the paper.

Section snippets

Terra-firme forest types and their corresponding Enhanced Vegetation Index annual cycles

We used the Enhanced Vegetation Index (EVI) from the TERRA-MODIS ‘16-Day L3 Global 500 m product (MOD13A1 c5)’ from January 2000 to December 2009 which is commonly used for monitoring the spatial and temporal variations of tropical forests (Xiao et al., 2006). Being more responsive to canopy structural variations, EVI provides improved sensitivity for high biomass vegetation compared to NDVI which is more chlorophyll-sensitive thus rapidly saturates for tropical forests (Huete et al., 2002).

Computation of the regional indexes of EVI, rainfall, cloudiness and solar radiation for the forests, and of minimum and maximum temperature

Rainfall, cloudiness, and solar radiation levels were first determined for each of the 11 types of terra-firme forests present in our study area. This was done into four steps. First, climate data were linearly interpolated on the forest map grid (which is at 500 m resolution). Second, the non terra-firme forest pixels were masked. Third, the climate values of pixels falling into the same forest type were averaged to obtain one climate index per parameter (rainfall, cloudiness and solar

The mean annual cycles of EVI, rainfall, total cloudiness, solar radiation and temperature

In agreement with Gond et al. (2013), the mean annual cycle of forests' greenness as pictured by EVI over our study region is bimodal (EVI, Fig. 2a, average of the 11 forest types). As compared to EVI dynamics observed for the Amazonian forests, the amplitude is smaller (∼0.15 vs ∼0.25) and levels are lower (∼0.51 vs ∼0.63, Huete et al., 2002, Huete et al., 2006). Both seasons of higher greenness (i.e., EVI values are above the annual mean, i.e., ∼0.45) are from beginning of March to mid-June

Discussion

Our study's goal is to better understand, for a 5°lat × 7°lon test-region in Central Africa, the forests' photosynthetic activity mean annual cycle and its drivers. Because in-situ eco-climatic data for the region are scarce or inexistent, our study relied on the analysis of remote sensing data of EVI, rainfall, cloudiness and solar radiation. For temperature we worked with in-situ measurements at three stations located in our study area. These data cover different periods, and space and time

Conclusion

In the absence of in-situ measurements, this study's aim was to analyse high resolution (in space and time) remote sensing products documenting key climatic parameters – namely cloudiness, solar radiation and rainfall – to understand the mean annual cycle of forests' greenness for a test-region in Central Africa spreading across Southeastern Cameroon, the Southwestern Centrafrica Republic, the Northern Congo Republic and the Northwestern Congo Democratic Republic. Several key-points emerge from

Acknowledgements

Calculations were performed using HPC resources from DSI-CCUB (Université de Bourgogne, Dijon, FRANCE) and the Matlab® software with the statistical toolbox built-in functions. The authors acknowledge the Centre National d'Etudes Spatiales (CNES) for its financial support (VEGREENE project), ICARE Data and Service Center (Lille, France) for giving access to the 2010–2014 MSG Cloud Analysis product archives. They are thankful to Claire Thomas, TRANSVALOR, for her support with regards to the

References (43)

E. Aguilar
Changes in temperature and precipitation extremes in western central Africa, Guinea Conakry, and Zimbabwe, 1955–2006
J. Geophys. Res.
(2009)
J. Betbeder et al.
Landscape wetland mapping using multi sensors approach
J. Sel. Top. Appl. Earth Obs. Remote Sens.
(2014)
J. Bi et al.
Sunlight mediated seasonality in canopy structure and photosynthetic activity of Amazonian rainforests
Environ. Res. Lett.
(2015)
P. Blanc et al.
The HelioClim project: surface solar irradiance data for climate applications
Remote Sens.
(2011)
C. Couralet et al.
Phenology in functional groups of central African rainforest trees
J. Trop. For. Sci.
(2013)
C. De Wasseige et al.
Retrieval of tropical forest structure characteristics from bi-directional reflectance of SPOT images
Remote Sens. Environ.
(2002)
M. Derrien et al.
MSG/SEVIRI cloud mask and type from SAFNWC
Int. J. Remote Sens.
(2005)
M. Derrien et al.
Improvement of cloud detection near sunrise and sunset by temporal-differencing and region-growing techniques with real-time SEVIRI
Int. J. Remote Sens.
(2010)
V. Gond
Vegetation structure and greenness in Central Africa from Modis multi-temporal data
Philos. Trans. R. Soc. Lond. B. Biol. Sci.
(2013)
K. Guan et al.
Seasonal coupling of canopy structure and function in African tropical forests and its environmental controls
Ecosphere
(2013)

K. Guan

Photosynthetic seasonality of global tropical forests constrained by hydroclimate

Nat. Geosci.

