BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.cienciaenmexico.com//igisc-2026//talk//KDTZMQ
BEGIN:VTIMEZONE
TZID:CST
BEGIN:STANDARD
DTSTART:20001029T030000
RRULE:FREQ=YEARLY;BYDAY=5SU;BYMONTH=10;UNTIL=20001029T080000Z
TZNAME:CST
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
END:STANDARD
BEGIN:STANDARD
DTSTART:20010930T030000
RRULE:FREQ=YEARLY;BYDAY=5SU;BYMONTH=9;UNTIL=20010930T080000Z
TZNAME:CST
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
END:STANDARD
BEGIN:STANDARD
DTSTART:20021027T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CST
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000402T030000
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=4;UNTIL=20000402T090000Z
TZNAME:CDT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20010506T030000
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=5;UNTIL=20010506T090000Z
TZNAME:CDT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20020407T030000
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=4;UNTIL=20220403T090000Z
TZNAME:CDT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-igisc-2026-KDTZMQ@pretalx.cienciaenmexico.com
DTSTART;TZID=CST:20261112T130500
DTEND;TZID=CST:20261112T132000
DESCRIPTION:Tropical deforestation represents one of the most critical envi
 ronmental threats of the 21st century\, with direct implications for biodi
 versity\, the hydrological cycle\, and regional and global carbon balance.
  In Latin America\, Colombia stands as one of the countries facing the gre
 atest pressure on forest cover\, with the department of Guaviare ranking a
 mong the territories with the highest rates of forest loss in the country\
 , driven by the expansion of the agricultural frontier\, extensive cattle 
 ranching\, and the presence of illicit crops. These pressures have resulte
 d in the loss of approximately 452\,000 hectares of forest cover between 2
 001 and 2024\, equivalent to 8.6% of the department's total forest area. I
 n this context\, spatial modeling of deforestation susceptibility emerges 
 as a fundamental tool for guiding conservation strategies and land-use pla
 nning policies. However\, most existing studies apply methods that assume 
 spatial independence among observations\, ignoring that deforestation oper
 ates as a contagion process with strong spatial autocorrelation — a limi
 tation that leads to biased estimates of conditioning factors and suscepti
 bility maps with reduced discriminative precision.\n\nThis study evaluated
  whether explicitly incorporating spatial autocorrelation through the Inte
 grated Nested Laplace Approximation with a Stochastic Partial Differential
  Equation approximation (INLA-SPDE) improves the deforestation susceptibil
 ity mapping capacity compared to a traditional logistic regression (LR) mo
 del. The analysis was conducted on 113\,504 deforestation and non-deforest
 ation points in the department of Guaviare\, Colombia\, for the period 201
 2–2024\, derived from the forest cover loss catalog of the Institute of 
 Hydrology\, Meteorology and Environmental Studies (IDEAM). Twelve explanat
 ory variables of biophysical and anthropogenic nature were used: slope\, e
 levation\, aspect\, mean annual precipitation\, distance to roads\, distan
 ce to agricultural areas\, distance to rivers\, distance to drainage syste
 ms\, distance to populated centers\, distance to protected areas\, populat
 ion density\, and mean annual temperature.\n\nThe exploratory data analysi
 s — both statistical and spatial — confirmed the relevance of the spat
 ial modeling approach. The global Moran's I index yielded a value of I = 0
 .830 (p < 2.2 × 10⁻¹⁶)\, demonstrating significant positive spatial 
 autocorrelation. LISA indicators identified well-defined High-High and Low
 -Low spatial clusters\, while Ripley's L(d) and F(d) functions confirmed a
  consistent spatial clustering pattern across multiple scales\, reinforcin
 g the evidence of non-random spatial structure in the deforestation phenom
 enon. This body of evidence justified the use of models that explicitly in
 corporate spatial dependence. Correlation and multicollinearity analysis (
 VIF) determined the exclusion of mean temperature — the variable with th
