Spatial-temporal dynamics of transboundary forest disturbance-recovery and its influencing factors in the central Himalayas

Figures & data

Figure 1. Topography map and location of the central Himalayas (the digital boundary of the GRB from (Nie et al. 2017) and Tibetan Plateau from (Zhang et al. 2021).

Table 1. Datasets used in this study.

	Data	Abbreviation	Data sources	Spatial resolution	Time series
Datasets	Landsat	–	http://landsat.usgs.gov	30m	1995–2018
	MODIS land cover type collection	MCD12Q1	https://lpdaac.usgs.gov/products/mcd12q1v006/	500m	2001–2019
	Climate change initiative-land Cover	CCI-LC	https://www.esa-landcover-cci.org/	300m	1992–2015
	GlobaLand30	–	http://www.globeland30.com	30m	2000, 2010 and 2020
	Finer resolution observation and monitoring-global land cover	FROMGLC	http://data.ess.tsinghua.edu.cn/data/Simulation/	30m	2015 and 2017
Natural factors	Annual maximum temperature	MAT	https://www.climatologylab.org/terraclimate.html/	2.5 arc minutes	1995–2018
	Annual minimum temperature	MNT	https://www.climatologylab.org/terraclimate.html/	2.5 arc minutes	1995–2018
	Average annual precipitation	PRE	https://www.climatologylab.org/terraclimate.html/	2.5 arc minutes	1995–2018
	Palmer drought severity index	PDSI	https://www.climatologylab.org/terraclimate.html/	2.5 arc minutes	1995–2018
	Elevation	DEM	https://earthexplorer.usgs.gov/	30m	–
	Aspect	ASP	Calculated according to elevation	30m	–
Anthropogenic factors	Population density	POP	https://www.worldpop.org/	100m	2000, 2007 and 2018
	Distance to roads	RDD	https://www.openstreetmap.org Euclidean distance from nearest roads	–	2017
	Distance to residential areas	RAD	https://www.icimod.org/ Euclidean distance from nearest residential areas	–	2000
	Distance to rivers	RVD	https://www.icimod.org/ Euclidean distance (m) from nearest rivers	–	2000
	Forest management		https://zenodo.org/record/5879022	100m	2015

Figure 2. The flow chart to generate forest disturbance recovery mapping and influencing factors analysis, which includes (a) annual forest probability estimation module, (b) forest disturbance and recovery change mapping module, (c) forest spatial and temporal evolution analysis module and (d) analysis of factors influencing changes in forest dynamics module.

Table 2. Interaction categories of two factors and the interaction relationship.

Description	Interaction
$\mathrm{q}\mathrm{(}\text{X1}\mathrm{\cap }\text{X2}\mathrm{)}\mathrm{<}\text{ Min }\text{(q(}\text{X1}\mathrm{)}\text{,q}\mathrm{(}\text{X2}\mathrm{)}\mathrm{)}$	Nonlinear − weaken
$\mathrm{Min}(q(X1),q(X2))<q(X1\cap X2)<\mathrm{Max}(q(X1),q(X2))$	Univariate − weaken
$q(X1\cap X2)>\mathrm{Max}(q(X1),q(X2))$	Bivariate-enhance
$q(X1\cap X2)=q(X1)+q(X2)$	Independent
$q(X1\cap X2)>q(X1)+q(X2)$	Nonlinear − enhance

Table 3. Accuracy assessment of forest change detection for each period.

Period		1	2	3	4	UA	PA	F1	OA
1995–2003	1	228	4	5	0	96.20	96.20	96.20	89.11
	2	4	182	15	20	82.35	94.30	87.92
	3	4	4	102	10	85.00	82.26	83.61
	4	1	3	2	77	92.77	71.96	81.05
2004–2011	1	228	3	8	1	95.00	96.20	95.60	92.78
	2	4	182	1	9	92.86	94.30	93.57
	3	4	8	124	9	85.52	93.23	89.21
	4	1	0	0	83	98.81	81.37	89.25
2012–2018	1	230	1	3	0	98.29	97.05	97.66	94.78
	2	2	189	5	1	95.94	97.93	96.92
	3	4	3	105	12	84.68	91.30	87.87
	4	1	0	2	93	96.88	87.74	92.01

Note: 1: Non-forest; 2: Stable forest; 3: Forest disturbance; 4: Forest recovery; UA: User Accuracy; PA: Producer Accuracy; OA: overall accuracy. The F1 represents the harmonic mean of the PA and the UA.

Figure 3. Annual forest disturbance-recovery area changes and cumulative percentage from 1995 to 2018.

Figure 4. Spatial patterns of transboundary forest disturbance-recovery in each detection period. (a) 1995–2018, (b) 1995–2003, (c) 2004–2011, (d) 2012–2018.

Figure 5. Annual disturbance and recovery rates of forests in India, Nepal, and China for each detection period. (a) Forest disturbance rate; (b) forest recovery rate.

Figure 6. Forest change intensity at the district level in the study area between 1995 and 2018. DD: Dehra dun; TG: Tehri Garhwal; RP: Rudra prayag; PG: Pauri Garhwal; BS: Bageshwar; CP: Champawat; UN: Udham Singh Nagar. The intensity of change (P) was divided into five categories: extremely low (p ≤ 1.44), low (1.44 < p ≤ 1.63), medium (1.63 < p ≤ 1.84), high (1.84 <p ≤ 2.7), extremely high (p ≥ 2.7).

Figure 7. PD values of the influencing factors from 1995 to 2018 in the transboundary forest. All PD values were statistically significant at the 1% level. The error bars represent the 1 – σ uncertainty. MXT: annual maximum temperature; MNT: annual minimum temperature; PRE: annual precipitation; DEM: elevation; ASP: aspect; POP: population density; RDD: distance to roads; RAD: distance to residential areas; RVD: distance to rivers.

Figure 8. Interaction effects of factors influencing transboundary forest dynamic change in each detection period. (a) 1995–2003; (b) 2004–2011; (c) 2012–2018; MXT: annual maximum temperature; MNT: annual minimum temperature; PRE: annual precipitation; DEM: elevation; ASP: aspect; POP: population density; RDD: distance to roads; RAD: distance to residential areas; RVD: distance to rivers.

Figure 9. Occurrence frequency of transboundary forest disturbance and restoration along elevation gradient from 1995 to 2018 in the central Himalayas.

Nie Y, Sheng Y, Liu Q, Liu L, Liu S, Zhang Y, Song C. 2017. A regional-scale assessment of himalayan glacial lake changes using satellite observations from 1990 to 2015. Remote Sens Environ. 189:1–13. doi: 10.1016/j.rse.2016.11.008.

 Web of Science ®Google Scholar

Zhang Y, Li B, Liu L, Zheng D. 2021. Redetermine the region and boundaries of Tibetan plateau. Geograph Res. 40(6):1543–1553.

 Google Scholar

Data availability statement

The data that support the findings of this study are available from the corresponding author, upon reasonable request.