(2015)

T. Hilker et al.

Remote sensing of tropical ecosystems: atmospheric correction and cloud masking matter

Remote Sens. Environ.

(2012)

T. Hilker et al.

On the measurability of change in Amazon vegetation from MODIS

Remote Sens. Environ.

(2015)

A. Huete et al.

Overview of the radiometric and biophysical performance of the MODIS vegetation indices

Remote Sens. Environ.

(2002)

A.R. Huete

Amazon rainforests green-up with sunlight in dry season

Geophys. Res. Lett.

(2006)

G.J. Huffman

The TRMM multi-satellite precipitation analysis: quasi-global, multi-year, combined-sensor precipitation estimates at fine scale

J. Hydrometeorol.

(2007)

B. Jackson et al.

Mesoscale convective systems over Western Equatorial Africa and their relationship to large-scale circulation

Mon. Weather Rev.

(2009)

R.J. Joyce et al.

CMORPH: a method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution

J. Hydrometeorol.

(2004)

P.H. Kamsu-Tamo et al.

Convection activity over the Guinean coast and Central Africa during northern spring from synoptic to intra-seasonal timescales

Clim. Dyn.

(2014)

M. Lefèvre

McClear: a new model estimating downwelling solar radiation at ground level in clear-sky conditions

Atmos. Meas. Tech.

(2013)

S.L. Lewis

Increasing carbon storage in intact African tropical forests

Nature

(2009)

Cited by (20)

