Impact of Drought Severity and Topography on Tropical Forests of Costa Rica

Kaiya Weatherby1,2, Gang Zhao2, Huilin Gao2, Georgianne Moore3, Kelly Brumbelow2

1Department of Earth and Environment, Boston University, Boston, MA 022152Zachry Department of Civil and Environmental Engineering, Texas A&M University, College Station, TX 77845

3Department of Ecosystem Science and Management, Texas A&M University, College Station, TX 77845

Acknowledgements:• Funding for this Research Experiences for Undergraduate program is provided by the National Science

Foundation’s Division of Earth Sciences (EAR-1659848).

• Eugenio Gonzalez – Director at Texas A&M Soltis Center

Introduction

• Understanding tropical rainforest responses to extreme climate events is

critical given the amount of ecosystem services these forests provide for

their flora and fauna as well as humans.

• However, these response mechanisms are still not very well understood,

especially when it comes to tropical montane rainforests.

• Characterized with large variations of vegetation properties, precipitation

amount, and topographical gradients, Costa Rica serves as an ideal test

bed for evaluating the impacts of drought conditions on tropical

rainforests.

Fig 1. Costa Rica trends in various parameters. a) Time series of the monthly average PDSI (Palmer Drought

Severity Index) values for Costa Rica. b) map of Costa Rican forest types c) 50-year average precipitation

map using TerraClimate data. d) 20-year average EVI (Enhanced Vegetation Index) map using MODIS data.

e) Elevation map using SRTM Digital Elevation data.

c) d)

Data & Methods• Various remotely sensed satellite and meteorological datasets available on

Google Earth Engine were utilized.

• EVI (Enhanced Vegetation Index) quantifies forest greenness

• LST (Land Surface Temperature) temperature of the canopy

• ET (Evapotranspiration) evaporation plus plant transpiration data from

MODIS sensor

• PDSI (Palmer Drought Severity Index) index for drought severity based

on precipitation and temperature data from the TerraClimate dataset

• *Data from the months of January to April (dry months) were used for

analysis, with the intention of avoiding seasonal variation

Categories and Regions• The country was divided up into regions based on drought severity (PDSI)

during the drought years. Region 1 (dark gray) represents the least severe

drought, Region 2 (light gray) moderately severe drought, and Region 3

(white) most severe drought

• To account for elevational differences, elevation categories were created.

For simplicity, the categories consist of ‘High’ (black), ‘Middle’(gray), and

‘Low’ (white) elevation.

e)

Fig 2. a) Regions based on PDSI values from drought years. Region 1, dark gray, has a PDSI range of -2.5 to

-3.5; Region 2, light gray, ranges from -3.5 to -4.5; Region 3, white, is less than or equal to -4.5 b) High

elevation, black, is defined to be greater than 1200m; Middle elevation, gray, is between 300m and 1200m;

Low elevation, white, is less than or equal to 300m . c) The non-forested area of Costa Rica was excluded

for data collection/analysis

a) b) c)

Results• The PDSI yearly average clearly drops during 2012 or 2013 and does not

return to ”normal” conditions until 2017.

• EVI seems to gradually decline during these years, and increases notably in

2017.

• Comparison of trends suggest the possibility of a lagged-effect

Results

Conclusion

References• Jadin, I, et al. “International Trade, and Land Use Intensification and Spatial Reorganization

Explain Costa Rica’s Forest Transition.” Environmental Research Letters, vol. 11, no. 3,

2016, p. 035005., doi:10.1088/1748-9326/11/3/035005.

• Yang, Yan, et al. “Post-Drought Decline of the Amazon Carbon Sink.” Nature

Communications, vol. 9, no. 1, 2018, doi:10.1038/s41467-018-05668-6.

b)

Objective

The primary objective of this study is to explore the responses of Costa

Rican forests to drought conditions while taking differences in severity of

drought and elevation into account.

• Figure 4 shows EVI changes by elevation for Costa Rica as a whole, and

by drought region (as defined in the method section)

• Most cases demonstrate a gradual decline in EVI during 2013 – 2016,

followed by a resurgence in 2017

• Notably, the lowest elevation of region 3 has the smallest EVI values, and

the highest elevation has the greatest EVI values. This is most likely due

to the different vegetation types that are present in this region (further

discussed in the “Discussion” section)

Fig 3. EVI and PDSI Trends in Costa Rica

Fig 4. a) EVI time series for all forested areas in

Costa Rica by elevation. b) EVI time series for

drought region 1 by elevation. c) EVI time series for

drought region 2 by elevation. d) EVI time series for

drought region 3 by elevation.

Fig 5. Trends in LST and ET compared to EVI trend in drought region 3, lowest elevation.

• Figure 6 shows LST was poorly correlated to EVI in all elevations.

• Differences in EVI trends were seen when zooming into each drought

region and elevation category.

• While there was an overall decline in EVI for all areas, some drought-

elevation subcategories showed irregular patterns (like the example

above).

• As shown in Figure 5 and 6, EVI and ET appear to display similar

behavior in lower elevations. This correlation between the two

parameters was found to be weaker in other elevations.

Fig 6. Scatter plot showing correlation between EVI and ET

Discussion

• Due to the variation in topography and precipitation patterns, Costa Rican

forest types are diverse.

• Region 3, according to PDSI patterns, experienced the most severe

drought conditions, yet the EVI trends are among the least affected.

• This most likely has to do with the fact that region 3 includes the tropical

dry forest type (among other forest types), coupled by the fact that this

part of the country receives less rainfall on average

• It is possible that although this region experienced the most severe

drought conditions, the vegetation here may not be susceptible to such

conditions in the same way that tropical wet/moist forests are.

• As for the highest elevation EVI being highest for drought region 3, this

can be a result of having tropical dry forests in lower elevations and

having tropical wet/moist forests in higher elevations all within region 3.

• EVI trends show a gradual overall decrease for all elevations during the

years 2013-2016, when PDSI values are well below normal conditions

(possibility of a lagged-effect of drought on forest greenness)

• EVI and ET appear to be more correlated in lower elevation tropical

forests, compared to other elevations.

a)

b) c)

d)

a)