Jordan Alexis Caraballo-Vega

Mr. Jordan Alexis Caraballo-Vega

orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
3 April 2023 in Geophysical Research Letters
Verified
Journal Article
Open access
AMAndrew L. Mullen
Andrew L. Mullen
JWJennifer D. Watts
Jennifer D. Watts
BRBrendan Rogers
Brendan Rogers
MCMark Carroll
Mark Carroll
CEClayton D Elder
Clayton D Elder
EKEthan Kyzivat
Ethan Kyzivat
Small water bodies (i.e., ponds; 0.01 km 2 ). We leverage high‐resolution (3 m) optical satellite imagery from Planet Labs and deep learning methods to map seasonal changes in pond and lake areal extent across four regions in Alaska. Our water maps indicate that changes in open water extent over the snow‐free season are especially pronounced in ponds. To investigate potential impacts of seasonal changes in pond area on carbon emissions, we provide a case study of open water methane emission budgets using the new water maps. Our approach has widespread applications for water resources, habitat and land cover change assessments, wildlife management, risk assessments, and other biogeochemical modeling efforts.

10 November 2025 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
AGAntonia Gambacorta
Antonia Gambacorta
AKAlexander Kotsakis
Alexander Kotsakis
DGDave Gershman
Dave Gershman
NSNarges Shahroudi
Narges Shahroudi
RRRobert Rosenberg
Robert Rosenberg
WBWilliam G. Blumberg
William G. Blumberg
This study presents a first-of-its-kind comprehensive data fusion approach combining hyperspectral microwave (HMW) with backscatter lidar (BSL) measurements for improved atmospheric thermodynamic sounding, with particular emphasis on the Earth’s Planetary Boundary Layer (PBL). This is a simulation-based trade study to demonstrate the enhancement of HMW over traditional microwave (MW) only measurements and the additional benefits of incorporating BSL with both approaches. This pioneering HMW+BSL fusion methodology represents a major advancement, achieving superior performance compared to traditional thermodynamic remote sensing approaches. Specifically, this configuration demonstrates significant enhancement in PBL temperature bias vertical stability and reduces standard deviation error (SDV) by 30% compared to traditional MW-only performance. Water vapor retrievals show similar improvements, with SDV reductions of 50% in the PBL and bias values consistently maintained below the 10% requirement threshold of the PBL DSI program, compared to PoR errors exceeding 30% bias in challenging cloudy regimes. Case studies across diverse oceanic regions reveal particular advantages of this data fusion approach in complex atmospheric conditions, especially in regions dominated by marine stratocumulus clouds and strong temperature inversions where conventional passive-only retrievals are challenging. Beyond thermodynamic profile improvements, our analysis demonstrates remarkable advances in the detection of PBL height (PBLH), with the HMW+BSL configuration achieving mean absolute errors within the 100 meter requirement threshold of the PBL DSI program, representing a step-change improvement over passive-only approaches. This work directly addresses observational gaps identified in the 2017 Earth Science Decadal Survey, positioning our integrated sensing approach as both a near-term enhancement to existing Earth observation capabilities and a pathfinder for future PBL mission architectures.

10 September 2024 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
PMPaul Montesano
Paul Montesano
MMMatthew Macander
Matthew Macander
JCJordan Alexis Caraballo-Vega
Jordan Alexis Caraballo-Vega
MFMelanie J. Frost
Melanie J. Frost
CNChristopher Neigh
Christopher Neigh
MCMark Carroll
Mark Carroll
Scientific analysis of Earth's land surface change benefits from well-characterized multispectral remotely sensed data for which models estimate and remove the effects of the atmosphere and sun-sensor geometry. Top-of-atmosphere (TOA) reflectance in commercial very high resolution (< 5 m; VHR) spaceborne imagery routinely varies for unchanged surfaces because of signal variation from these effects. To reliably identify critical broad-scale environmental change, consistency from surface reflectance (SR) versions of this imagery must be sufficient to identify and track the change or stability of fine-scale features that, though small, may be widely distributed across remote and heterogeneous domains. Commercial SR products are available, but typically the model employed is proprietary and their use is prohibitively costly for large spatial extents. Here we (1) describe and apply an open-source workflow for the scientific community for fine-scaled empirical estimation of surface reflectance from multispectral VHR imagery using reference from synthetic Landsat surface reflectance, (2) examine SR model results and compare with corresponding TOA estimates for a large batch with varying acquisitions in Arctic and Sub-Arctic regions, (3) assess its consistency at pseudo-invariant calibration sites, and (4) quantify improvements in classification of land cover in a Sahelian region.