Central African biomass carbon losses and gains during 2010–2019
2024, One Earth
Disturbance, vegetation productivity, and recovery are crucial for aboveground biomass carbon (AGC) dynamics. Here, we use multiple satellite-based datasets to analyze the drivers of AGC dynamics in Central Africa. During 2010–2019, deforestation induced a gross AGC loss of 102.2 ± 17.1 Tg C year⁻¹, which was counterbalanced by an AGC increase of 116.9 ± 41.1 Tg C year⁻¹, leading to a net gain of 14.6 ± 3.8 Tg C year⁻¹. Compared to anthropogenic and soil factors, changes in climate-related factors (e.g., radiation) are more important for the non-deforestation AGC changes. A large AGC increase was found in the northern savannas. In moist forests, strong biomass recovery and growth largely compensated the carbon loss from deforestation and degradation. Considering the increasing resource demand due to rapid population growth, reconciling natural conservation and economic development in Central Africa remains challenging and depends on climate changes and country-specific social-economic conditions.
Examination of the amount of GEDI data required to characterize central Africa tropical forest aboveground biomass at REDD+ project scale in Mai Ndombe province
2023, Science of Remote Sensing
The Global Ecosystem Dynamics Investigation (GEDI) is the first spaceborne LiDAR designed to improve quantification of vegetation structure and forest aboveground biomass (AGB) including in the tropics where forest AGB inventory data are limited. GEDI is a sampling instrument on the International Space Station (ISS) and does not provide data on a regular, systematic basis. Reducing Emissions from Deforestation and Degradation and enhancement of carbon stocks (REDD+) projects require forest AGB inventories to quantify avoided carbon emissions achieved by conserving forest biomass. Although there is high confidence that GEDI can retrieve measurements that allow estimation of AGB at scale, less is known about how well its operational deployment performs for measurement of AGB to support REDD+ projects. This includes an understanding of the appropriate time period required to collect sufficient GEDI observations for reliable forest AGB assessment. This paper describes the first study to examine the amount of GEDI data needed to characterize tropical forest AGB at REDD+ project scale. In tropical Africa, the average REDD+ project size documented by the Center for International Forestry Research is equivalent to a square area of approximately 50 × 50 km (250,000 ha). Recently available good quality GEDI footprint-level AGB product data acquired over a 31 month period over Mai Ndombe province in the west of the Democratic Republic of the Congo were considered. A global 30 m percent tree cover product, updated with contemporary mapped forest cover loss, was used to map the intact forest across the province. Fifteen 50 × 50 km test sites, representing example REDD+ project areas with >80% forest cover and good quality AGB forest footprint data distributed across each site, were selected. The sites were selected from five AGB stratum defined from the GEDI data, and with three sites selected per stratum that had low, medium and high semivariogram sill values that reflect increasing within-site AGB spatial variation. The overall mean GEDI AGB (OMGA) was derived from all the good quality forest GEDI footprint AGB values acquired over the 31 months of GEDI operation at each site. The expected minimum number of GEDI orbits ( $n_{o r b i t s}^{p}$ ) required to characterize the OMGA to within p = ±5%, ±10%, and ±20% was derived by considering different combinations of GEDI orbits randomly selected from the 31 months of GEDI data. The expected minimum number of days ( $n_{d a y s}^{p}$ ) required to characterize the AGB over each site was derived by multiplying the site $n_{o r b i t s}^{p}$ values with a scalar coefficient of 13.03 days. The scalar coefficient was found by counting the temporal intervals between successive GEDI orbits containing good quality forest AGB data and is equivalent to the average number of days required to obtain a GEDI orbit containing good quality forest AGB data at 50 × 50 km scale. Among the 15 sites, observation periods ranging from 65 to 221 days (0.18 – 0.61 years), 143 – 534 days (0.39 – 1.46 years), and 390 – 742 days (1.07 – 2.03 years) were required to characterize the AGB to within ±20%, ±10%, and ±5% of the site OMGA, respectively. The Intergovernmental Panel on Climate Change (IPCC) recommended accuracy requirement for forest AGB estimates is 10%. Thus, to meet this accuracy requirement the findings of this study indicate that at least 534 days (1.46 years) would be required for REDD+ site monitoring using GEDI in Mai Ndombe province. In other central African tropical forest localities these observations periods may be different depending on the forest AGB and spatial variation, cloud cover, ephemeral surface water presence, and GEDI AGB retrieval sensitivity to the forest conditions.
On the determination of coherent solar climates over a tropical island with a complex topography
2020, Solar Energy
Many tropical islands aim at developing a greener self-sufficient energy production systems based on renewable energy, notably solar-generated electricity. This work explores the mean diurnal and annual solar cycles over La Réunion island (southwest Indian Ocean: 21°S, 55.5°E), and their spatial behavior, using the Solar surfAce RAdiation Heliosat – East (SARAH-E) satellite-derived data at high spatial ( $0.05 ° \times 0.05 °$ ) and time (hourly) resolutions over period 1999–2016. Comparisons of the SARAH-E data with ground-based measurements over the period 2011–2015 show differences of $~ 15$ % for diurnal-seasonal variations. The solar resource over the island displays strong spatial variability, with differences larger than 100 $W m^{- 2}$ between coastal and mountainous zones. The mean solar resource is lower on the island than on the nearby sea by $~ 20$ %. The strongest interactions between the diurnal and annual cycles are found at the windward mid-slopes and near the active volcano, in line with the well-known cloud processes encountered there. A clustering of solar zones, based on diurnal-seasonal cycles, structures the island into a dipole that opposes the western to the eastern side of the island.
Estimating hourly land surface downward shortwave and photosynthetically active radiation from DSCOVR/EPIC observations
2019, Remote Sensing of Environment
Citation Excerpt :
However, we find that EPIC-based estimates have higher R2 than CERES at the equatorial site, and one possible reason is the EPIC has the highest spatial resolution in the tropical area. Overall, these results show that our radiation estimates based on a single EPIC instrument well capture the diurnal cycles under different sky conditions at different climate zones, and especially suitable for capturing the complex climate change patterns in wet tropical regions (Philippon et al., 2019; Philippon et al., 2016), with an at-least similar performance as the multi-sensor initiative CERES. Therefore, negative δRMSE and δbias indicate EPIC-based estimates can more accurately characterize the diurnal variation of SW and PAR, due to better capturing of variability of atmosphere and cloud conditions.
The direct and diffuse components of downward shortwave radiation (SW), and photosynthetically active radiation (PAR) at the Earth surface play an essential role in biochemical (e.g. photosynthesis) and physical (e.g. energy balance) processes that control weather and climate conditions, and many ecological processes. Space-based observations have the unique advantage of providing reliable estimates of SW and PAR globally with sufficient accuracy for constructing Earth's radiation budget and estimating land-surface fluxes that control these processes. However, most existing space-based SW and PAR estimations from sensors onboard polar-orbiting and geostationary satellites have inherently low temporal resolution and/or limited spatial coverage of the entire Earth surface. The unique location/orbit of Earth Polychromatic Imaging Camera (EPIC) onboard the Deep Space Climate Observatory (DSCOVR) provides an unprecedented opportunity to obtain global estimates of SW and PAR accurately at a high temporal resolution of about 1–2 h. In this study, we developed and used a model (random forest, RF) to estimate global hourly SW and PAR at 0.1° × 0.1° (about 10 km at equator) spatial resolution based on EPIC measurements. We used a combination of EPIC Level-2 products, including solar zenith angle, aerosol optical depth, cloud optical thickness, cloud fraction, total column ozone and surface pressure with their associated quality flags to drive the RF model for estimating SW and PAR. We evaluated the model results against in situ observations from the Baseline Surface Radiation Network (BSRN) and Surface Radiation Budget Network (SURFRAD). We found the EPIC SW and PAR estimates at both hourly and daily time scales to be highly correlated and consistent with these independently obtained in situ measurements. The RMSEs for estimated daily diffuse SW, direct SW, total SW, and total PAR were 19.10, 38.47, 33.52, and 14.09 W/m², respectively, and the biases for these estimates were 1.71, −0.77, 1.04 and 4.11 W/m², respectively. We further compared the estimated SW and PAR with the Clouds and the Earth's Radiant Energy System Synoptic 1° × 1° (CERES SYN1deg) products and found a good correlation and consistency in their accuracy, spatial patterns and latitudinal gradient. The EPIC SW and PAR estimates provide a unique dataset (i.e. observations from single instrument from pole-to-pole for the entire sunlit portion of Earth) for characterizing their diurnal cycles and their potential impact on photosynthesis and evapotranspiration processes.
Late Holocene forest contraction and fragmentation in central Africa
2018, Quaternary Research (United States)
Historical tree phenology data reveal the seasonal rhythms of the Congo Basin rainforest
2024, Plant-Environment Interactions

