North Carolina's coastal forest gave way to marsh and ghost forest three times faster per year after 2010

The 1.5 compares two totals gathered over windows of very different length. The later window, 2010 to 2021, covers 11 years. The earlier one, 1985 to 2010, covers 25 ^[2]. Put on a yearly basis, conversion to marsh, ghost forest and shrub ran at about 680 hectares a year before 2010 and about 2,170 a year after [d1, d2]. That is roughly 3.2 times the earlier pace ^[15]. Divided directly, the raw totals give about 1.4 ^[16].

Ghost forest is what is left when salt water floods a stand and kills the trees, and it grew faster still ^[5]. It gained 3,087 hectares in the first window and 7,561 in the second ^[4]. Per year, that is about 123 hectares against 687, a ratio near 5.6 ^[18].

The thing this doesn't tell you is whether the faster pace is a trend or a step. The study splits 36 years into two bins ^[3], and the seam at 2010 sits inside a run of extreme weather on the Albemarle-Pamlico Peninsula ^[20]. "This area experienced severe drought from 2007 to 2011, and then also Hurricane Irene in 2011," Tajudeen said ^[7]. "While these events occurred long before 2021, some of the areas simply never recovered. Despite being protected, a combination of these extreme events along with rising sea levels has pushed them into new ecological states, including becoming ghost forests." ^[8] If trees started dying in 2011 and the deaths were counted over the following decade, the post-2010 average would rise even if the underlying sea-level-driven rate held steady. The study's summary is careful on this point: loss began accelerating after 2010 and may still be speeding up ^[20].

A second caveat comes from measurement. The NC State team trained a convolutional neural network to separate forest from marsh, ghost forest and shrub in Landsat and Sentinel-2 imagery ^[9]. Sentinel-2 resolves 10 metres to Landsat's 30. In 2021, a year both sensors covered, Sentinel-2 produced the more accurate maps [c10, c11]. Only Landsat reaches back to 1985 ^[10]. The early baseline rests on the coarser sensor, and the published summary does not include accuracy figures for either.

The totals also leave a remainder. Conversions across the two windows sum to 40,844 hectares, out of total forest loss of about 64,220 and a 1985 forest of roughly 300,000 [c1, d6, d7]. The other 23,376 hectares, about 36% of the loss, fall outside the marsh, ghost forest and shrub classes in the summary ^[17].

For coastal land managers, I think the planning case holds, with conditions attached. A budget runs on yearly figures, and the yearly ratio is more than double the 1.5 in the study's own framing ^[15]. Most of the heavily affected ground lies within a kilometre of the coast ^[6], and in Tajudeen's account protection did not save it ^[8]. I'd use the post-2010 rate of about 2,170 hectares a year as the working figure for the study area until a year-by-year series shows whether loss eased once the drought and Irene had passed ^[13].