 e highest redundancy relative to precipitation (ρ = 0.93) — and motivat
 ed a log transformation of variables with highly skewed distributions\, fo
 llowed by Z-score standardization of all covariates prior to model fitting
 .\nData partitioning was performed differently for each model: a stratifie
 d random split (80/20) was applied for the LR model\, while spatial block 
 cross-validation using 10\,000 × 10\,000 m blocks was implemented for INL
 A-SPDE\, consistent with the estimated spatial correlation range of approx
 imately 5\,000 m. This approach was adopted to prevent the overestimation 
 of predictive performance associated with spatial contiguity between train
 ing and test sets\, ensuring that model evaluation reflects genuine genera
 lization capacity toward unobserved areas.\n\nResults demonstrate the supe
 riority of the spatial model across all evaluation criteria. The INLA-SPDE
  model achieved an AUC of 0.9705 compared to 0.9544 for the LR model\, rep
 resenting an improvement of 1.61 percentage points attributable exclusivel
 y to the incorporation of the latent spatial component. The estimated Gaus
 sian field presented a correlation range of 5\,290 m and a standard deviat
 ion of σ = 3.38\, indicating a relevant residual spatial structure not fu
 lly captured by the observed covariates. Evaluation using the SCAI index r
 evealed that the INLA-SPDE model concentrated 88.20% of the observed defor
 estation within the very high susceptibility class (SCAI = 0.205)\, compar
 ed to 73.20% for the LR model (SCAI = 0.241)\, with a sharper separation b
 etween classes at both extremes of the susceptibility distribution.\n\nThe
  comparative coefficient analysis identified three behavioral groups. A fi
 rst group — distance to agricultural areas\, distance to roads\, mean pr
 ecipitation\, and distance to drainage systems — maintained or intensifi
 ed its effect after controlling for spatial dependence\, establishing thes
 e variables as structurally stable determinants of deforestation in Guavia
 re. A second group — distance to rivers and elevation — collapsed towa
 rd zero or reversed sign in INLA-SPDE\, revealing that their effects in th
 e logistic model were largely spurious\, acting as proxies for unmodeled s
 patial heterogeneity rather than reflecting genuine causal relationships. 
 A third group — slope and distance to protected areas — exhibited sign
  reversals accompanied by changes in statistical significance\, uncovering
  real effects that had been masked by uncontrolled autocorrelation in the 
 classical specification.\n\nThe overall performance of both models surpass
 es the range reported in the literature for classical logistic methods (AU
 C between 0.68 and 0.87)\, machine learning approaches such as Support Vec
 tor Machine (AUC = 0.907) and artificial neural networks (AUC between 0.86
 9 and 0.921)\, and multicriteria approaches such as VIKOR (AUC = 0.938). T
 he advantage of the INLA-SPDE model is particularly significant in the Col
 ombian Amazonian context\, where deforestation exhibits a pattern of conti
 guous spatial expansion from pre-existing deforestation nuclei — a struc
 ture that the latent Gaussian field efficiently captures through its estim
 ated correlation range\, and that classical logistic regression fundamenta
 lly cannot represent under its independence assumption.\n\nThese results p
 osition INLA-SPDE as a methodologically robust and predictively superior a
 lternative for deforestation susceptibility mapping in territories with ma
 rked spatial structure. Furthermore\, they open the possibility of extendi
 ng this framework toward spatiotemporal models capable of projecting the f
 uture dynamics of the forest frontier in the department of Guaviare\, cont
 ributing to the development of evidence-based early warning systems for fo
 rest conservation in the Colombian Amazon.
DTSTAMP:20261122T124241Z
LOCATION:Sala Geomática 7
SUMMARY:Spatial Modeling of Deforestation Susceptibility Using INLA-SPDE: A
  Comparative Assessment Against Traditional Logistic Regression in Guaviar
 e\, Colombia - Edgar Libardo Barahona Cantor
URL:https://pretalx.cienciaenmexico.com/igisc-2026/talk/KDTZMQ/
END:VEVENT
END:VCALENDAR