19 February 2026 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
MLMinh Tri Le
Minh Tri Le
KWKonrad Wessels
Konrad Wessels
MWMargaret Wooten
Margaret Wooten
JCJordan Alexis Caraballo-Vega
Jordan Alexis Caraballo-Vega
MBMolly Brown
Molly Brown
CNChristopher Neigh
Christopher Neigh
With the significantincrease in the availability of 2-3 daily Harmonized Landsat and Sentinel-2 imagery (HLS), this satellite image time series (SITS) can capture the temporal and spatial features for accurate cropland mapping. This study evaluated the performance of four SITS deep learning (DL) classification approaches, including, convolutional LSTM, convolutional GRU, and 3D UNet, alongside a novel DPC-UNet architecture, for regional cropland mapping inSenegal, using 30 m HLS data. The study focused on the Eastern Casamance (ECAS) and Eastern Transition Zone (ETZ) regions. The study created extensive training data from multidate, 2 m WorldView imagery (WV) covering 15 locations (1500 km$^{2}$) and independently generated ${>}$2000 validation points from the WV images. All the SITS DL models generated cropland maps with high overall accuracy ${>}$ 92% in ECAS and ${>}$ 80% in ETZ. All models displayed relatively low precision of 76 to 80% that was caused by the well-documented overestimation of croplands due to confusion with fallow or abandoned fields and bare sand along drainage lines. The relative performance of the models differed between regions, however, the ConvLSTM marginally had the highest F1 score and overall accuracy in both regions. DPC-UNet displayed the most overestimation in ECAS, but the least in the ETZ. The cropland maps produced by the DL models otherwise displayed strong agreement with the benchmark Global Land Analysis and Discovery (GLAD) cropland map and may also be detecting new expanding fields along the edges of the core cropland clusters. Given their performance and rate of progress, it is likely that SITS DL models will become the standard for operational cropland mapping in the near future, especially with the eminent, dense frequency, 10m HLS data.

4 August 2026 in Environmental Research Letters
Verified
Journal Article
Open access
PDF file
PMPaul Montesano
Paul Montesano
MFMelanie Frost
Melanie Frost
JCJordan Alexis Caraballo-Vega
Jordan Alexis Caraballo-Vega
MMMatthew Macander
Matthew Macander
GFGerald Frost
Gerald Frost
MCMark Carroll
Mark Carroll
A reconfiguration of vegetation structure patterns in the boreal forest-tundra ecotone and Low Arctic tundra is underway, driven by shrubification, tree expansion, permafrost thaw, and disturbance regimes linked to rapid warming. The short-statured vegetation in these regions presents challenges for resolving the heterogeneity of plant functional types (PFTs), even when very high resolution spaceborne imagery is used with earth observation artificial intelligence (AI EO ) models. In this study, we resolve height variation of high northern latitude PFTs by fine-tuning an AI EO model built for global applications using a large variety of domain-specific airborne training and validation data, and field observations. We map, validate, and quantify vegetation heights at 2 m horizontal and decimeter vertical scales, and estimate the area of height classes across 40% of Alaska’s boreal-tundra ecotone. Results show most of this area is low stature vegetation <0.35 m in height (63.4% ± 5.2%, 95% confidence interval), while various low shrub (11.9% ± 1.2%, 6.3%± 1.1%) and tall shrub and forest (10.9% ± 1.9%, 3.1% ± 0.6%, 1.8 ± 0.5%, 2.6 ± 0.8%) classes distinguish height intervals useful for vegetation structure and plant functional type monitoring in Alaska. Together, canopies <1.5 m tall comprise approximately 82% of the mapped landscape, nearly 18% of which are some form of low shrubs 0.35–1.5 m in height, while height classes ⩾1.5 m occupy the remaining ∼18% of the mapped area. Validation reveals that predictions underestimate the tallest surfaces of the vegetation canopy. This work addresses a gap in resolving the fine-scaled vegetation heterogeneity associated with key PFTs, leveraging the complementarity of commercial and public remote sensing data, to provide a basis for assessing the variation in the routes and rates of Arctic and sub-Arctic vegetation change.

11 December 2025 in arXiv
Verified
Preprint
Open access
PDF file
BPBrian P. Powell
Brian P. Powell
JCJordan Alexis Caraballo-Vega
Jordan Alexis Caraballo-Vega
MCMark Carroll
Mark Carroll
TMThomas Maxwell
Thomas Maxwell
APAndrew Ptak
Andrew Ptak
JMJorge Martinez-Palomera
Jorge Martinez-Palomera
We report the discovery that binary encoding allows neural networks to extrapolate periodic functions beyond their training bounds. We introduce Normalized Base-2 Encoding (NB2E) as a method for encoding continuous numerical values and demonstrate that, using this input encoding, vanilla multi-layer perceptrons (MLP) successfully extrapolate diverse periodic signals without prior knowledge of their functional form. Internal activation analysis reveals that NB2E induces bit-phase representations, enabling MLPs to learn and extrapolate signal structure independently of position.