View all citing articles on Scopus

¹: Present address: PHyREV, LTHE, UMR 5564CNRS/IRD/UGA, Batiment OSUG-B, 70 rue de la physique, Saint Martin d’Hères 38 400, France.

View full text

Analysis of the diurnal cycles for a better understanding of the mean annual cycle of forests greenness in Central Africa

Highlights

Abstract

Introduction

Section snippets

Terra-firme forest types and their corresponding Enhanced Vegetation Index annual cycles

Computation of the regional indexes of EVI, rainfall, cloudiness and solar radiation for the forests, and of minimum and maximum temperature

The mean annual cycles of EVI, rainfall, total cloudiness, solar radiation and temperature

Discussion

Conclusion

Acknowledgements

Changes in temperature and precipitation extremes in western central Africa, Guinea Conakry, and Zimbabwe, 1955–2006

J. Geophys. Res.

Landscape wetland mapping using multi sensors approach

J. Sel. Top. Appl. Earth Obs. Remote Sens.

Sunlight mediated seasonality in canopy structure and photosynthetic activity of Amazonian rainforests

Environ. Res. Lett.

The HelioClim project: surface solar irradiance data for climate applications

Remote Sens.

Phenology in functional groups of central African rainforest trees

J. Trop. For. Sci.

Retrieval of tropical forest structure characteristics from bi-directional reflectance of SPOT images

Remote Sens. Environ.

MSG/SEVIRI cloud mask and type from SAFNWC

Int. J. Remote Sens.

Improvement of cloud detection near sunrise and sunset by temporal-differencing and region-growing techniques with real-time SEVIRI

Int. J. Remote Sens.

Vegetation structure and greenness in Central Africa from Modis multi-temporal data

Philos. Trans. R. Soc. Lond. B. Biol. Sci.

Seasonal coupling of canopy structure and function in African tropical forests and its environmental controls

Ecosphere

Photosynthetic seasonality of global tropical forests constrained by hydroclimate

Nat. Geosci.

Remote sensing of tropical ecosystems: atmospheric correction and cloud masking matter

Remote Sens. Environ.

On the measurability of change in Amazon vegetation from MODIS

Remote Sens. Environ.

Overview of the radiometric and biophysical performance of the MODIS vegetation indices

Remote Sens. Environ.

Amazon rainforests green-up with sunlight in dry season

Geophys. Res. Lett.

The TRMM multi-satellite precipitation analysis: quasi-global, multi-year, combined-sensor precipitation estimates at fine scale

J. Hydrometeorol.

Mesoscale convective systems over Western Equatorial Africa and their relationship to large-scale circulation

Mon. Weather Rev.

CMORPH: a method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution

J. Hydrometeorol.

Convection activity over the Guinean coast and Central Africa during northern spring from synoptic to intra-seasonal timescales

Clim. Dyn.

McClear: a new model estimating downwelling solar radiation at ground level in clear-sky conditions

Atmos. Meas. Tech.

Increasing carbon storage in intact African tropical forests

Nature